Attitudes matter: The 2021 National Community Attitudes towards Violence against Women Survey (NCAS), Findings for Australia

ANROWS acknowledgement

This material was produced with funding from the Australian Government. Australia’s National Research Organisation for Women’s Safety (ANROWS) gratefully acknowledges the financial and other support it has received from the Australian Government, without which this work would not have been possible. The findings and views reported in this paper are those of the authors and cannot be attributed to the Australian Government.

Acknowledgement of Country

ANROWS acknowledges the Traditional Owners of the land across Australia on which we work and live. We pay our respects to Aboriginal and Torres Strait Islander Elders past and present, and we value Aboriginal and Torres Strait Islander histories, cultures, and knowledge. We are committed to standing and working with Aboriginal and Torres Strait Islander peoples, honouring the truths set out in the  Warawarni-gu Guma Statement .

Peer review process

The quality of ANROWS publications is ensured through a rigorous peer review process that is consistent with the principles of the  Committee on Publication Ethics (COPE) Ethical Guidelines for Peer Review . This report has been assessed by at least two peer reviewers with relevant academic expertise.

© ANROWS 2023

With the exception of the ANROWS branding, content provided by third parties, and any material protected by a trademark, all material presented in this publication is licensed under a Creative Commons Attribution Non-Commercial 3.0 Australia (CC BY-NC 3.0 AU) licence.

The full licence terms are available at  creativecommons.org/licenses/by- nc /3.0/au/ legalcode

Published by

Australia’s National Research Organisation for Women’s Safety Limited (ANROWS)

PO Box Q389, Queen Victoria Building, NSW 1230 |  www.anrows.org.au

ABN 67 162 349 171

ISBN: 978-1-922645-65-4 (paperback)

ISBN: 978-1-922645-64-7 (PDF)

Please note that there is the potential for minor revisions of this report.

Please check the online version at  www.anrows.org.au  for any amendments.

Project lead

Dr Christine Coumarelos

Director, Research Program (NCAS), ANROWS

Research team

Dr Nicole Weeks

Senior Research Officer (NCAS), ANROWS

Dr Shireen Bernstein

Senior Research Officer (NCAS), ANROWS

Dr Natalie Roberts

Senior Research Officer (NCAS), ANROWS

Dr Nikki Honey

Executive Director, Quantitative Research Consulting, Social Research Centre

Kate Minter

Senior Research Officer (NCAS), ANROWS

Dr Erin Carlisle

Senior Research Officer (NCAS), ANROWS

Author acknowledgement

We thank our research partner, the Social Research Centre (SRC), led by Dr Nikki Honey. The researchers and statisticians at the SRC who contributed to the data collection and statistical analysis included Andrew Ward, Laura Rimington, Sandra Ropero and Sebastian Misson.

We also thank the respondents who took the time to participate in the survey.

Acknowledgement of lived experiences of violence

ANROWS acknowledges the lives and experiences of the women and children affected by domestic, family and sexual violence who are represented in this report. We recognise the individual stories of courage, hope and resilience that form the basis of ANROWS research.

Caution: Some people may find parts of this content confronting or distressing. Recommended support services include 1800RESPECT (1800 737 732), Lifeline (13 11 14) and, for Aboriginal and Torres Strait Islander people, 13YARN (13 92 76).

This report addresses work covered in ANROWS’s National Community Attitudes towards Violence against Women Survey (NCAS) Research Program. Please consult  the ANROWS website  for more information on this research program.

ANROWS research contributes to the shared vision to end gender-based violence in one generation of the  National Plan to End Violence against Women and Children 2022–2032 (National Plan 2022–2032) and the six National Outcomes of the  National Plan to Reduce Violence against Women and their Children 2010–2022 (National Plan 2010–2022). This research provides prevention and early intervention key indicators for the National Plan 2022–2032 and addresses National Outcome 1 – Communities are safe and free from violence, and National Outcome 2 – Relationships are respectful, of the National Plan 2010–2022.

Suggested citation

Coumarelos, C., Weeks, N., Bernstein, S., Roberts, N., Honey, N., Minter, K., & Carlisle, E. (2023).  Attitudes matter: The 2021 National Community Attitudes towards Violence against Women Survey (NCAS), Findings for Australia. (Research report 02/2023). ANROWS.

Contents

Shortened forms and data symbols

Shortened forms

Data symbols and table and figure notations

Key terms

About this report

Executive summary

Research design and analysis

Key findings and implications

1 Introduction: Violence against women and the need for action

1.1 Climate of violence against women

1.2 Facilitators of a climate of violence

1.3 Deconstructing the climate of violence: Prevention

2 Research design

2.1 Aims of the 2021 NCAS

2.2 2021 NCAS instrument

2.3 Sampling

2.4 Demographics of the final sample

2.5 Analysis and reporting

2.6 Strengths and limitations

3 Findings: Benchmarking understanding and attitudes

3.1 Benchmarking broad understanding and attitudes

3.2 Types of violence in focus: Benchmarking understanding and attitudes

3.3 Conclusion: Benchmarking understanding and attitudes

4 Findings: Understanding of Violence against Women Scale (UVAWS)

4.1 Understanding of violence against women over time by gender

4.2 Understanding of violence against women: UVAWS subscales

4.3 Understanding of violence against women: Assessing the importance of demographics

4.4 Conclusions about understanding of violence against women

5 Findings: Attitudes towards Gender Inequality Scale (AGIS)

5.1 Attitudes towards gender inequality over time by gender

5.2 Attitudes towards gender inequality: AGIS subscales

5.3 Attitudes towards gender inequality: Assessing the importance of demographics and understanding

5.4 Conclusions about attitudes towards gender inequality

6 Findings: Attitudes towards Violence against Women Scale (AVAWS)

6.1 Attitudes towards violence against women over time by gender

6.2 Attitudes towards violence against women: AVAWS subscales

6.3 Attitudes towards violence against women: Assessing the importance of demographics, understanding and attitudes

6.4 Conclusions about attitudes towards violence against women

7 Findings: Specific types of violence against women

7.1 The AVAWS and type of violence scales

7.2 Domestic violence

7.3 Sexual violence

7.4 Technology-facilitated abuse

7.5 Stalking: Technology-facilitated and in person

7.6 Conclusions about types of violence against women

8 Findings: Bystander response

8.1 2021 NCAS bystander scenarios

8.2 Bystander response to each scenario

8.3 Anticipated peer support or criticism

8.4 Barriers to bystander intention to intervene

8.5 Likely bystander responses: Assessing the importance of demographics, understanding and attitudes

8.6 Conclusions about bystander intention to intervene

9 Findings: People and contexts

9.1 Gender

9.2 Age

9.3 Sexuality

9.4 Disability

9.5 Country of birth and English proficiency

9.6 Formal education

9.7 Main labour activity

9.8 Socioeconomic status of area

9.9 Major cities, regional and remote areas

9.10 Gender composition of occupation and social contexts

10 Implications for ending violence against women and for further research

10.1 Key NCAS findings

10.2 Implications for ending violence against women

10.3 Implications for future research

11 Conclusion

12 References

13 Appendix A: NCAS Panel of Experts and Advisory Group

13.1 NCAS Panel of Experts

13.2 NCAS Advisory Group and other stakeholder advisors

Shortened forms and data symbols

Shortened forms

Acronym	Meaning
ABS	Australian Bureau of Statistics
ABS Standard	ABS Standard for Sex, Gender, Variations of Sex Characteristics and Sexual Orientation Variables 2020
AFP	Australian Federal Police
AGIS	Attitudes towards Gender Inequality Scale
AHRC	Australian Human Rights Commission
AIHW	Australian Institute of Health and Welfare
ANROWS	Australia’s National Research Organisation for Women’s Safety
AVAWS	Attitudes towards Violence against Women Scale
CASVAWS	Community Attitudes Supportive of Violence against Women Scale (2017 NCAS)
CDC	Centers for Disease Control and Prevention
Change the Story	Change the story: A shared framework for the primary prevention of violence against women in Australia (2nd ed.) (Our Watch, 2021a)
COAG	Council of Australian Governments
DSD	Disorder/difference of sex development
DVS	Domestic Violence Scale
eSafety	eSafety Commissioner
GEAS	Gender Equality Attitudes Scale (2017 NCAS scale)
GVIS	Gendered Violence and Inequality Scale
LGBTQ+	An evolving acronym that stands for lesbian, gay, bisexual, transgender, queer/questioning, asexual and other sexuality- or gender-diverse people
LOTE	Language other than English
MESC	Main English–speaking country
National Plan 2010–2022	National Plan to Reduce Violence against Women and their Children 2010–2022
National Plan 2022–2032	National Plan to End Violence against Women and Children 2022–2032
NCAS	National Community Attitudes towards Violence against Women Survey
N-MESC	Non-main English–speaking country
PSS	Personal Safety Survey
RDD	Random digit dialling
Recognise DV Subscale	Recognise Domestic Violence Subscale
Recognise VAW Subscale	Recognise Violence Against Women Subscale
SEIFA	An ABS index that measures socioeconomic conditions by geographic area (Index of Relative Socio-Economic Advantage and Disadvantage)
SAS	Sexual Assault Scale
SHS	Sexual Harassment Scale
SVS	Sexual Violence Scale
Technical report	Coumarelos, C., Honey, N., Ward, A., Weeks, N., & Minter, K. (2023).  Attitudes matter: The 2021 National Community Attitudes towards Violence against Women Survey (NCAS), Technical report. ANROWS.
TFAS	Technology-Facilitated Abuse Scale
Understand Gendered DV Subscale	Understand Gendered Domestic Violence Subscale
UVAWS	Understanding of Violence against Women Scale
VLRC	Victorian Law Reform Commission
WHO	World Health Organization
WGEA	Workplace Gender Equality Agency

Data symbols and table and figure notations

Data symbol	Meaning
*	Indicates a statistically significant result, meaning we can be confident (with 95% certainty) that the difference observed in the survey sample is meaningful and likely to represent a true difference in the Australian population ( p < 0.05) that is not negligible in size (Cohen’s  d ≥ 0.2 or equivalent)
^	Indicates an item was asked of one half of the sample
~	Indicates an item was asked of one quarter of the sample
>	Significantly higher than the reference group in the regression analysis
<	Significantly lower than the reference group in the regression analysis
ns	Not significantly different to the reference group in the regression analysis
REF	The reference group in the regression analysis, which was compared to all other groups for a variable

Key terms

Term	Definition
Affirmative consent	The free choice to engage in a sexual activity involving mutual and ongoing communication. This definition reinforces that consent cannot be presumed, must be actively sought and actively communicated, and can be withdrawn at any point. In practice, an individual seeking to have sex with another person must obtain clear, expressed consent from them before (and while) engaging in a sexual act (NSW Government Communities and Justice, 2022).
Attitudes	Evaluations of a particular subject (e.g. person, object, concept) that usually exist along a continuum from less to more favourable. The NCAS measures attitudes towards violence against women, including attitudes towards specific types of violence such as domestic violence and sexual violence, as well as attitudes towards gender inequality.
Backlash	The resistance, hostility or aggression with which strategies to redress gender inequality or prevent violence are met by some people in the community (typically a minority).
Benevolent and hostile sexism	Benevolent sexism encompasses attitudes towards women that are seemingly positive but nonetheless imply women’s inferiority to men based on perceptions of women as fragile, emotionally sensitive or needing help and protection.  Hostile sexism encompasses overtly negative, resentful or misogynistic attitudes towards women who violate traditional gender roles and threaten male dominance. Both forms of sexism serve to justify and maintain the patriarchy and traditional gender roles (Glick & Fiske, 1997).
Bivariate analysis	A statistical analysis that examines the direct or straightforward relationship between two variables only, such as an outcome of interest (e.g. understanding of violence against women) and one other variable (e.g. a demographic factor such as age), without taking into account the effect of any other variables.
Bystander	Somebody who observes, but is not directly involved in, a harmful or potentially harmful event and could assist or intervene (Webster et al., 2018a).
Bystander response	How bystanders react to witnessing a scenario such as disrespect or abuse. The NCAS examined whether bystanders would be bothered by various scenarios and whether they would intervene. Prosocial bystander actions attempt to improve the situation and can include confronting the perpetrator’s unacceptable, gendered and violence-condoning attitudes and behaviour, as well as supporting the victim and survivor. In this report, the two prosocial responses examined were showing disapproval then and there or showing disapproval in private later.
Cisgender	People who identify their gender as matching the sex that was recorded or presumed for them at birth (Transhub, 2021). For further information on the classification of cis and trans respondents in this survey, see Section 2.2.
Coercive control	A pattern of behaviours used to manipulate, intimidate, isolate and control a partner and create an uneven power dynamic in the relationship (Council of Australian Governments [COAG], 2022; Meeting of Attorneys-General, 2022). Coercive control is often a significant part of a person’s experience of domestic violence. A focus on coercive control reflects a shift from specific, isolated incidents (of primarily physical violence) to a recognition that individual acts can be used by perpetrators to form a broader pattern of abusive behaviours that reinforce and strengthen the control and dominance of one person over another (COAG, 2022).
Domestic violence	Refers to violence within current or past intimate partner relationships, which causes physical, sexual or psychological harm. Domestic violence can include physical, sexual, emotional, psychological and financial abuse, and often occurs as a pattern of behaviour involving coercive control. The term “domestic violence” is often used interchangeably with “intimate partner violence”. “Domestic violence” is used in this report, as many historical NCAS items use this terminology to describe violence between partners. ( Note: some broader definitions of domestic violence in the literature include violence between other family members.)
Elder abuse	The abuse or neglect of an elderly person that causes them harm or distress and occurs within relationships of trust that usually involve a power imbalance, including relationships with family, carers, friends and acquaintances (Australian Institute of Health and Welfare [AIHW], 2019b; Qu et al., 2021; World Health Organization [WHO], 2022a). Often the elderly person is dependent on their abuser, such as for assistance with health and care needs, finances or affairs, or to avoid isolation, which creates a power imbalance that can maintain the abuse and deter help-seeking (Adib et al., 2019; Joosten et al., 2017).
Emotional and psychological abuse	Forms of abuse that may include verbal, non-verbal or physical acts by the perpetrator that are intended to exercise dominance, control or coercion over the victim; degrade the victim’s emotional or cognitive abilities or sense of self-worth; or induce feelings of fear and intimidation in the victim (National Family and Domestic Violence Bench Book, 2022).
Family violence	A broader term than “domestic violence”. Refers not only to violence between intimate partners but also to violence between family members. For Aboriginal and/or Torres Strait Islander peoples and communities, “family violence” encapsulates the broader issue of violence within extended families, kinship networks and community relationships, as well as intergenerational issues. “Family” may also refer to “chosen families”, as found in LGBTQ+ communities.
Financial abuse	Also termed economic abuse. A type of violence that often occurs alongside other types of domestic violence, such as physical or emotional abuse. It involves using money in ways to cause harm, such as by withholding funds, preventing a person being involved in financial decisions that affect them, preventing them from getting a job, controlling all household spending and many other tactics to restrict a victim’s and survivor’s freedom and independence.
Gender	The socially constructed and learned roles, norms, behaviours, activities and attributes that a society considers appropriate for people, usually based on their biological sex. Gender has historically been constructed as a binary between “man” and “woman” or “masculinity” and “femininity”, and as a hierarchy of “men” over “women”. These binaries and hierarchies can produce inequalities and discrimination based on gender. As a social construct, gender is not fixed: the acceptable roles and behaviours associated with “man” and “woman” can vary from society to society and can change over time. Gender identities of “man” and “woman” are often associated with the social expectations for members of the biological sex categories “male” and “female”. Where people identify their gender as matching their biological sex assigned or presumed for them at birth, this is called “cisgender”. However, many people do not subscribe to cisgender norms and describe their gender identity in terms that do not accord with the rigidity of the gender binary. For further information on how gender is used in the NCAS survey and this report, see Section 2.2.
Gendered drivers of violence	The underlying causes that create the necessary conditions in which violence against women occurs. The drivers relate to the particular structures, norms and practices arising from gender inequality in public and private life, as well as from other forms of social discrimination and oppression against certain groups of women, including racism, classism, ableism, ageism, heteronormativity and cissexism, etc.
Gender equality	Relates to equal opportunities for all genders to access social, economic and political resources, including legislative protection. Effectively, it describes equality of opportunity.
Gender-ignoring	A perspective that focuses on the importance of being “fair” by treating everyone the same but fails to recognise the gendered norms and gendered differences within structures and systems that drive gender-based inequalities and violence.
Gender norms and stereotypes	Shared standards of acceptable behaviour and overgeneralised concepts that are associated with genders within a community, culture or group (The Good Society, 2022).
Gender-transformative approaches	Approaches that challenge and attempt to change problematic gender stereotypes, scripts, norms, the gender binary and the gender hierarchy, which facilitate and maintain gender inequality (Our Watch, 2019b, 2021a).
Hegemonic masculinity	A type of masculinity that perpetuates unequal relations between men and women. It involves adhering to and exaggerating stereotypically masculine traits, including aggression and men’s domination (Messerschmidt, 2019).
Heteronormativity	The belief that heterosexuality is the preferred and “natural” sexual orientation, which assumes that gender is binary (i.e. men and women). Heteronormativity functions to legitimise social and legal institutions that devalue, marginalise and discriminate against people who deviate from this normative principle (e.g. gay men, lesbians, bisexuals, trans people; American Psychological Association, 2022). The dominance of heteronormative and cisnormative models of domestic and family violence also makes it harder to recognise this violence in LGBTQ+ communities. This bias can contribute to a culture of silence that leads to LGBTQ+ people staying in abusive relationships and not accessing services and other vital support (LGBTIQ+ Health Australia, 2022).
Heterosexual sex scripts	Socially constructed frameworks or “scripts” that guide sexual activity and sexual behaviour. These scripts dictate what one should be doing as a sexual partner (Simon & Gagnon, 1986) and reinforce the widely and implicitly accepted standards for what sex “should” be and look like (Pham, 2016). While individuals shape their own sex scripts in light of their own identity and experiences, sex script theory argues that sexual partners perform sexual encounters according to highly gendered “roles” within the dominant script. More traditional heterosexual sex scripts position men as the active and aggressive initiators of sex, while positioning women as passive sex objects and gatekeepers. In so doing, these scripts privilege men’s sexuality by prioritising men’s sexual gratification and penile–vaginal penetrative sex as the sex act or “real” sex (S. Jackson, 2006; Medley-Rath, 2007).
Hostile sexism	See “Benevolent and hostile sexism”.
Hypersexuality	An aspect of dominant masculinity whereby men are perceived as having high sex drives and are expected to be sexually demanding and dominant in their sexual relationships with women to demonstrate their masculinity. Hypersexuality is linked to objectifying attitudes towards women and beliefs that privilege men’s entitlement to sex with women.
Intersectionality	The interactions between multiple systems and structures of oppression (such as sexism, racism, classism, ageism, ableism, heteronormativity and cissexism), which can be reflected in policy, practices, services and legal contexts. Intersectionality acknowledges that some people are subject to multiple forms of oppression and the experience is not just the sum of its parts. An intersectional approach is a lens for seeing how various forms of inequality can often operate together and exacerbate each other (Kimberlé Crenshaw quoted in K. Steinmetz, 2020).
Intersex/DSD (disorder/difference of sex development)	A term relating to people born with a variation of sex characteristics that do not fit typical definitions of male or female bodies (Intersex Human Rights Australia, 2022). For further information on the intersex item in this survey, see Section 2.2.
Men	A gender identity. In this report, the term is used for respondents who identified as men when asked to state how they describe their gender.
Microaggressions	Everyday, subtle and sometimes overt, intentional or unintentional interactions or behaviours that communicate some type of bias towards historically marginalised groups, including women. People who enact microaggressions may not even be aware of their bias.
Misogyny	A strong dislike of or contempt for women.
Multiple linear regression analysis	A statistical analysis that examines the relationship of a (continuous) outcome variable of interest (e.g. understanding of violence against women) to  multiple factors (or input variables) considered together (e.g. multiple demographic characteristics). Unlike bivariate analysis, multiple regression analysis has the advantage that it can determine which of multiple factors: are  independently related to or “predict” the outcome variable,  after accounting for any relationships between the factors are  most important in predicting the outcome variable.
Multiple logistic regression analysis	A form of multiple regression where the outcome variable is a dichotomous rather than continuous variable.
Multivariate analysis	A type of statistical analysis that examines the interrelationships between three or more variables.
Non-binary people	A gender identity that sits outside the gender binary of “men” and “women”. The term is often used as an umbrella term that encompasses a range of diverse gender identities. In this report, “non-binary” is used as a collective term for respondents who, when asked to state how they describe their gender: explicitly identified as non-binary provided another response that was consistent with a gender identity outside the gender binary. The latter group of respondents was very small ( n = 3). Because this group was too small to be reported on separately, this cohort of respondents has been included within the umbrella term “non-binary” for the purposes of this report.
Non-physical violence	Forms of violence and abuse which do not involve inflicting or threatening physical harm. These forms can include coercive control, financial abuse, psychological or emotional abuse, spiritual abuse or technology-facilitated abuse, among others.
Normalisation of violence	Where violence is seen and treated as normal or is rationalised or excused as part of everyday life.
Physical violence	The use or threat of physical force with the intent to cause physical or psychological harm, such as physical injury, intimidation or fear. ”Violence against women” is broader than “physical violence” and can include other forms of abuse and coercive control.
Prosocial bystander	A bystander who chooses a prosocial action in response to witnessing disrespect or abuse. See “Bystander” and “Bystander response”.
Rate of false allegations of sexual assault	The empirical evidence indicates that most sexual assault allegations are genuine and false allegations are rare. However, the precise rate of false allegations is difficult to establish due to inconsistent recording and classification, study limitations, and because most sexual assaults go unwitnessed (c.f. Kelly, 2010). Although estimates have varied, a meta-analysis of the higher-quality studies estimated that only 5 per cent of sexual assaults reported to police are false (Ferguson & Malouff, 2016). This figure may underestimate false reports to police as it was based on reports “confirmed” to be either false or genuine. However, estimates of false allegations also typically exclude the vast majority of genuine sexual assaults (about 9 in 10) that go unreported to police (Australian Bureau of Statistics, 2017).
Representative sample	A sample of respondents whose demographic profile is similar enough to that of the broader population to be confident that conclusions about the sample apply to the broader population. Random selection is typically used as a means of achieving a representative sample.
Scale	A psychometrically validated group of survey items that measure aspects of the same construct or topic. In the NCAS, scales are used to summarise and demonstrate understanding and attitudes at an  overall or broad level. In this report, the scales are used to measure or assess overall change in understanding or attitudes over time, relationships between understanding and attitudes, and relationships between understanding or attitudes and other factors (such as demographic factors).
Sexism	Attitudes, stereotypes, prejudice and other cultural elements that promote discrimination based on gender. See also “Benevolent and hostile sexism”.
Sexual assault	A form of sexual violence. Sexual activity that happens where consent is not freely given or obtained, is withdrawn or the person is unable to consent due to their age or other factors. Sexual assault occurs any time a person is forced, coerced or manipulated into any sexual activity, including coercing a person to engage in sexualised touching, kissing, rape and pornography.
Sexual harassment	A form of sexual violence. An unwelcome sexual advance, sexualised comment, intrusive sexualised question, request for sexual favours or other unwelcome conduct of a sexual nature that makes a person feel offended, humiliated or intimidated. Can include, but is not limited to, staring or leering, indecent texts, emails or posts, indecent exposure, inappropriate comments, non-consensual sharing of intimate images and unwanted touching.
Sexuality	The experience of sexual attraction, behaviour and identity (Carman et al., 2021). In this report, when sexuality is discussed in relation to NCAS results, it refers to responses to the item, “How would you describe your sexuality?”, with the stated options of “heterosexual/straight, “lesbian”, “gay”, “bisexual or pansexual”, “queer”, “another term (please specify)”, “prefer not to say”.
Sexual violence	An umbrella term that encompasses sexual activity without consent being obtained or freely given. It occurs any time a person is forced, coerced or manipulated into any unwanted sexual activity, such as touching, sexual harassment and intimidation, forced marriage, trafficking for the purpose of sexual exploitation, sexual abuse, sexual assault and rape.
Significant	Throughout this report, “significant” is used to refer to “statistically significant” results where we can be confident (with 95% certainty) that the difference observed in the survey sample is meaningful and likely to represent a true difference in the Australian population ( p < 0.05) that is not negligible in size (Cohen’s  d ≥ 0.2 or equivalent). Significant findings in this report are denoted by the * symbol.
Social norms	Shared standards of acceptable behaviour that may be an informal understanding within groups or across broader society that govern behaviour, or may take the form of codified rules and conduct expectations.
Split-sampling	A method of maximising the range of topics explored in the survey. It involves randomly allocating or “splitting” the sample into groups and asking these groups specific sets of items to allow more items to be asked in total. The present sample was randomly allocated into four subsets of respondents. Key items were asked of the whole sample. However, certain items were asked of only one of the subsets (i.e. one quarter of the sample) or two of the subsets (i.e. one half of the sample).
Stalking	A form of violence that can occur in person or via the use of technology. It involves a pattern of repeated behaviour with the intent to maintain contact with, or exercise power and control over, another person. Examples of stalking behaviours include tracking or following someone (in person or online) and loitering.
Socioeconomic status of area	An ABS measure of the socioeconomic conditions in geographic areas in terms of people’s access to material and social resources, and their opportunity to participate in society (SEIFA quintiles; ABS, 2018).
Subscale	A component of a psychometrically validated scale that taps into a particular aspect of the construct underlying the scale, such as an aspect of understanding or attitudes towards violence against women or gender inequality. Factor analyses were used to subdivide items within a scale into subscales based on which items were answered most similarly to one another by respondents, most likely because they are more conceptually related. Subscales were also validated using Rasch analysis.
Technology-facilitated abuse	An umbrella term used to refer to forms of abuse where technology is the conduit or means of enacting or exercising abuse. Examples of technology-facilitated abuse include harassment, stalking, impersonation and threats via technology, as well as image-based abuse and other forms of abuse online (eSafety Commissioner [eSafety] 2022a; Powell & Henry, 2019).
Time series analysis	Comparison of results over several waves of the NCAS. The results are compared at the scale level and the individual item level. Where possible, the results are compared across four waves of the NCAS: 2009, 2013, 2017 and 2021.
Thematic examination	A type of analysis used to draw out qualitative themes in survey items. This approach is based on identifying and interpreting patterns of meaning within data.
Transgender	“Trans” is an inclusive umbrella term meaning people whose gender is different from the sex recorded or presumed for them at birth and is not contingent on how they socially, medically or legally affirm their gender (Transhub, 2021). For further information on the classification of cis and trans respondents in this report, see Section 2.2.
Trauma-informed care	A strengths-based framework that is grounded in an understanding of and responsiveness to the impact of trauma. It emphasises the physical, psychological and emotional safety of victims and survivors, as well as first responders and service providers, and creates opportunities for survivors to rebuild a sense of control and empowerment (Hopper et al., 2010).
Univariate analysis	The data analysis of a single variable or item. For example, the frequency distribution of gender.
Victims and survivors	Refers to those who have experienced violence. We use this term to recognise both the harm experienced and the resilience of those who experience violence. The term recognises the diverse experiences of violence, although we acknowledge that not all people who experience violence will use this term to describe themselves.
Violence against women	Violence that is specifically directed against a woman because she is a woman or that affects women disproportionately. It includes any act of violence based on or driven by gender that causes, or could cause, physical, sexual or psychological harm or suffering to women, including threats of harm or coercion, in public or in private life.
Women	A term describing a gender identity. In this report, the term is used for respondents who identified as women when asked to state how they describe their gender.

About this report

This report is one among a suite of ANROWS resources produced for the 2021 NCAS. Other reports and documents on NCAS findings include:

• Minter, K., Carlisle, E., & Coumarelos, C. (2021). Chuck her on a lie detector”: Investigating Australians’ mistrust in women’s reports of sexual assault (Research report, 04/2021). Sydney: ANROWS.
• Carlisle, E., Coumarelos, C., Minter, K., & Lohmeyer, B. (2022). “ It depends on what the definition of domestic violence is”: How young Australians conceptualise domestic violence and abuse (Research report, 09/2022). ANROWS.
• Coumarelos, C., Weeks, N., Bernstein, S., Roberts, N., Honey, N., Minter, K., & Carlisle, E. (2023).  Attitudes Matter: The 2021 National Community Attitudes towards Violence against Women Survey (NCAS), Summary for Australia. ANROWS.
• Coumarelos, C., Honey, N., Ward, A., Weeks, N., & Minter, K. (2023).  Attitudes matter: The 2021 National Community Attitudes towards Violence against Women Survey (NCAS), Technical report. ANROWS.
• Attitudes matter: The 2021 National Community Attitudes towards Violence against Women Survey (NCAS), Findings for Aboriginal and/or Torres Strait Islander respondents (forthcoming).
• Attitudes matter: The 2021 National Community Attitudes towards Violence against Women Survey (NCAS), Findings for Australian states and territories (forthcoming).
• Attitudes matter: The 2021 National Community Attitudes towards Violence against Women Survey (NCAS), Findings for people from non-English speaking backgrounds (forthcoming).
• Attitudes matter: The 2021 National Community Attitudes towards Violence against Women Survey (NCAS), Findings for young Australians (forthcoming).

Executive summary

Research design and analysis

The survey sample consisted of 19,100 Australians aged 16 years or over, who were interviewed via mobile telephone. Most mobile numbers in the final sample were selected via random digit dialling (81%), and the remainder were listed mobile numbers.

The 2021 instrument included:

• demographic items
• items measuring understanding of the nature of violence against women
• items measuring attitudes towards violence against women and gender inequality
• scenario-based items examining bystander responses when witnessing abuse or disrespect against women.

Most items were retained from the 2017 NCAS (Webster et al., 2018a) to ensure reliable measurement of changes over time. Some new items were introduced on key and emerging topics of interest, such as technology-facilitated abuse and forms of domestic violence shaped by intersecting inequalities, including disability, ethnicity and sexuality.

Understanding and attitude items were grouped into nine psychometric scales, validated via Rasch analysis and factor analysis. The 2021 NCAS reports on three main scales, namely:

• the Understanding of Violence against Women Scale (UVAWS), which measures recognition of problematic behaviours as violence and understanding of the gendered nature of violence against women
• the Attitudes towards Gender Inequality Scale (AGIS), which measures rejection of problematic attitudes regarding gender inequality
• the Attitudes towards Violence against Women Scale (AVAWS), which measures rejection of problematic attitudes regarding violence against women.

The main scales include subscales that measure different thematic aspects of the broad concepts underlying the scales. In addition, the 2021 NCAS included five scales to measure and allow comparisons between attitudes towards each of five types of violence. These five scales are the Domestic Violence Scale (DVS), the Sexual Violence Scale (SVS), the Sexual Assault Scale (SAS), the Sexual Harassment Scale (SHS) and the Technology-Facilitated Abuse Scale (TFAS).

Respondents’ scores on each scale were used to calculate the average level of understanding of violence against women and rejection of problematic attitudes, as well as changes in understanding and attitudes over time. The proportion of respondents with “advanced” versus “developing” understanding or attitudes according to each scale is also reported. Respondents were classified as having “advanced” understanding if they recognised that all the behaviours measured by a scale “always” or “usually” constitute domestic violence or violence against women. They were classified as having “advanced” attitudes if they “strongly” or “somewhat” disagreed with all the problematic attitudes measured by a scale. ^[1] Bivariate and regression analyses were also conducted to examine the factors significantly related to understanding and attitudes regarding violence against women and gender inequality, including demographic factors and particular aspects of understanding and attitudes.

Key findings and implications

Benchmarking understanding and attitudes over time

Understanding and attitudes regarding violence against women are improving slowly, but further progress is needed.

There has generally been slow but statistically significant improvement in community understanding of violence against women and attitudinal rejection of gender inequality and violence against women since 2013, according to all NCAS scales. Most scales also showed statistically significant improvement between 2017 and 2021, indicating improvement in understanding of violence (UVAWS), rejection of gender inequality (AGIS) and rejection of sexual violence (SVS). However, although rejection of domestic violence (DVS) was stronger in 2021 compared to 2013, it plateaued between 2017 and 2021.

There is room for further progressive change across the Australian population, as fewer than half of the respondents demonstrated “advanced” understanding of violence against women or “advanced” rejection of problematic attitudes regarding gender inequality and violence against women.

Further, while there was high recognition that violence against women is a problem in Australia (91%), there was less understanding that violence against women is a problem in one’s own suburb or town (47%). This finding suggests a misconception that violence tends to occur generally outside one’s own networks, rather than everywhere, which may impede recognition that violence is a community-wide problem requiring action at all levels of society.

Understanding of violence against women

Understanding of the diverse forms of violence against women has slowly improved but there is less recognition of non-physical abuse and coercive control than physical forms of violence.

Most respondents correctly recognised that both the physical and non-physical behaviours examined “always” constitute domestic violence or violence against women (66–92%). Behaviours threatening physical injury or a forced medical procedure, such as forced contraception or abortion, were the most readily recognised as being domestic violence “always” (81–92%). However, there was less recognition of non-physical forms of domestic violence involving financial and emotional abuse or control, including tracking via technology (66–75%). Violence involving the exploitation of aspects of a partner’s identity or experience, such as chronic health conditions, sexual diversity, religion and migrant status, were also less well recognised (66–73%). Similarly, there is room to further improve understanding that broader violence against women (outside intimate and domestic relationships) includes electronic harassment and abuse, such as via texts, emails, social media and sending unwanted sexual images (68–78% of respondents recognised these as “always” forms of violence against women).

Understanding of the gendered nature of domestic violence lags behind recognition of individual violent behaviours.

Contrary to evidence from police, court and hospital admissions data and victimisation surveys, considerable proportions of respondents incorrectly believed that men and women equally perpetrate domestic violence (41%) and equally experience physical harm (21%) and fear (28%) from domestic violence. Thus, a concerning portion of the population may be conceptualising domestic violence through a “gender-ignoring” lens, which focuses on the importance of being fair by apportioning blame equally to each gender but fails to recognise the gendered norms and biases within practices, structures and systems that facilitate gender-based inequalities and violence against women.

Attitudes towards gender inequality

Community attitudes towards gender inequality are slowly improving but some attitudes that support gender inequality persist in a sizeable minority of the population.

Most Australians reject attitudes that perpetuate gender inequality, including attitudes that reinforce rigid gender roles, limit women’s personal autonomy in relationships, undermine women’s leadership in public life, normalise sexism and deny gender inequality experiences. However, some problematic attitudes persist in all these areas for a concerning minority of Australians. For example, a sizeable minority agreed that women mistakenly interpret innocent remarks as sexist (41%), that women prefer men to be in charge in relationships (19%) and that there is no harm in sexist jokes (15%).

Demographic factors and understanding of violence explain only some of the differences in people’s attitudes towards gender inequality.

Less than half of the variation in attitudes towards gender inequality was explained by understanding of violence against women and demographic factors, suggesting that other factors are more influential in determining these attitudes. Higher understanding of violence against women was associated with significantly higher rejection of gender inequality.

Attitudes towards violence against women

Community attitudes towards violence against women are improving very slowly but some attitudes condoning this violence persist in a sizeable minority of the population.

Overall, attitudes rejecting violence against women have significantly improved since 2009 and 2013. However, between 2017 and 2021, despite a significant increase in the rejection of sexual violence, there was no significant improvement in overall rejection of violence against women, largely reflecting a plateau in the rejection of domestic violence.

Nonetheless, most Australians reject attitudes that support violence, including attitudes that minimise violence and shift blame away from perpetrators, mistrust women’s reports of violence, and objectify women and disregard their consent. However, some problematic attitudes persist in all these areas for a concerning minority of Australians. For example, sizable minorities of respondents agreed with attitudes that:

• mistrust women’s reports of violence, agreeing that women make up or exaggerate claims of domestic violence to gain an advantage in custody battles (37%) or use sexual assault allegations as a way of “getting back at men” or due to regretting consensual sex
  (24–34%)
• objectify women and disregard consent, agreeing that a sexually aroused man may not realise the woman doesn’t want to have sex (25%) and that a woman who gives her partner a naked picture of herself is partly responsible if he shares it without her consent (21%)
• minimise violence against women and shift blame, agreeing that much of what is called domestic violence is a normal reaction to day-to-day stress and frustration (23%) and that a woman can make a man so angry he “accidentally” hits her (19%).

Attitudes towards violence against women are closely related to attitudes towards gender inequality and modestly related to understanding of violence and demographic factors.

Attitudes towards gender inequality (AGIS scores) were the strongest significant predictor of attitudes towards violence against women (AVAWS scores). These results indicate that people with higher rejection of gender inequality also tend to have higher rejection of violence against women. Although understanding of violence against women and demographic factors were also significant predictors of attitudes towards violence, their contribution was smaller.

Types of violence against women

Attitudes towards diverse types of violence show some improvement, but challenges remain.

Australians’ attitudinal rejection of sexual violence, including sexual assault and sexual harassment, improved between 2017 and 2021. Although attitudinal rejection of domestic violence was higher in 2021 than in 2013, there was no significant improvement between 2017 and 2021. Nonetheless, in 2021 all types of violence examined by the NCAS were rejected to a similar degree. Concerningly, however, various myths and misconceptions about each type of violence are held by a minority of the community, as outlined below.

Domestic violence

Misconceptions about domestic violence are evident among a minority of the community.

These misconceptions include:

• violence can be justified or excused in certain circumstances (6–23%)
• it is easy to leave violent relationships (6–25%)
• domestic violence is a matter that should be handled privately or within the family (2–12%).

Many Australians do not know how to access domestic violence services.

Two in five respondents indicated they would not know where to go if they needed outside support for someone experiencing domestic violence.

Sexual assault

Problematic myths and stereotypes about sexual assault, sexual consent, and victims and survivors are evident among a sizeable minority of respondents.

For example, some respondents agreed with:

• hostile gendered stereotypes of women as malicious, vengeful and untrustworthy, who lie about sexual assault as a way of “getting back at men” (34%) or because they later regret consensual sexual interactions (24%)
• problematic heterosexual sex scripts that privilege men’s entitlement to sex and position women as the “gatekeepers” who must resist men’s advances, including attitudes that disregard consent because an aroused man “may not realise” the woman does not want to have sex (25%)
• rape myths that sexual assault is primarily committed by strangers (18%) or that “genuine” sexual assault victims immediately report their assault to police (7%) and have evidence of physical injuries (5%).

Sexual harassment

Misunderstanding of sexual harassment as flattering, benign or warranted persists among some Australians.

Some respondents shifted blame to victims and survivors for sexually harassing behaviours involving non-consensual sharing of an intimate image (21%) or touching (10%) or minimised the seriousness of sexual harassment. In addition, a minority of respondents agreed that some non-consensual sexual behaviours that objectify women and disregard consent are “flattering” and desirable, including cat calls (13%) and the uninvited persistent pursuit of a woman (13%).

Technology-facilitated abuse

A minority of Australians do not appreciate the gravity and impacts of technology-facilitated abuse.

Most Australians (89%) are aware that it is a criminal offence to post or share a sexual picture of an ex-partner on social media without their consent. However, the seriousness and psychological impact of technology-facilitated abuse on victims and survivors is not appreciated by some Australians. For example, a minority of respondents:

• minimised the seriousness of technology-facilitated abuse, agreeing that consent could be disregarded in some circumstances, such as when a woman sends an intimate image to her partner and he shares it without her consent (21%)
• did not recognise some forms of technology-facilitated abuse, such as sending an unwanted sexual picture (9%) and targeting women on social media (6%).

Stalking: Technology-facilitated and in person

Most, but not all, Australians recognise stalking behaviour.

Most respondents recognised technology-facilitated and in-person stalking as violence always or usually (83–89%). However, a minority did not recognise this behaviour as violence against women or domestic violence (4–7%).

Bystander responses

Most Australians would intend to intervene prosocially in response to witnessing abuse and disrespect but prosocial bystander intervention is context-dependent.

The NCAS asked respondents how they would react if they witnessed a sexist joke or verbal abuse. Most respondents indicated they would respond prosocially by saying something to show their disapproval when witnessing a friend verbally abusing their partner (92%) or if a work friend (59%) or boss (63%) told a sexist joke. However, prosocial bystander responses depended on:

• the type of behaviour, with virtually all respondents being bothered by verbal abuse (99%) but significantly fewer respondents being bothered by sexist jokes told by a friend (69%) or a boss (86%)
• the presence of a power differential between the bystander and the perpetrator, with significantly fewer respondents saying they would show public disapproval to a boss (35%) than a friend (58–64%), despite more being bothered by the boss scenario
• anticipated peer support, with significantly more respondents saying they would show public disapproval if they anticipated peer support rather than criticism or silence
• the gender composition of respondents’ networks, with respondents, particularly men, who had men-dominated occupations and social networks being significantly less likely to report prosocial bystander responses
• attitudes and understanding, with respondents being significantly more likely to be bothered by sexist jokes if they had higher rejection of gender inequality and recognised that violence against women is a problem in Australia.

Prosocial bystander responses can be impeded by multiple barriers, including personal, context-specific and structural barriers.

The most commonly reported barriers by respondents who said they would be bothered by the abuse or disrespect but would not intervene included fear of negative consequences (75–91%), feeling uncomfortable (75–79%), not knowing what to say (60–62%), feeling it would make no difference (34–52%) and feeling that it was not one’s business to intervene (30–58%). These barriers reflect context-specific and structural barriers, as well as personal skills such as confidence and competence to intervene.

People and contexts

Understanding, attitudes and bystander responses relevant to violence against women are related to multiple, complex factors, including demographic factors. However, demographics explain only a fraction of the picture.

Regression analysis revealed that demographic factors were statistically significant predictors of respondents’ understanding of violence, attitudes towards gender equality and towards violence against women, and bystander responses. However, together, all the demographic factors examined explained no more than 20 per cent of the differences (i.e. the variance) in respondents’ understanding, attitudes and bystander responses, suggesting that the majority of these differences (at least 80%) are explained by other factors. Each individual demographic factor explained no more than 5 per cent of these differences. Demographic factors were less closely related to attitudes towards violence against women than were attitudes towards gender inequality. The modest effect of demographic factors in predicting understanding, attitudes and bystander responses should be kept in mind when reviewing the results below.

Gender was the most important demographic predictor of understanding of violence against women and attitudes towards gender inequality. Women and non-binary respondents demonstrated higher understanding of violence against women and higher rejection of gender inequality compared to men.

Age was the strongest significant demographic predictor of attitudes towards violence against women, with 25- to 34-year-old respondents demonstrating significantly higher rejection and respondents older than 75 demonstrating significantly lower rejection of violence against women compared to all respondents on average.

The strongest significant demographic predictor of being bothered by sexist jokes was gender, with women being more likely to be bothered than men. The strongest significant demographic predictor of bystander intention to intervene prosocially depended on the type of disrespect or abuse that the bystander witnessed. When a friend told a sexist joke, respondents with gender-balanced social networks were more likely to intervene. When a boss told a sexist joke, younger respondents aged 16 to 34 years were less likely to intervene and older respondents aged 65 to 74 years were more likely to intervene. When witnessing a friend verbally abusing a partner, employed respondents were more likely to intervene compared to respondents who were retired or unable to work.

1 Introduction: Violence against women and the need for action

The National Community Attitudes towards Violence against Women Survey (NCAS) is the world’s longest-running survey of such attitudes. Six iterations of the NCAS have been conducted since its implementation in 1987, including the current iteration conducted in 2021. As a large-scale, representative population survey, the NCAS seeks to benchmark and elucidate the Australian population’s understanding and attitudes regarding violence against women, ^[2]  attitudes regarding gender inequality and the likelihood of a person intervening if they were to witness such violence (Webster et al., 2018a). By monitoring changes in community understanding and attitudes over time, the NCAS provides data on key indicators for prevention and early intervention outlined in the National Plan 2022–2032 (COAG, 2022).

The present report details the 2021 NCAS results for the Australian population. ^[3]  This chapter provides the background context for the NCAS results by outlining the nature of violence against women. The chapter discusses:

• the ongoing climate of violence against women in Australia and the urgent need to reduce and prevent this violence (Section 1.1)
• the multiple factors and intersecting modes of discrimination and oppression that underpin the culture that drives and perpetuates violence against women (Section 1.2)
• the preventability of violence against women and the role of the NCAS in informing prevention initiatives (Section 1.3).

1.1 Climate of violence against women

Across the world, violence against women, including violence within intimate, domestic and family relationships, is a widespread social, health and economic problem (Our Watch, 2021a; World Health Organization [WHO], 2021). Violence against women constitutes a fundamental violation of human rights and exacts a significant cost to individuals and communities. Violence against women takes many forms, including physical, sexual, emotional, psychological, social, cultural, spiritual, financial and technology-facilitated violence, abuse or control. Violence against women also occurs in many different contexts and can be perpetrated by someone known to the victim and survivor or by a stranger. These contexts include homes, workplaces, social environments, the public domain, residential care facilities or institutions, and the virtual or online world (Our Watch, 2021a).

Prevalence of violence against women

Worldwide, more than one quarter (27%) of ever married or partnered women aged 15 to 49 years report being subjected to some form of physical or sexual violence by their intimate partner (WHO, 2021). In the European Union, more than one third (37%) of incidents of physical violence against women take place at home (European Union Agency for Fundamental Rights, 2021).

Violence against women similarly continues to be a pervasive problem within Australia. As Figure 1-1 shows, the prevalence rates of physical violence, sexual harassment, sexual violence and emotional abuse against Australian women are alarmingly high (Australian Bureau of Statistics [ABS] 2017; Australian Human Rights Commission [AHRC], 2018a). One Australian woman is murdered by her intimate partner every 10 days (Serpell et al., 2022). National victimisation statistics for 2016 also indicate that 31 per cent of women had experienced physical violence, with physical assault experienced by 27 per cent of women and physical threat experienced by 10 per cent of women (ABS, 2017).

Figure 1-1: Estimated prevalence of distinct types of violence against women in Australia

Estimated prevalence	Type of violence experienced by women since the age of 15 years
1 in 2 women	Sexual harassment by a man or woman
1 in 4 women	Emotional abuse by a current or former partner
1 in 5 women	Sexual violence
1 in 6 women	Physical violence by a partner
1 in 6 women	Stalking

Note: Data from the 2016 Personal Safety Survey (PSS), based on prevalence since the age of 15 (ABS, 2017).

Gendered nature of violence against women

Across the world, population-level data confirms domestic violence is predominantly gendered. Women are overwhelmingly the victims of violence in intimate relationships and men are overwhelmingly the perpetrators of this violence (European Union Agency for Fundamental Rights, 2014; WHO, 2021).

In Australia, population-level health data and victimisation surveys similarly demonstrate that men are the main perpetrators of interpersonal violence and women are more often the victims (ABS, 2017; Australian Institute of Health & Welfare [AIHW], 2022a, 2022b; Serpell et al., 2022). Additional analysis of the 2012 Personal Safety Survey (PSS) indicated that 94 per cent of women who experienced violence since the age of 15 did so at the hands of a man (Diemer, 2015; The Men’s Project & Flood, 2018). In addition, men are also more likely to perpetrate acts that result in serious injury or fatality (AIHW, 2019b). Compared to Australian men, Australian women are:

• almost three times more likely to experience violence by a current or former partner (AIHW, 2022b)
• about four times more likely to experience sexual violence (ABS, 2017; AIHW, 2022c)
• more than eight times more likely to experience sexual violence by a partner (ABS, 2017)
• almost one and a half times more likely to experience emotional abuse (ABS, 2017)
• more than six times as likely to be hospitalised as a result of domestic violence perpetrated by a spouse or domestic partner (AIHW, 2022a, 2022c)
• almost four times more likely to be murdered by a partner (Serpell et al., 2022). ^[4]

Demographic factors correlated with risk of victimisation

The intersections of a range of structural and systemic forms of oppression and discrimination produce particular forms and patterns of violence against women, increase the prevalence or severity of this violence, and limit or undermine individual and systemic consequences for the use of this violence (see also Section 1.2). A wide range of demographic factors have been associated with increased risk of women experiencing violence, including cultural, ethnic, age, ability, gender and sexuality factors (Kulkarni, 2019; K. Morgan et al., 2016; Our Watch, 2021a; Our Watch et al., 2015; Sokoloff & Dupont, 2005; Thiara et al., 2011).

Risk of violence: Race and ethnicity factors

All forms of violence against Aboriginal and/or Torres Strait Islander women occur at higher rates and are more likely to result in severe impacts than violence against non-Indigenous Australian women (ABS, 2017; AIHW, 2018; Bartels, 2010; Closing the Gap Clearinghouse, 2013; eSafety Commissioner [eSafety], 2017; Our Watch, 2018b; Powell et al., 2022). In 2016–17, Aboriginal and/or Torres Strait Islander women aged 15 and over were 34 times more likely to be hospitalised for domestic or family violence compared to other Australian women, with the rate being even higher for Aboriginal and/or Torres Strait Islander women living in remote areas (AIHW, 2019b). Aboriginal and/or Torres Strait Islander women also experience online harm and abuse at much higher rates than the general population (eSafety, 2017). Although the internet and mobile phones are an important source of connection and support for women living in remote areas, inadequate support and education relating to identifying technology-facilitated abuse and a lack of accessible services compound their risk of victimisation (C. Brown et al., 2021). Violence against Aboriginal and/or Torres Strait Islander women is perpetrated by both Indigenous and non-Indigenous men and is linked to the impacts of colonisation, as will be discussed further in the forthcoming 2021 NCAS Aboriginal and/or Torres Strait Islander report.

Women from some cultural or religious backgrounds are at heightened risk of specific forms of violence that, while illegal and unacceptable in Australia, are still carried out based on specific cultural or religious imperatives in some contexts. Such culturally sanctioned forms of violence and abuse include forced and subservient marriage, marital rape, dowry-related violence, female genital mutilation and child marriage (Adinkrah, 2011; Gethin, 2019; Lyneham & Bricknell, 2018; Ogunsiji et al., 2018; WHO, 2022b).

Forced marriage in Australia became an offence under Commonwealth law in 2013. The number of forced marriage referrals to the Australian Federal Police (AFP) has been increasing with growing awareness of the issue. In the 2018–19 financial year, there were 91 referrals compared to 11 in 2013–2014, when the offence was first introduced (AFP, 2019). In 2020–2021, 51 per cent of forced marriage reports involved victims under the age of 18 years (AFP, 2021).

Risk of violence: Age

While women can experience violence across their lifespan, research suggests some differences in the types of violence experienced by women at different life stages, with increased prevalence of particular types of violence at certain ages.

Younger women are at higher risk of many forms of violence, including stalking, sexual harassment, sexual assault and intimate partner violence, compared to both younger men and older women. According to the 2016 PSS, women aged 18 to 24 had the highest rates of experiencing stalking by male perpetrators, sexual harassment and intimate partner violence over the previous 12 months (ABS, 2017, 2019b). Women aged 25 to 34 are also more likely to be hospitalised for assault by a domestic partner than women and men of all other age groups (AIHW, 2022b). In addition, the PSS indicated that women under the age of 35 had the highest rates of sexual assault victimisation over the previous 12 months (ABS, 2017; AIHW, 2019b). Similarly, the Australian Longitudinal Study on Women’s Health found that unwanted sexual activity was more likely to be experienced by younger women aged 18 to 23 (27%) than women aged 62 to 67 (10%; AIHW, 2019b). Previous studies have also shown that victimisation (and multiple victimisation) is increasingly common among college and university students (Cénat et al., 2021; DeKeseredy, Schwartz, et al., 2018; Heywood et al., 2022; Sabina & Straus, 2008; Snyder et al., 2018).

Some young women are also at risk of experiencing specific types of violence related to both their age and cultural background, such as child or early marriage, forced marriage or female genital mutilation (AIHW, 2018).

Elder abuse has increasingly been conceptualised as a form of violence or abuse that can include some distinctive features that are less likely to be evident in forms of violence or abuse experienced at younger ages. Elder abuse is typically defined as mistreatment or neglect of an elderly person that causes them harm or distress and occurs within relationships of trust that usually involve a power imbalance, including relationships with family, carers, friends and acquaintances (AIHW, 2019b; Qu et al., 2021; WHO, 2022a). Often the elderly person is dependent on the abuser, such as for assistance with their health, care needs, finances or affairs, or to avoid isolation, which creates a power imbalance that can maintain the abuse and deter help-seeking (Adib et al., 2019; Joosten et al., 2017; Qu et al., 2021). According to the National Elder Abuse Prevalence Study, Australian women have significantly higher rates of elder abuse of any type compared to Australian men, although there are differences in the prevalence of the different types of abuse, with women being more likely to experience neglect, sexual abuse and psychological abuse, and men being more likely to report financial and physical abuse (Qu et al., 2021). There is also emerging evidence that older women may be more likely to experience specific types of violence because of economic dependence on male partners and lifetime economic inequalities that lead to poverty and insecure housing (United Nations Department of Economic and Social Affairs, 2013).

Risk of violence: Sexuality and gender identity and experience factors

In Australia, prevalence data on violence against lesbian, gay, bisexual, trans, queer and questioning, and other gender- and sexuality-diverse (LGBTQ+) people has only begun to emerge relatively recently, after the lack of inclusion of sexuality and diverse gender identity options in health services data and population research and data collections such as the Census (AIHW, 2022d; Campo & Tayton, 2015b; LGBTIQ+ Health Australia, 2021). However, evidence over the past decade indicates that LGBTQ+ people are more likely to experience sexual violence and family violence and are also less likely to recognise, report and receive appropriate support in response (DeKeseredy et al., 2021; Edwards, Sylaska, Barry, et al., 2015; Edwards, Sylaska, & Neal, 2015; Horsley, 2015; Messinger, 2017; Peitzmeier et al., 2020; Snyder et al., 2018).

A national survey of 6,835 LGBTQ+ Australians in 2020 found that more than 4 in 10 (42%) respondents reported ever being abused in some way by their partner and almost half (49%) reported ever being coerced or forced into sexual acts by their partner (A. O. Hill et al., 2020). Further, cis men were most often the perpetrators of intimate partner violence (57%) and sexual assault (84%; A. O. Hill et al., 2020). The Australian Longitudinal Study on Women’s Health reported that women identifying as bisexual or as mainly or exclusively lesbian were more likely to report having experienced sexual violence in their lifetime than those who identified as mainly or exclusively heterosexual (Townsend et al., 2022). Trans and gender-diverse people aged 16 and over reported experiencing sexual assault or coercion at rates that were nearly four times higher than the general Australian population (Callander et al., 2019). A national survey in 2017 of 4,122 Australians who were active online found that respondents who identified as lesbian, gay or bisexual (19%) were more likely than heterosexual respondents (11%) to have experienced image-based abuse (eSafety, 2017). Some studies also suggest that LGBTQ+ students, particularly trans students of colour, as well as international students, are at an elevated risk of experiencing sexual and intimate partner violence (Bonistall Postel, 2020; Coulter et al., 2017; DeKeseredy et al., 2021).

Lesbian, bisexual and trans women can experience additional unique forms of violence as a result of their gender identity or sexual orientation, including threats to publicly reveal a partner’s sexual orientation or gender identity, and withholding of a partner’s essential medication or hormones (A. O. Hill et al., 2020). Existing research on the experience of violence by gender has almost exclusively focused on men and women, and has not recognised the full diversity of gender identities (Donovan & Barnes, 2019; McKay et al., 2019).

In 2021, for the first time, the NCAS presents results for non-binary and gender-diverse respondents. ^[5] In addition to providing greater inclusivity in population-level research, this change will contribute to the evidence base on gender diversity and attitudes towards interpersonal violence.

Risk of violence: Disability factors

Evidence indicates that women with disability have an increased prevalence of certain types of violence or abuse (Lund, 2020; Mailhot Amborski et al., 2021; Tomsa et al., 2021). For example, 1 in 3 (32%) Australian women with disability have experienced emotional abuse from a current or previous partner since the age of 15, compared with around 1 in 5 (19%) Australian women without disability (AIHW, 2019b). Further, ANROWS research found that women with disability or illness were more likely to report having experienced sexual violence in their lifetime than those without disability (Townsend et al., 2022).

In addition, elder abuse is also increased for elderly people with disability or poor physical or mental health (Qu et al., 2021).

Impacts of violence against women

Violence against women produces a profound and long-term toll on women’s health and wellbeing, on families and communities, and on our broader society. Table 1-1 describes some of the innumerable individual and broader societal impacts of violence against women.

Table 1-1

The impacts of violence against women

Health and wellbeing	Social and psychological	Economic
Intimate partner violence has significant acute and chronic health impacts on women, with causal links to depressive disorders, anxiety disorders, alcohol use disorders, early pregnancy loss, physical injury, homicide, self-inflicted injuries and suicide (AIHW, 2019a)	Violence against women engenders significant social and psychological costs for victims and survivors, their families and the broader community (KPMG, 2016)	The total economic cost of violence against women in Australia in 2015–16 was estimated to be at least $22 billion, and possibly as much as $26 billion, given the under-representation in national prevalence estimates of Aboriginal and/or Torres Strait Islander women, pregnant women, women with disability and homeless women (KPMG, 2016)
In Australia, over 29,000 people (68% of whom were women) were hospitalised for family and domestic violence between 2010–11 and 2017–18 (AIHW, 2021b)	The potential consequences of violence against women include child abuse and neglect, and adverse impacts on emotional wellbeing, cognitive functioning, learning and the ability to develop positive relationships (AIHW, 2019a; Australia’s National Research Organisation for Women’s Safety [ANROWS], 2018)	The total cost includes costs related to pain, suffering and premature mortality; consumption-related activities (e.g. replacing damaged property, defaulting on debts, moving costs); production and employment; health services; justice and other services; transfer payments (e.g. tax and social welfare costs); and impacts on children witnessing or experiencing domestic and family violence (KPMG, 2016).
In 2016–17, almost 2 in 3 (63% or 2,200) hospitalisations of women due to assault by a partner were for injuries to the head or neck, including brain injuries (AIHW, 2019a)	A study of women who had experienced intimate partner violence found they had increased risk of perpetrating child abuse if their own victimisation had resulted in post-traumatic stress disorder, emphasising the need for timely support for victims and survivors of intimate partner violence (R. E. Anderson et al., 2018)	Victims and survivors are likely to bear about half ($11.3 billion) the total cost (KPMG, 2016)
In Australia, one woman is murdered by her intimate partner every 10 days (Serpell et al., 2022)	Children exposed to domestic and family violence have increased risk of both perpetrating and experiencing such violence as an adult, as well as experiencing adverse psychological health outcomes (Agüero et al., 2022; Orr et al., 2022; Reading, 2008; Wagner et al., 2019)	Experiencing intimate partner violence impedes women’s progress in employment and their long-term career prospects because of time o work and the need to relocate frequently to preserve safety (A. Adams et al., 2012; Franzway et al., 2015; S. Meyer, 2016)
	Both nationally and internationally, domestic and family violence is among the leading causes of financial and housing instability, including homelessness, for women and children (Baker et al., 2010; Postmus et al., 2020; Warren & McAuliffe, 2021)

These insights regarding the prevalence and adverse impacts of violence against women reveal that considerable progress is needed to meet the target of the new National Plan 2022–2032 to end violence against women and children within one generation (COAG, 2022). The central objective of the National Plan 2010–2022 was to realise a sustained and significant reduction in the levels of violence against women in Australia. The new National Plan 2022–2032 seeks to go further with its ambitious “towards zero” agenda and violence reduction target. Unfortunately, this objective is proposed in a context in which understanding and attitudes among many people in the national and international community continue to facilitate, create, reinforce and normalise violence against women, in what can be described as a “climate of violence”.

Key events regarding violence against women since 2017

In many ways, events since the previous iteration of the NCAS in 2017 have amplified the focus on violence against women in Australia and overseas. Table 1-2 presents examples of key events that attracted media and public discussion but is not meant to be an exhaustive list of relevant events. As detailed below, the most noteworthy global event has been the COVID-19 pandemic. While some key events exemplify the culture of violence against women, others constitute important steps towards changing this culture of violence.

Table 1-2

Chronology of key events exemplifying the climate of violence in Australia since 2017

Year	Key events
2017	NCAS: the fieldwork for the 2017 NCAS was conducted Russia decriminalised acts of domestic violence that do not cause severe injuries or are reported only once a year (Margolis, 2017) #MeToo movement, originally conceived by Tarana Burke, gained worldwide impetus after a tweet by Alyssa Milano following sexual abuse allegations against Hollywood producer Harvey Weinstein (Sayej, 2017)
2018	Despite allegations of having perpetrated sexual assault, Brett Kavanaughwas appointed as associate justice of the Supreme Court of the United States (BBC, 2018) Following an initial mistrial in 2017, veteran Hollywood actor Bill Cosby was convicted on three felony counts of aggravated indecent assault in April 2018. This conviction was overturned in June 2021 (Francescani & Fisher 2021) Australia: Following the murders of Eurydice Dixon and Aya Maasarwe in Melbourne, a senior police officer was criticised for stating that women should take steps to stay safe rather than placing the onus on perpetrators (SBS, 2018) Australia: the federal Enhancing Online Safety (Non-consensual Sharing of Intimate Images) Act 2018 was passed, giving eSafety a range of enforcement options to require rapid removal of image-based abuse material and to hold perpetrators to account (eSafety, 2018) Australia: a murder-suicide of seven family members occurred in Margaret River, Western Australia (Carmody, 2018)
2019	Financier Jeffrey Epstein was arrested in New York following allegations of sexual abuse dating back to 2005 (Friedman, 2019) The state of Alabama in the United States banned abortion in all circumstances, including rape and incest, unless the pregnancy poses serious health risks (Elliott & Wamsley, 2019) Australia: National Rugby League player Jack de Belin was suspended while facing sexual assault charges (Dean, 2019) Australia: Brittany Higgins, a Liberal Party staff member, alleged that she was raped by a fellow staff member in the Parliament House office of the Defence Industry Minister, Linda Reynolds. The accused perpetrator denied the allegation (The West Australian, 2022)
2020	The COVID-19 pandemic spread worldwide, resulting in mass lockdowns, restrictions and deaths (Australian Journal of Managed Care, 2021) Following Jeffrey Epstein’s suicide while awaiting trial, the United States Attorney’s Office announced the unsealing of federal felony charges against his partner, Ghislaine Maxwell. The jury trial against her commenced in 2021 (The United States Attorney’s Office Southern District of New York, 2022). Harvey Weinstein was found guilty and sentenced to 23 years in prison (Australian Broadcasting Corporation News, 2020). In an attack in Toronto, Canada, in which a woman was killed and another injured, the perpetrator was eventually charged with terrorist activity due to links to “incel” (misogynistic male extremist) ideology. This case was the first time that criminal charges were laid for incel activity, with the acts being defined as “domestic terrorism” (Goden, 2020) Australia: COVID-19 response measures, including border closures, travel restrictions and home-schooling, were first implemented in some Australian states and territories (Storen & Corrigan, 2020). Australia: recommendations by the NSW Law Reform Commission to reform sexual consent laws were tabled in State Parliament, including that a lack of physical and verbal resistance should not be seen to constitute sexual consent (New South Wales Law Reform Commission, 2020) Mastercard and Visa stopped allowing their cards to be used on Pornhub in 2020 following allegations that Pornhub facilitated and distributed material on child sexual abuse, non-consensual sexual activity, image-based abuse and victims of sex trafficking (Price, 2022) Australia: Hannah Clarke and her three children were brutally murdered by her ex-partner (Robertson, 2020) Australia: Ann-Marie Smith died after being left by carers in the same cane chair for 12 months (Boisvert, 2020) Australia: the Australian Communications and Media Authority found that radio broadcaster Alan Jones breached decency rules when he said that New Zealand Prime Minister Jacinda Ardern “should have a sock shoved down her throat” (Cockburn, 2020, para. 3)
2021	NCAS: The fieldwork for the 2021 NCAS was conducted Ghislaine Maxwell was found guilty on five counts of abuse, including sex trafficking of a minor (Bekiempis, 2021) Virginia Giuffre filed a lawsuit against Prince Andrew, Duke of York, for sexual assault. Giuffre’s lawsuit alleged that she had been forced to have several sexual encounters with Prince Andrew in the early 2000s at the age of 17, after being sex trafficked by convicted sex offender Jeffrey Epstein (Giuffre v. Prince Andrew, 2021) Australia: In March , Attorney-General Christian Porter’s strong denial of allegations that he had committed rape in 1988 as a teenager generated considerable media attention and public debate on sexual violence. The police dropped the investigation into these allegations due to “insufficient admissible evidence” to proceed (BBC News, 2021a) In September, Porter resigned from office after revealing he had accepted an anonymous donation to help cover his personal legal fees (Norman, 2021) Australia: Australian Football League Hawthorn forward Jonathon Patton was stood down after multiple women accused him of inappropriate sexual conduct (Colangelo, 2021) Australia: Activist Grace Tame was named Australian of the Year for her advocacy for survivors of sexual assault (Mitchell & Kelly, 2021) Australia: Television host and former White Ribbon chairman Andrew O’Keefe was charged for domestic assault against his partner (Hislop, 2021) Australia: Thousands marched across Australia for March4Justice in the wake of sexual assault allegations against parliamentary staff (BBC News, 2021b) Australia: The process of criminalising coercive control began in Australian jurisdictions (Department of Communities and Justice, 2021) Australia: NSW Police Commissioner Mick Fuller was criticised for proposing that a sexual consent app could address the high rate of sexual assaults (McGowan, 2021) Australia: An online petition launched by activist Chanel Contos called for better sexual consent education in Sydney private schools and received more than 5,000 accounts of sexual assault (Chrysanthos, 2021)
2022	NCAS: Events in 2022 occurred after completion of the 2021 NCAS fieldwork In 2020, Hollywood actor Johnny Depp lost a libel lawsuit in the United Kingdom against the Sun newspaper following its publication of an article in which actor Amber Heard described her abuse by an unnamed partner. However, in 2022, Depp won his defamation case against Heard in the United States based on the same article, and Heard was publicly vilified, with some commenters suggesting the vitriol represented a #MeToo backlash (BBC News, 2022; Jacobs & Bednar, 2021). The United States Supreme Court overturned Roe v Wade , the 1973 landmark decision that had legalised abortion nationwide, thus facilitating the right for states to ban abortionsin all but a few extraordinary circumstances (Totenberg & McCammon, 2022). Australia: The National Plan 2022–2032 was released (COAG, 2022). Australia: The National principles to address coercive control: Consultation draft was released in September (Meeting of Attorneys-General, 2022). Australia: The coronial inquest into the killing of a 28-year-old woman by her ex-partner in the Northern Territory found that police told her to “stop calling them” (para. 1) five days before she was murdered (Park, 2022). Australia: The federal Online Safety Act 2021 (Cth) commenced on 23 January, enhancing and expanding eSafety’s functions and powers (Lavan, 2021). Australia: Following a mistrial, the Australian Capital Territory Director of Public Prosecutions announced that the plan for a second trial relating to the alleged sexual assault of Brittany Higgins had been dropped after expert medical advice warned it posed a “significant and unacceptable risk” to Brittany Higgins’ life (Grattan, 2022, para. 1). In delivering this announcement, the Director of Public Prosecutions stated, “During the investigation and trial as a sexual assault complainant, Ms Higgins has faced a level of personal attack that I have not seen in over 20 years of doing this work” (Knaus, 2022, para. 8). Following this announcement, Brittany Higgins announced her intentions to pursue damages against two Liberal Party ministers and the Commonwealth (Hartcher, Massola, Clun & Thompson, 2022). In December 2022, Brittany Higgins settled a personal injury claim against the Commonwealth (K. Murphy & Knaus, 2022).

COVID-19 pandemic

For many women, the pandemic coincided with the onset or escalation of violence and abuse. The balance of evidence indicates that the pandemic exacerbated violence against women and its adverse impacts (Boxall & Morgan, 2021b; Dalton, 2020; Gosangi et al., 2020; Kourti et al., 2021). Indeed, some people have described gender-based violence in the era of COVID-19 as a “twin” or “shadow” pandemic (Dlamini, 2021; Pfitzner et al., 2020; Sri et al., 2021). Given the difficulties in reporting victimisation to authorities and in recruiting participants for prevalence studies during lockdowns, estimates vary about the impact of COVID-19 on violence against women. In Australia:

• The ABS reported that the number of police-recorded victims of family and domestic violence–related sexual assault increased by 13 per cent in 2020 (ABS, 2021e).
• A survey of more than 10,000 Australian women aged 18 and over found that around 1 in 10 women had experienced physical violence by their partner since the beginning of the pandemic (AIHW, 2021a).
• An online survey of 15,000 Australian women found that, during a three-month period in the initial stages of the pandemic, 4.6 per cent of respondents reported experiencing physical or sexual violence by a current or former cohabiting partner. Almost 6 per cent of women reported experiencing coercive control and 11.6 per cent reported at least one form of emotionally abusive, harassing or controlling behaviour (Boxall et al., 2020). Notably, two thirds of the women who reported experiencing physical or sexual violence by a current or former cohabiting partner said the violence had started or escalated in the three months prior to the survey (Boxall et al., 2020).
• During COVID-19, eSafety also noted a significant increase in online abuse. From early March 2020, reports to eSafety regarding online harms surged, with reports of image-based abuse almost doubling (eSafety, 2020a).

Several factors may have contributed to the observed increases in violence against women during the COVID-19 pandemic. These factors include situational stressors, such as lockdowns necessitating close, ongoing contact between victims and perpetrator; job losses leading to economic hardship; reduced access to support services (particularly face-to-face services); and a range of other individual exacerbating factors (Boserup et al., 2020; Nancarrow, 2020; Zhang, 2020). The experience of violence was also compounded for many women by the disruption of social and support networks that might ordinarily facilitate external intervention (Boserup et al., 2020; Boxall & Morgan, 2021b; Freeman, 2020; Parkinson, 2019).

Similarly, the pandemic may also have influenced community attitudes towards violence against women, for example, via changes that may have occurred in social and occupational networks or from other changes to activities or lifestyle. While the 2021 NCAS can be used to investigate whether attitudes towards violence against women have changed since 2017, it cannot be used to identify the specific factors responsible for any change in attitudes or the extent of the influence of any factor, such as the COVID-19 pandemic.

Key events since 2017 exemplifying a culture of violence against women

Beyond the pandemic, a context of tolerance, wilful ignorance and endorsement of violence against women has persisted both internationally and within Australia. These attitudes and behaviours were exemplified by a series of high-profile legal cases, legislative changes, incidents of violence and media reports in Australia and overseas (Table 1-2).

Key events since 2017 exemplifying a climate for change

The period since 2017 also saw increased momentum and advocacy with the emergence of pivotal movements and steps towards legislative reforms focused on the rejection of violence against women. The events between 2017 and 2022 have brought violence against women to the forefront of public consciousness. The #MeToo movement spread swiftly and widely across the Internet in 2017, and soon made its way into courtrooms and the broader international community (Chandra & Erlingsdóttir, 2021; Hillstrom, 2019). Created by activist Tarana Burke to generate solidarity among marginalised Black women, the hashtag expanded to become a statement of defiance and a call to action against all forms of gendered violence (Chandra & Erlingsdóttir, 2021). Although the American film producer Harvey Weinstein was not sentenced until March 2020, revelations of his abuse spanning 30 years began to appear years before. Unaware of Burke’s movement, on 15 October 2017, actress Alyssa Milano, an ardent opponent of Weinstein, tweeted a request to her followers: “If you’ve been sexually harassed or assaulted write ‘me too’ as a reply to this tweet” (Gill & Rahman-Jones, 2020).

The tweet inspired a cascade of disclosures about abuse, harassment and sexual assault. The attention and momentum of the movement provided further evidence of the ubiquity of gendered violence across the globe and at every social level. However, while inspiring in its ability to provide a forum for women to speak out about their experiences, not all women or people of all genders necessarily feel the #MeToo movement has offered them the space or the necessary support to speak out (Moran, 2018).

Closer to home, Australian victims and survivors, including Grace Tame, Australian of the Year in 2021, and Brittany Higgins (among many others), promoted awareness and pushed for critical law reforms in the prevention of violence against women. This advocacy in part contributed to the Independent Review into Commonwealth Parliamentary Workplaces by the Sex Discrimination Commissioner. The resulting report found that more than half of all people in Commonwealth parliamentary workplaces have experienced at least one incident of bullying, sexual harassment, or actual or attempted sexual assault (AHRC, 2021). Given this abuse was disproportionately aimed at women staff members and Members of Parliament, the report recommended a host of reforms, including gender targets to address gender inequality in parliamentary workplaces (AHRC, 2021).

Community pressure and advocacy resulted in changes to the way sexual assault is understood, recognised and legislated for in Australia (see e.g. Rape and Sexual Assault Research and Advocacy, 2021; Teach Us Consent, 2021; The STOP Campaign, 2022). Since the 2017 NCAS, most Australian states and territories have amended or reviewed their laws to increase clarity about what constitutes consensual sexual activity. The earliest of these amendments, in New South Wales, require “affirmative” consent; that is, taking active steps to ensure that the other person is a willing participant in any sexual act. The New South Wales amendments also recognise the importance of ongoing sexual consent, noting that either party can withdraw their consent at any time and that consent may need to be verified at each stage of sexual activity. According to New South Wales Attorney General Mark Speakman, “the consent reforms are not just about holding perpetrators to account but changing social behaviour with clearer rules of engagement to drive down the rate of sexual assaults” (NSW Government Communities and Justice, 2022). With similar objectives in mind, other states and territories have also amended or reviewed their sexual consent laws, although the legislative changes to date are not uniform across jurisdictions and a nationally consistent legal definition of sexual consent is yet to be realised (ACT Government, 2022; Premier of Victoria, 2022).

Similarly, with respect to domestic violence, there have been significant shifts towards acknowledging and addressing coercive control as a form of domestic and family violence within legislation in Australia. Recognised as an abusive pattern of behaviour used to establish and maintain power over another person, coercive control can include limiting a person’s access to money, controlling who they see, threats and intimidation, persistent texting and tracking their movements, and a range of other behaviours (COAG, 2022). While coercive control is often a key aspect of intimate partner violence, it can also be perpetrated outside intimate partner relationships, including by extended family members (Langton et al., 2020; Vaughan et al., 2016). Currently, Australia generally only allows redress for coercive control via civil law. However, recognition that coercive control is typically a key and serious aspect of domestic and family violence has led to steps in some Australian jurisdictions to criminalise coercive control. Advocates of criminalising coercive control argue that it would help prevent the escalation of domestic violence and provide better protection for victims. There have also been concerns, however, that criminalisation may be ineffective and may have unintended negative consequences, such as law enforcement unfairly targeting marginalised communities and increased victim reluctance to report domestic violence. The Australian Government’s National principles to address coercive control: Consultation draft was released in September 2022 and aims to facilitate a coordinated national approach to coercive control in terms of criminalisation, as well as primary prevention, early intervention, response and recovery. It provides guidance to states and territories to consider their approaches to coercive control in consultation with victims and survivors and with careful consideration of potential unintended consequences of criminalisation and impacts on their communities (ANROWS, 2021; Meeting of Attorneys-General, 2022).

1.2 Facilitators of a climate of violence

While early research focused on individual pathology as a driver of violence against women, contemporary theory and research recognises that violence against women is a complex problem that is underpinned by multiple factors across all levels of society (Centers for Disease Control and Prevention [CDC], 2022; Heise, 1998; Our Watch, 2021a).

Social ecology of violence against women

The socioecological model of violence against women considers the complex interplay between a multitude of factors across society which can place people at greater risk or buffer them from experiencing or perpetrating violence. As Figure 1-2 shows, violence against women is a consequence of complex interactions among many factors at all the different levels of society: the individual and relationship level, the organisational and community level, the system and institutional level, and the societal level (CDC, 2022; Heise, 1998). The model considers the factors that may increase risk of victimisation and perpetration of violence at each level. It also considers the interaction between different factors, both within and across levels, emphasising that different factors may shape, influence and reinforce one another to together facilitate violence against women.

Crucially, the socioecological model recognises both gender inequality and other inequalities resulting from oppression and discrimination as  key underlying drivers of violence against women. Further, the model allows consideration of how gender inequality and other inequalities intersect and interact to create the broad social context that condones and allows violence against women to perpetuate. Table 1-3 summarises the types of factors at each level of the socioecological model that may facilitate violence against women. The next sections discuss the critical role of gender inequality and other structural inequalities in driving violence against women.

Figure 1-2: The socioecological model of violence against women

Source: Adapted from Our Watch (2021a), p. 34.

Table 1-3

Socioecological factors that contribute to or facilitate violence against women

Societal	System and Institutional	Organisational and community	Individual and relationship
Broad societal factors can facilitate or create a context in which violence is encouraged or inhibited, such as via social and cultural norms that endorse or normalise gender inequality and violence against women (Flood, 2020; Rizzo et al., 2020; Sabol et al., 2020; Tomsen & Gadd, 2019) For example, dominant (or hegemonic) patterns of masculinity associated with control, dominance, aggression and hypersexuality have been found to be associated with violence against women (Collins, 2012; Gallagher & Parrott, 2011; McCarthy et al., 2018; P. K. Morrison et al., 2018; Peralta & Tuttle, 2013; Willie et al., 2018) Societal factors that create the context for the marginalisation and discrimination faced by particular groups of women, including Aboriginal and/or Torres Strait Islander women, women from LGBTQ+ communities, migrant women and women with disability, can also perpetuate violence against women from these communities (C. Brown et al., 2021; Carman et al., 2020; Dyson et al., 2017; Langton et al., 2020; Mailhot Amborski et al., 2021; Our Watch, 2018b; Tomsa et al., 2021) Broad health, economic, educational and social policies can also serve to maintain or disrupt gender, economic and social inequalities (CDC, 2022; H. Lowe et al., 2022)	Formal and informal arrangements in policies, systems and institutions may support and maintain, or challenge, the conditions that facilitate the perpetration or experience of violence, including gender inequality and other intersecting sources of inequality and oppression (Hardesty & Ogolsky, 2020; Our Watch, 2021a; Song et al., 2020) Formal structures include rules and legislation that fail to address violence against women and gender inequality, while informal structures include patriarchal hierarchies that serve to embed and maintain inequalities for women, particularly those who experience intersecting oppressions and discrimination (Our Watch, 2021a; Pease, 2021) Examples at this level include policies and practices that hinder active participation and leadership of women based on sexism, racism, classism, ableism etc (Burton et al., 2020; T. Clark et al., 2021; Hideg & Shen, 2019; Liu, 2021; Our Watch, 2021a; Sokoloff & Dupont, 2005) Similarly, rewarding hegemonic masculinity traits such as hypersexuality, dominance and aggression in systems and institutions creates an environment in which women are targets for objectification, hostility and denigration, increasing the acceptability and likelihood of violence against women (Dahl et al., 2015; Murnen, 2015; Our Watch, 2019b; Pease, 2021; Rizzo et al., 2020)	Organisational and community norms, structures and practices that endorse or fail to challenge gender inequality, other inequalities and violence can influence large numbers of people. Therefore, the characteristics of schools, workplaces and neighbourhoods can increase the likelihood of becoming either a victim or a perpetrator of violence (Banyard et al., 2019; Copp et al., 2019; C. Jackson & Sundaram, 2018; Kidman & Kohler, 2020; Yeo et al., 2021) Dominant forms of masculinity and heteronormativity which are associated with violence can also be expressed and maintained at this level (Carman et al., 2020; The Men’s Project & Flood, 2018) Examples include organisational and social responses to workplace sexual harassment that suggest harassment is based on men’s inability to control their sexual desires or that women should be flattered by male attention (Carman et al., 2020; Hlavka, 2014; E. A. Taylor et al., 2018)	The individual’s unique experiences, attitudes, knowledge, skills and relationships may affect their likelihood of becoming either a perpetrator or a victim of violence (Bell & Higgins, 2015; Cano-Gonzalez et al., 2020; Hamai et al., 2021; Jouriles et al., 2014; Kimber et al., 2015; Ogilvie et al., 2022; Reyes et al., 2017; White & Geffner, 2022) Other individual factors that may be associated with both perpetration and victimisation include alcohol use, income, education level, psychopathology (including depression, anxiety, post-traumatic stress disorder and personality disorders) and poor self-esteem (Armenti et al., 2018; Cortés-Treviño et al., 2022; Graham et al., 2018; Mannell et al., 2021; Renner et al., 2015; Spencer et al., 2019) Similarly, individual attitudes towards gender inequality, rigid gender roles and the use of violence to solve interpersonal disputes may also be associated with the perpetration of violence against women (Flood, 2019b; Latzman et al., 2018; Our Watch, 2019b) At the relationship level, a person’s closest social circle of peers, their partners and their family members influence the person’s behaviour and understanding of violence against women. Specifically, membership in social networks characterised by violence-and rape-supportive norms is associated with increased risk for perpetration among men. These peer associations reinforce a shared hostility and aggression towards women that is associated with violence against women and failure to act prosocially when witnessing this violence (Corboz et al., 2016; DeKeseredy, Hall-Sanchez, et al., 2018; Flood, 2008, 2019a; Ha et al., 2019; Leen et al., 2012)

Note: Informed by the socioecological model of violence against women (CDC, 2022) and Change the Story (Our Watch, 2021a, p. 34).

Gender inequality as a driver of violence against women

Many forms of violence against women, whether physical, sexual, emotional, psychological or economic, are underpinned by gender inequality, which can be manifested in the gender norms, structures, systems and practices that privilege men (Flood, 2019b; Our Watch, 2021a; Webster et al., 2018a; WHO, 2022c). The Change the Story framework notes that important drivers of violence against women:

arise from gender-discriminatory institutional, social and economic structures, social and cultural norms, and organisational, community, family and relationship practices that together create environments in which women and men are not considered equal, and violence against women is both more likely, and more likely to be tolerated and even condoned. (Our Watch, 2021a, p. 36)

Gender inequality is a social problem in which women and men do not have equal social standing, value, power, resources or opportunities in society, providing a key context that facilitates and maintains violence against women (Our Watch, 2021a). Australia lags behind many countries on various indicators of gender equality (AHRC, 2018b; Workplace Gender Equality Agency [WGEA], 2022a). Compared to Australian men, Australian women are paid less, are less likely to hold managerial and senior executive positions, and have less superannuation savings (AIHW, 2016b; Riach et al., 2018; WGEA, 2022a). In addition, inadequate parental leave, inflexible work conditions and sparse advancement opportunities can have significant consequences on women’s financial security by prohibiting career progression and forcing women to change occupations or restart their careers elsewhere, inevitably impacting earnings, savings and overall economic security (Riach et al., 2018; Safe Steps, 2016). Gender inequality can also impact other factors of safety, poverty and housing stability through factors such as commercial rent affordability, limited social housing and rental discrimination against single mothers (Blunden & Flanagan, 2021; S. Meyer, 2016; Rowley & James, 2018; Safe Steps, 2016; Summers, 2022; Warren & McAuliffe, 2021).

Despite these tangible inequities, many men are threatened by women’s attempts to achieve economic, political, social and relational equality (Gotell & Dutton, 2016; Lombardo et al., 2021; Skewes et al., 2018). A recent global study found that, in Australia, 32 per cent of men and 11 per cent of women agreed that feminism has resulted in men losing economic, political or social power, while 22 per cent agreed that gender inequality “doesn’t really exist” (IPSOS, 2022). The report also noted that 14 per cent of Australians agreed that violence against women is often provoked by the victim or survivor and that women often make up or exaggerate claims of abuse or rape (IPSOS, 2022).

Drawing on past NCAS results (Webster et al., 2018a) and a large range of available international and national evidence, the Change the Story framework outlines the key gendered drivers of violence against women (Figure 1-3; Our Watch, 2021a). These gendered drivers of violence include attitudes that condone violence against women, support rigid gender roles, tolerate disrespect or aggression towards women, and endorse limits to women’s decision-making and independence (Our Watch, 2021a). As discussed further below, these gendered drivers are informed by two key operating principles:

• sexist ideology, defined by rigid gendered beliefs that justify existing systems and structures and maintain patriarchal social relations (Our Watch, 2021a)
• misogyny, which functions to enforce patriarchal social relations wherever they are challenged (Manne, 2017).

Figure 1-3: Gendered drivers and reinforcing factors in violence against women

Source: Adapted from Our Watch (2021a), p. 10.

Sexist ideology

The gendered drivers of violence are underpinned by sexist ideology that devalues women and assumes that they are less deserving of respect or independence (Our Watch, 2021a). The normalisation and entrenchment of sexist ideology creates the social gender inequality conditions that increase the likelihood of violence against women (Our Watch, 2021a; WHO, 2022c). Sexism can be overtly “hostile” and misogynistic, or it can be more subtle and seemingly “benevolent”, in that it is enacted under the guise of men’s role to protect and provide for women.

Crucially, attitudes supportive of gender inequality have been associated with the actual perpetration of violence (Ozaki & Otis, 2017; Pöllänen et al., 2018; Reed et al., 2018; Verroya et al., 2022; Wahid et al., 2018). Strong associations have been noted between sexist attitudes and behaviours, forms and patterns of masculinity that promote men’s dominance, and men’s perpetration of violence against women (Chung, 2005; Our Watch, 2019b; Rizzo et al., 2020; Webster et al., 2018a). Similarly, men who adhere to rigid gendered beliefs are more likely to commit violence against women, demonstrate sexist and violence-supportive attitudes and behaviours, and use violence as a means of achieving control in their intimate relationships (Peralta & Tuttle, 2013; Rollero et al., 2019).

Men’s peer relations that normalise disrespect or aggression towards women also function as a gendered driver of violence against women. Research has demonstrated the phenomenon of “coercive joining”, whereby internalisation of antisocial behaviours occurs through daily conversations with peers (e.g. involving sexist jokes, objectification of women, homophobia and adherence to dominant ideas of masculinity; Burrell, 2021; Dishion & Tipsord, 2011; Our Watch, 2021a; Webster et al., 2018a). A longitudinal community study found that coercive relationship conversations with friends at age 16 predicted sexually coercive behaviour at ages 23 and 24 for both men and women (Frías & Angel, 2013).

Traditional gender norms and patriarchal attitudes regarding entitlement and benevolent sexism are factors that can normalise and perpetuate the use of violence against women (Bouffard, 2010; Viki & Abrams, 2002). Specifically, perceptions that women are less capable and need the protection of men and that men are better suited to complex decision-making foster gendered beliefs that prevent women’s independence in both their private and the public domain.

The Change the Story framework also identifies four reinforcing factors which do not drive violence on their own but can contribute to or exacerbate violence against women (Figure 1-3; Our Watch, 2021a):

• Condoning of violence in general. This reinforcing factor “normalises” violence and results in violence being seen as a normal part of everyday life. This normalisation of violence can occur throughout society, including via formal laws and structures, media and public discourse representations, and the responses of families and communities (Bernstein et al., 2022a; Bonomi et al., 2013; Makin & Morczek, 2016; Tranchese & Sugiura, 2021; Wright & Tokunaga, 2016). For example, there is extensive evidence that exposure to media violence can cause both short- and long-term increases in aggressive and violent behaviour (J. J. Allen et al., 2018; Bandura, 1977; L. Berkowitz, 1993; Huesmann et al., 2003). Society’s tolerance for violence reinforces seeing violence and aggression by men as desirable masculine traits (Bernstein et al., 2022a; Bonomi et al., 2013; A. L. Smith et al., 2019).
• Experience of, and exposure to, violence. Direct experience of violence victimisation as a child, as well as witnessing violence against other family members as a child, can have profound and compounding effects and increase the likelihood of further violence victimisation or even perpetration (Flood, 2020; Madruga et al., 2017).
• Factors that weaken prosocial behaviour. A range of factors can have a detrimental impact on social norms, which in turn can reduce the likelihood that people will adopt prosocial behaviours that are intended to benefit others or society as a whole. These factors may increase the risk of experiencing or perpetrating violence in the absence of protective factors and include neighbourhood-level poverty, disadvantage and isolation; environments dominated by men’s peer relations; natural disasters and crises; alcohol use; and gambling (Berdahl, Cooper, et al., 2018; Dowling et al., 2016; Pabayo et al., 2020; Parkinson, 2019; Wilson et al., 2017).
• Resistance or backlash to violence prevention and gender equality. Resistance is a common response to social change among some members of the community and can occur in the form of a negative reaction to the increasing empowerment and agency of women (Bandyopadhyay et al., 2020; Caridad Bueno & Henderson, 2017; Flood, 2019c). Some men become antagonistic and violent towards women partners (or women in general), as they are convinced that improving women’s rights must inevitably come at the expense of their own. Backlash may take the form of denying that the problem of gender inequality or violence against women exists, the disavowal of responsibility, inaction, appeasement, co-option and repression (Flood, 2019c, 2020). These responses may be seen at the individual or group level and are often strongest among members of the privileged group (e.g. men) than the disadvantaged group (e.g. women). Resistance and backlash exist on a continuum and may manifest in diverse ways within an organisational or institutional context. More passive forms of resistance can appear in a diverse range of organisational behaviours and attitudes, practices, structures and systems (Respect Victoria & Our Watch, 2022).

Misogyny

Misogyny is a moral manifestation of sexist ideology and functions to enforce patriarchal social relations wherever they are challenged (Manne, 2017). Specifically, misogyny often represents a more hostile and extreme form of backlash and resistance to gender equality (Respect Victoria & Our Watch, 2022). Similarly, some researchers describe gendered violence as a reliable indicator of the presence of systemic misogyny in society (Flood et al., 2020; Manne, 2017). Characterised by hostility, denigration, objectification and violence towards women, misogyny serves as an organising principle that is revealed by the dominance of men and the subordination of women across society, including in politics, business and popular culture, and in the sphere of private life (Manne, 2017; Tranchese & Sugiura, 2021; Vickery & Everbach, 2018). Some argue that while sexism is a supporting ideology for inequality, misogyny operates to enforce inequality.

Misogyny can be disguised or obvious. Misogyny is made more insidious and more difficult to counter because it is often perpetuated subconsciously. That is, the way individuals are embedded in a culture and internalise its customs and social mores can facilitate their complicity in misogynistic social systems (Manne, 2017). In contrast, other forms of misogyny are more explicit. In recent years, a group of heterosexual men calling themselves “involuntary celibates”, or “ïncels”, have constructed a violent political ideology based on the “unfairness” of desired women refusing to have sex with them (L. Bates, 2021). Incels often endorse notions of white supremacy and believe they are superior to women, who should make themselves sexually available to men (L. Bates, 2021). Their misogynistic ideology has inspired violent attacks, including a 2014 attack in Isla Vista, California, intended to instigate a “war on women” that resulted in six fatalities, and an attack in Toronto in 2018 that resulted in 10 fatalities. Some incel supporters celebrated the Toronto attack, calling for other incels to follow up with “acid attacks” and “mass rape” (Tye, 2021).

It is important to address both extremist and everyday misogyny. Everyday misogyny can manifest as microaggressions and disrespect towards women, both within and outside public discourse. The media is a common everyday source of misogyny and disrespect of women. The objectification and dehumanisation of women has led to violence against women becoming usual in mainstream television, movies, music videos, video games and internet pornography (Beck et al., 2012; Bernstein et al., 2022a; Fox & Potocki, 2016; Kahlor, 2011; Rhodes et al., 2018; Seabrook et al., 2019). The exposure of young people to objectifying, degrading and violent depictions of sexual behaviour via electronic platforms before they are developmentally capable of integrating such exposure into a healthy sexual identity has been raised as an area of particular concern. Such exposure may encourage misogynistic, violent “scripts” in young people’s own sexual behaviour and highlights the importance of age-appropriate sex education (Davis et al., 2018; Flood, 2009; Martellozzo et al., 2016; Massey et al., 2021; Peterson et al., 2022). Further, consumption of aggressive or violent internet pornography has been found to be associated with increased likelihood of perpetrating intimate partner violence (Beymer et al., 2021; Brem et al., 2021; DeKeseredy & Hall-Sanchez, 2017; Tarzia & Tyler, 2020).

Other inequalities as drivers of violence against women: An intersectional approach

As already noted, gender inequality is not the sole driver of violence against women, nor the most important driver of violence and abuse against women in all contexts (Our Watch, 2021a). As Figure 1-4 shows, violence against women occurs within a context of  intersecting and mutually compounding forms of oppression, discrimination, and unequal power and privilege, which operate within and across each level of the social ecology (Our Watch, 2021a). An intersectional approach to violence against women recognises that different forms of oppression and privilege in a society, including those due to gender inequality, interact to produce different life experiences and uniquely different outcomes for diverse groups in society. Whereas a non-intersectional approach incorrectly assumes that all women’s experiences of violence are the same by virtue of their gender  as women and their gender only, an intersectional approach highlights the larger systemic and structural factors that can increase the risk of victimisation among some groups or can act as protective factors. Thus, an intersectional approach focuses on the broad, intersecting influences of violence rather than focusing solely on “static” factors within individuals (Koh et al., 2021).

Figure 1-4: The intersecting drivers of violence against women

Source: Our Watch (2021a), p. 46.

Notably, an intersectional approach posits that different types of oppression, discrimination and subordination can be experienced by some people  simultaneously, rather than as discrete oppressions, and can interact to produce  distinct forms of inequalities for some marginalised groups. It has been argued that “no form of subordination ever stands alone” and no one form of oppression is the same as any other (Matsuda, 1990, p. 1189). An intersectional lens recognises that inequalities, and the abuse and violence that results from these inequalities, can be all at once racialised, gendered, classed, abled, etc. For example, racist gender discrimination, or gendered racial discrimination, can occur differently for people of different genders.

Examples of these distinct, intersecting forms of oppression include:

• the specific form of racist misogyny towards Black women, termed “misogynoir” (Bailey & Trudy, 2018)
• the specifically colonial, racist and sexist dehumanisation of Aboriginal and/or Torres Strait Islander women (Cripps, 2021; Watego, 2021; Watego et al., 2021)
• the specifically infantilising (ageist), sexist and ableist social media backlash against climate activist Greta Thunberg (Park et al., 2021)
• the range of specifically racist, xenophobic, queerphobic and heterosexist microaggressions faced by LGBTQ+ people of colour (Arayasirikul & Wilson, 2019; Nadal, 2019a, 2019b).

Importantly, for some women, intersecting dimensions of oppression can have profound effects on their risk and experience of violence. Intersecting inequalities can increase the prevalence or severity of violence; produce different manifestations of violence and differential outcomes; and weaken individual and structural consequences for the use of violence against marginalised women (Annamma et al., 2018; Carman et al., 2020; Crenshaw, 1989, 1991; Fiolet et al., 2019; Ghafournia & Easteal, 2018; Kulkarni, 2019; Lockhart & Danis, 2010; E. M. Morgan & Zurbriggen, 2016; Our Watch, 2018a, 2018b, 2021a; Our Watch & Women with Disabilities Victoria, 2021; Sokoloff & Dupont, 2005; Thiara et al., 2011). “Demographic factors correlated with risk of victimisation” in Section 1.1 outlines some examples of particular groups of marginalised women who have increased risk of violence overall or increased risk of particular types of violence. Such groups include Aboriginal and/or Torres Strait Islander women, women from certain cultures, women with disability and LGBTQ+ women. Some examples of how different intersecting inequalities can produce specific barriers to help-seeking or different outcomes for particular groups of women are as follows:

• Aboriginal and/or Torres Strait Islander women face elevated rates of child removal and incarceration, as well as greater challenges in accessing support services for domestic or family violence due to inequitable service provision, greater risk of misidentification as aggressors and increased risk of harm from law enforcement institutions (Cox et al., 2014; Cramp & Zufferey, 2020; Fiolet et al., 2019; Langton et al., 2020; Nancarrow et al., 2020; Olsen & Lovett, 2016; Prentice et al., 2016; Spangaro et al., 2016; Walter, 2016; Weldon & Kerr, 2020).
• Migrant and refugee women may fear seeking help for domestic or family violence due to their immigration status and may face challenges in gaining support due to a scarcity of culturally and linguistically appropriate services, institutional racism, a lack of education, more disadvantaged socioeconomic status, and restrictions on health or wellbeing support as a result of their visa status (Femi-Ajao et al., 2020; Fineran & Kohli, 2020; Hulley et al., 2021; L. Murray et al., 2019).
• Ableism can compound the gender inequality experienced by women and girls with disabilty and can result in their sexuality and reproductive rights being dismissed, and in their receipt of limited or negligible sexual and relationships education. Ableism can therefore act as a barrier to women and girls with disability recognising relationship abuse and knowing how to seek assistance (Frawley & Wilson, 2016; Serrato Calero et al., 2020; Stein et al., 2018; Streur et al., 2019).
• Trans people face poorer health outcomes as a result of violence due to service access inequality, gender insensitivity and transphobia, and barriers are further increased for trans women of colour (Callander et al., 2019; Calton et al., 2015; Ussher et al., 2020).

Intersectionality theory is an important consideration when researching and addressing violence against women because it requires “due consideration of the various axes of oppression and privilege” (Srinivasan, 2021, p. 17). A key insight of intersectionality is that singularly focused initiatives and interventions that treat women as a homogenous group can be problematic because this unidimensional focus often serves those who are least oppressed among the group while perpetuating the marginalisation and oppression directed towards others within it (Annamma et al., 2018; Srinivasan, 2021).

Individual attitudes and violence against women

According to the socioecological model, people’s attitudes are an individual-level factor that can interact with a broad range of other factors at different levels of society to facilitate violence against women (Figure 1-2 and Table 1-3). For example, an individual’s attitudes towards violence against women can be influenced by other individual-level factors such as their exposure to family violence and their peer and family relationships (Callaghan et al., 2018; Debowska et al., 2015; Ozaki & Otis, 2017; Seff, 2021). Individual attitudes to violence can also be shaped by, and can reflect, social norms about gender inequality and other inequalities that may be evidenced at the organisational, community, institutional and societal levels.

“Attitudes” are defined as evaluations of a particular subject (e.g. a person, concept, behaviour or event) and usually exist along a continuum from less to more favourable (Fishbein & Ajzen, 2010). Psychological theory describes an attitude as comprising three components:

• a cognitive component, reflecting thoughts and beliefs about the subject
• an affective component, reflecting feelings associated with the subject
• a behavioural component, reflecting the attitude’s influence on actual behaviour (Breckler, 1984; Eagly & Chaiken, 1993).

An attitude may be explicit or implicit – that is, the individual may or may not be consciously aware of their attitude and how it impacts their behaviour. Although attitudes are often enduring, they can also change given that they are a learned tendency to evaluate something in a particular way (Suedfeld, 2017). Thus, problematic attitudes are potentially mutable via new experiences and education (Albarracin & Shavitt, 2018).

Given that some research studies have shown links between people’s attitudes and their behaviour, attitudes provide a possible point of intervention for changing problematic behaviours (Albarracin & Shavitt, 2018; Suedfeld, 2017). However, it is important to note that the relationship between an individual’s attitudes and their behaviour is not straightforward, for a few reasons. First, the motivational bases and characteristics of the attitude, such as its intensity and importance, can affect how much the attitude will impact behaviour. For example, the cognitive basis for the attitude, including the extent and nature of evidence supporting the attitude, and the specific expectations surrounding the attitude can affect whether the attitude will translate into actual behaviour (Kelman, 2017).

Second, attitudes are only one of the factors that can influence behaviour. A prominent theory about the relationship between attitudes and behaviours is the Theory of Planned Behaviour, which has been used to predict a range of health-related behaviours (Ajzen, 1991; Ajzen & Fishbein, 1980). ^[6]  This theory argues that attitudes are one of six factors that can influence whether a person engages in a certain behaviour:

• attitudes: the degree to which a person has a favourable or unfavourable evaluation of the behaviour
• behavioural intention: the motivation (or strength of intention) to perform the behaviour
• subjective norms: the extent to which the individual believes most people would approve or disapprove of the behaviour, particularly peers and other key influences in the individual’s relational circle
• social norms: the customary codes of behaviour (such as among peers or within a larger cultural context, as described in Figure 1-2) which influence an individual’s assessment of the behaviour
• perceived power: perceptions of factors that facilitate or obstruct performance of the behaviour, which inform the person’s perceived control over each of these factors (Figure 1-2 and Figure 1-3)
• perceived behavioural control: the individual’s perception of the ease or difficulty of performing the behaviour, which can vary across situations (e.g. depending on their confidence or the inequalities that may operate in different situations).

Studies have shown that these factors involving attitudes, motivations, perceived norms, and perceived power and control are relevant to prosocial actions of bystanders who witness violence or disrespect towards women. A recent study found that bystander intent to intervene in a sexual assault was positively related to the bystander’s anticipated “efficacy” in the intervention (Papineau, 2020). This finding suggests that increasing bystander efficacy or confidence (e.g. through training) may increase intentions to intervene and prosocial behaviours when witnessing violence against women (Papineau, 2020). In another study examining intentions to intervene in bullying and dating violence, adolescents reported a higher proportion of barriers to acting than facilitating factors, with their perceptions of peer norms and social consequences being among their principal concerns (Casey et al., 2017). These barriers to intervening are discussed further in Chapter 8.

The NCAS examines individual understanding and attitudes regarding violence against women, gender inequality and intentions to intervene as a witness to violence against women. Given that the NCAS is a large-scale representative population survey, it thus provides a snapshot of the “normative” or typical attitudes and understanding of the Australian community about violence against women at a specific point in time. Further, given that attitudes are shaped by, and in part reflect, broader organisational, community, institutional and societal systems and structures, the NCAS functions as a gauge for how Australia is progressing in changing the broader climate that facilitates and maintains violence against women.

1.3 Deconstructing the climate of violence: Prevention

As discussed in Section 1.1, violence against women produces profound adverse consequences for women, their children and our wider society. However, these impacts can be reduced by taking decisive action to prevent violence before it starts, intervening early, responding appropriately to violence when it occurs, and supporting recovery and healing (COAG, 2022). Ending violence against women requires addressing the range of drivers and oppressions that enable and reinforce violence against women, including violence against the most marginalised groups of women who remain over-represented in victimisation data and who confront unique challenges in accessing support and assistance (Kulkarni, 2019; K. Morgan et al., 2016; Our Watch, 2021a; Sokoloff & Dupont, 2005; Thiara et al., 2011). The Change the Story framework recommends 12 types of actions that need to be undertaken to prevent violence against women by addressing the key drivers of violence, as well as the social contexts and reinforcing factors that facilitate violence (Our Watch, 2021a). These actions include challenging the condoning of violence; promoting women’s independence; building social norms that foster healthy personal identities; building healthy masculinities; promoting gender equality; addressing intersectional oppression and discrimination; building safe, fair and equitable organisations and institutions through policy and systems change; strengthening respectful relationships in both private and public life; challenging the normalisation of violence; reducing the impacts of violence; strengthening prosocial behaviour; and addressing backlash and resistance to positive change.

Initiatives for preventing violence against women have traditionally been divided into three types: primary, secondary and tertiary prevention (Our Watch, 2021a; VicHealth, 2017). Consistent with this approach, the National Plan 2022–2032 outlines four “domains” for action to end violence against women but uses more descriptive terminology to refer to the traditional types of “prevention”. Figure 1-5 shows the alignment between the three “traditional” types of prevention and the four domains of the National Plan. The National Plan 2022–2023 domains for action are:

1. Prevention (also described as primary prevention) – working to change the underlying social drivers of violence by addressing the attitudes and systems that drive violence against women and children to stop it before it starts.
2. Early intervention (also described as secondary prevention) – identifying and supporting individuals who are at high risk of experiencing or perpetrating violence and preventing violence from escalating or reoccurring.
3. Response (also described as tertiary prevention) – providing services and supports to address existing violence and support victims and survivors experiencing violence, including via crisis support and police intervention, and fostering a trauma-informed justice system that will hold people who use violence to account.
4. Recovery and healing (also described as tertiary prevention) – helping to reduce the risk of victim and survivor re-traumatisation, and supporting victims and survivors to be safe and healthy, and to recover from trauma and the physical, mental, emotional and economic impacts of violence (COAG, 2022).

For clarity, throughout this report, “primary prevention” is used to refer specifically to actions consistent with Domain 1 (Prevention) from the National Plan 2022–2032. In addition, “prevention” is used as a more general term that can include actions consistent with any, some or all of the domains of the National Plan 2022–2032 (COAG, 2022).

Figure 1-5: Ending violence against women: Prevention, early intervention, response, and recovery and healing

Source: Based on interventions outlined in the Change the Story framework (Our Watch, 2021a, p. 58) and the National Plan 2022–2032 (COAG, 2022).

While recognising and endorsing the drivers and reinforcers of violence and necessary actions articulated by Our Watch, the National Plan 2022–2032 also describes six guiding principles that inform action within the four domains to address violence against women (Our Watch, 2021a; COAG, 2022). The six guiding principles are:

• Advancing gender equality, which recognises that achieving gender equality is fundamental to both advancing human rights for Australians and addressing a key driver of violence against women. The National Strategy to Achieve Gender Equality is a federal government initiative that seeks to address the structural, social and economic barriers to advancing gender equality in Australia (COAG, 2022).
• Closing the Gap, which is an agreement by all Australian governments and the Coalition of Peaks, a representative body of over 80 Aboriginal and/or Torres Strait Islander community–controlled peak organisations and members. The objective of this agreement is to enable Aboriginal and/or Torres Strait Islander peoples and governments to work together to overcome the inequality experienced by Aboriginal and/or Torres Strait Islander peoples, including in relation to violence against women.
• Centring victims and survivors ensures that their lived experiences, perspectives and direct knowledge of the strengths and weaknesses of current systems, structures and interventions is acknowledged, heard and respected as a key ingredient of policy development and reform.
• Accountability, which is an intention to focus attention and expectations on the actions of people who choose to use violence. This involves trust and support for victims and survivors and avoiding victim-blaming in any context. Similarly, perpetrators are to be held accountable and supported to take responsibility for their violence with appropriate legal and social sanctions and consequences.
• Intersectionality, which recognises that violence against women exists in relation to multiple and intersecting structural and systemic forms of discrimination, such as racism, colonialism, ableism, homophobia, biphobia and transphobia, and ageism. This recognises that gender and gender inequality may be constructed and experienced differently and may not be the most significant factor in violence against all women. Actions from prevention through to recovery and healing must therefore respond to the diversity of women and children.
• Person-centred coordination and integration, which strives for trauma-informed, person-focused and holistically integrated responses from the specialised services and systems that support victims and survivors through their recovery and healing.

The National Plan 2022–2032 and the Change the Story framework recognise that efforts to end violence against women must occur at every level of the social ecology, including in key settings such as schools and universities, workplaces, clubs and sporting institutions, the media, and in the justice and health service system (COAG, 2022; Our Watch, 2021a). Table 1-4 outlines some key (but non-exhaustive) examples of prevention strategies that can be undertaken at each level of the social ecology and across the domains of the National Plan 2022–2032 to prevent violence against women. Across these levels, prevention strategies to address harmful systems, structures and norms that perpetuate inequalities, discrimination and oppression can be used to create new shared beliefs, expectations and practices. Research also suggests that violence prevention strategies are necessary to change social expectations and individual attitudes, publicise these changes and spark both the initiation and reinforcement of new norms and behaviours (Alexander-Scott et al., 2016). Activities to achieve these objectives may include schools-based programs to create more respectful and gender-equitable environments, interventions and education aimed at shifting the disrespectful portrayal of women in the media, community education and social marketing campaigns, and workplace initiatives promoting positive bystander responses (AHRC, 2018a; ANROWS, 2019; COAG, 2022; Easteal et al., 2015; Karageorgos & Boyle, 2021; Our Watch, 2021b; Sutherland et al., 2019).

Table 1-4

Strategies for preventing violence against women at different levels of the social ecology

Level	Strategy
Societal	Promote societal norms that reject violence against women, as well as efforts to strengthen women’s financial security, education and employment opportunities as well as their sexual, reproductive and overall health security and autonomy.
System and institutional	Promote women’s economic, legal and societal autonomy and address gender inequality and violence in all aspects of institutional and systems operation. For example, actively encourage women’s leadership and participation in public life by providing childcare support and parental leave that is not gender specific. Embed materials in the Australian school curriculum that address gender inequality and the drivers of violence against women. Use policy and legislative levers, such as the National Strategy to Achieve Gender Equality and the National principles to address coercive control, to address the drivers of violence against women (Meeting of Attorneys-General 2022a; Australian Government, 2022b).
Organisational and community	Ensure organisational, technological and community settings are safe places and promote equality for all people. Create policies and practices to actively encourage women’s participation and leadership, including reforming organisational human resourcing practices, policies and structures that perpetuate the gender pay gap. Encourage community cohesion in adopting prosocial norms and practices that support gender equality and reject stereotyping, discrimination and violence.
Individual and relationship	Promote attitudes, beliefs and behaviours that increase gender equality and prevent violence against women. This promotion may include skills training, social-emotional learning, parenting or family-focused prevention programs, healthy relationships education, programs to promote healthier masculinities and women’s autonomy in relationships, and peer programs to enhance communication and positive peer norms and problem-solving skills.

Note: Adapted from the socioecological model of violence against women (CDC, 2022) and Change the Story (Our Watch, 2021a, p. 34).

Evidence is also growing about increasing the effectiveness of prevention strategies by employing “gender-transformative” approaches. These approaches seek to implement changes across levels of organisations and communities by addressing both individual attitudes and beliefs and broader entrenched social ideologies related to acceptance of rigid gender norms, roles, stereotypes and scripts, such as male entitlement and rigid stereotypes of masculinity (Casey et al., 2018; Flood, 2019b, 2019c). Gender-transformative approaches go beyond simply challenging gender norms, structures and practices and instead seek to transform them in a way that frees all genders from rigid and problematic gender stereotypes. These approaches also consider the intersecting sources of discrimination and inequality that must also be addressed at all levels of the social ecology to truly achieve social transformation.

Alignment of the NCAS with the National Plan

The NCAS is aligned with both the directions set out by the National Plan 2022–2032 and the primary prevention focus of the Change the Story framework (Our Watch, 2021a). The NCAS examines attitudes towards different forms of violence against women, attitudes towards perpetrator accountability and attitudes towards victims and survivors of violence (COAG, 2022; Our Watch, 2021a; Webster et al., 2018a). Also, in accord with the National Plans (2010–2022 and 2022–2032), the NCAS examines a number of social factors that may contribute to violence against women (COAG, 2010b, 2022). Specifically, the NCAS instrument is premised on the idea that achieving the objective of ending violence against women in one generation is facilitated by the population:

• having a strong understanding of the nature of violence against women, including its diverse and nuanced forms (National Plan 2022–2032, “Early intervention key indicators”; COAG, 2022)
• strongly rejecting attitudes that condone gender inequality and violence against women (National Plan 2022–2032, “Prevention key indicators”; COAG, 2022) and
• being prepared to intervene when witnessing violence or abuse against women, when it is safe to do so (National Plan 2022–2032, “Early intervention key indicators”; COAG, 2022).

The next chapter, Chapter 2, outlines the aims and methodology of the 2021 NCAS, with more detail being provided in the Attitudes matter: The 2021 National Community Attitudes towards Violence against Women Survey (NCAS), Technical report (the Technical report; Coumarelos et al., 2023). Chapter 3 provides key benchmarks from the survey regarding community understanding and attitudes regarding violence against women. The report then provides detailed results from the 2021 NCAS for Australia as a whole regarding:

• understanding of violence against women (Chapter 4)
• attitudes towards gender inequality (Chapter 5)
• attitudes towards violence against women in general (Chapter 6)
• attitudes towards specific types of violence against women (Chapter 7)
• bystander responses when witnessing disrespect or violence (Chapter 8)
• understanding and attitudes held by different demographic groups (Chapter 9).

Finally, Chapter 10 provides the implications of the 2021 NCAS results for policy and prevention of violence against women.

2 Research design

2.1 Aims of the 2021 NCAS

The 2021 NCAS had five main aims. The first aim was to benchmark, as at 2021, the Australian population’s:

• understanding of the nature of violence against women
• attitudes towards gender equality
• attitudes towards violence against women
• intentions to intervene if they were to witness abuse or disrespect towards women.

The second aim was to determine if this understanding and these attitudes had improved in the four-year period since the previous NCAS in 2017. ^[7]  Together, the first two aims sought to measure progress towards breaking down the culture that facilitates violence against women.

The third aim of the 2021 NCAS was to identify any specific areas where there are bigger gaps in community understanding or more problematic community attitudes regarding violence against women. This third aim sought to inform education and intervention strategies about areas of particularly high priority.

The fourth aim of the 2021 NCAS was to identify demographic, attitudinal and contextual factors that are associated with problematic understanding and attitudes regarding violence against women. In terms of demographic factors, complementing the results presented in this report, separate papers (forthcoming) will provide additional results for three demographic groups identified as groups of interest in the National Plan 2022–2032:

• young people
• Aboriginal and/or Torres Strait Islander people
• people born in a non-main English–speaking country (N-MESC).

In terms of attitudinal factors that may be associated with a culture of violence against women, the 2021 NCAS examined how attitudes towards gender equality may be linked to understanding and attitudes regarding violence against women.

The final main aim of the 2021 NCAS was to benchmark understanding and attitudes regarding violence against women at the jurisdictional level and report on key results separately for each Australian state and territory (forthcoming report).

Ethics clearance for the project was provided by the University of Sydney Human Research Ethics Committee (ethics project number 2020/650).

2.2 2021 NCAS instrument

The 2021 NCAS instrument consists of 131 items. Figure 2-1 presents the key components of the 2021 instrument. ^[8]  To simplify reporting, each item was assigned an alphanumeric code (e.g. D1). The letter in the code identified the item’s thematic topic:

• D = domestic violence
• B = bystander response
• G = gender inequality
• S = sexual violence
• V = violence against women.

The number corresponds to the order of the items within a thematic topic in the 2021 NCAS instrument.

Demographic items

Self-reported demographic information about respondents was used to explore how understanding and attitudes may vary based on people’s characteristics, backgrounds, contexts and locations. Demographic information was also used to assess how closely the demographic profile of the sample matched that of the Australian population and to make any necessary adjustments through data weighting (Section 2.3). ^[9]

Items and scales measuring understanding and attitudes

The non-demographic items in the 2021 NCAS measure either understanding or attitudes relevant to violence against women. Most of the non-demographic items were grouped into various psychometric scales. Each scale measures understanding or attitudes regarding a particular type of violence against women, violence against women more broadly or gender equality. ^[10]  The strength of psychometrically validated scales is that they can measure a complex overall construct or concept that would be difficult to measure with a single item. As detailed in Section 2.5, analyses were conducted at both the item level and scale level.

The psychometric scales were validated via Rasch analysis and factor analysis. Rasch analysis is a form of statistical analysis that examines whether a scale comprises items that “sit well” together and thus likely measure aspects of the same broader construct.

Nine scales were used in reporting the 2021 NCAS (Figure 2-1). These nine scales can be categorised into three groups:

• the Gendered Violence and Inequality Scale (GVIS): the GVIS is an overarching “mega scale” that includes all knowledge and attitude items that sit in one of the other eight scales
• three “main scales”, namely:
• • the Understanding of Violence against Women Scale (UVAWS)
  • the Attitudes towards Gender Inequality Scale (AGIS)
  • the Attitudes towards Violence against Women Scale (AVAWS)
• five “type of violence scales”, namely:
• • the Domestic Violence Scale (DVS)
  • the Sexual Violence Scale (SVS), which consists of the Sexual Assault Scale (SAS) and Sexual Harassment Scale (SHS)
  • the Technology-Facilitated Abuse Scale (TFAS). ^[11]

The three main scales also comprise several subscales, which were identified via factor analysis. Factor analysis examines the relationships between item responses to identify whether the items in a scale can be grouped into different themes within the broader construct measured by the scale. Thus, the subscales allow for a closer investigation of the key themes within each of the main scales. ^[12]

Scale and subscale scores

Rasch analysis was used to compute a (rescaled Rasch) score for each respondent on each of the nine scales, based on their answers to the items within the scale. Scores on each scale could range from 0 to 100, with higher scores representing higher understanding of violence or higher rejection of problematic attitudes.

Similarly, each respondent also received a (rescaled Rasch) score on each subscale in the UVAWS, AGIS and AVAWS, based on their answers to the items in the subscale. Subscale scores could range from 0 to 100, with higher scores indicating stronger understanding or stronger rejection of problematic attitudes. Further information about each scale is provided below.

Gendered Violence and Inequality Scale (GVIS)

Higher scores on the GVIS indicate stronger understanding of violence against women and stronger rejection of gendered violence and gender inequality. The purpose of the GVIS was to anchor all other scales to each other so that they can be compared. All items that sat with one of the remaining eight scales also sat in the GVIS overarching scale (Figure 2-1). In 2021, the GVIS was rescaled to a theoretical range of 0 to 100, and all other scales were anchored to it. ^[13]

Main scales

The three main scales were based on the 2017 NCAS scales and, together, contain all the items in the GVIS. Items in the three main scales were mutually exclusive. That is, each item sat in only one of the three main scales. Figure 2-1 details the subscales of each main scale. Higher scores on the main scales and their subscales indicate higher understanding of violence against women (UVAWS), higher attitudinal rejection of gender inequality (AGIS) and higher attitudinal rejection of violence against women (AVAWS).

Type of violence scales

The five scales regarding different types of violence were developed for the 2021 NCAS. These type of violence scales predominantly draw on items from the AVAWS, with a small number of items from the UVAWS. The AVAWS and UVAWS combine items on different forms of violence against women, including domestic violence, sexual violence, technology-facilitated abuse and stalking. The type of violence scales were developed because policymakers, practitioners and researchers may be interested in the more specific attitudes that may relate to each type of violence against women, even though these types of violence can overlap. Examples of overlaps include that sexual violence can occur within or outside domestic relationships, and that technology-facilitated abuse can include domestic abuse, sexual abuse or abuse that is neither of a domestic nor sexual nature. Note that there were insufficient items (three items) to develop a separate scale on stalking.

Figure 2-1

Components of the NCAS instrument, 2021

Note: DV = domestic violence; VAW = violence against women.

Types of violence scales (continued)

As Figure 2-1 shows, the Technology-Facilitated Abuse Scale (TFAS) includes understanding and attitude items drawn from the UVAWS and AVAWS. All other type of violence scales consist only of attitude items from the AVAWS. Together, the Domestic Violence Scale (DVS) and the Sexual Violence Scale (SVS) comprise almost all of the items in the AVAWS (41 of 43 items).

As some items were relevant to more than one type of violence, they were included in multiple type of violence scales. Items from the Sexual Assault Scale (SAS) and Sexual Harassment Scale (SHS) were combined to form the Sexual Violence Scale (SVS). In addition, two attitude items were included in both the SVS and the TFAS.

The DVS and SVS do not have any overlapping items. The four items about sexual violence within a domestic relationship are included only in the SVS, not the DVS. The lack of overlap between the DVS and SVS allowed comparison of respondents’ scores on these scales.

Higher scale scores on the type of violence scales indicate higher understanding and attitudinal rejection of technology-facilitated abuse (TFAS), and higher attitudinal rejection of domestic violence (DVS), sexual assault (SAS), sexual harassment (SHS) and sexual violence in general (SVS).

Groups of items that do not sit in a scale

In addition to the nine scales, there are three groups of items in the 2021 instrument that are not part of a scale. These items are reported on at the individual item level. These three groups of items are:

• Bystander items – these items were used to examine whether respondents would be bothered by and would intend to intervene if they witnessed disrespect or abuse of women by asking about three different scenarios:
• • a male work friend telling a sexist joke (Scenario B1)
  • a male boss telling a sexist joke (Scenario B2)
  • a male friend verbally abusing his partner (Scenario B3).

The scenarios were deliberately designed to vary in terms of the type of disrespectful behaviour (a sexist joke versus verbal abuse) and the relationship of the perpetrator to the bystander (a male friend versus a male boss).

• The Aboriginal and/or Torres Strait Islander respondents’ module – these items aimed to ask Aboriginal and/or Torres Strait Islander respondents, in a culturally safe and appropriate manner, about their perceptions of service responses to violence against Aboriginal and/or Torres Strait Islander women. These results will be discussed in a forthcoming paper on Aboriginal and/or Torres Strait Islander respondents.
• Additional knowledge items – these items include two items about the law, two items about perception of violence against women as a problem in Australia and locally, an item about knowledge of support services for domestic violence, and an item about sexual assault knowledge. The results of these items are presented in break-out boxes in the relevant chapters. ^[14]

Changes since 2017 to demographic items

New or revised demographic items were included in 2021 on biological sex, gender, sexuality and disability to provide additional and more inclusive demographic information and to capture gender identity, diversity and experience more accurately, in keeping with current standards. These items were drafted in consultation with relevant organisations represented on the NCAS Advisory Group and other stakeholders, including government, peak and advocacy bodies.

The 2017 item on biological sex was altered into several items in 2021 to capture gender identity, diversity and experience more accurately, in keeping with current standards. A new item was also added to capture sexuality. In consultation with relevant stakeholders, the 2021 NCAS incorporated demographic items from the ABS Standard for Sex, Gender, Variations of Sex Characteristics and Sexual Orientation Variables 2020 (hereafter, “ABS Standard”; ABS, 2021h). The 2021 NCAS was the first large-scale data collection with a representative sample of the Australian population to implement the ABS Standard.

Sex

“Sex” refers to the biological sex recorded or presumed for a person at birth. To capture information on sex, respondents were asked, “What sex was recorded on your birth certificate when you were born?” We report on the number of male and female respondents in the 2021 sample (Section 2.4), but do not present results on understanding or attitudes based on biological sex.

Gender identity

“Gender identity” refers to people’s internal sense of their gender and how they describe themselves. To measure gender identity, respondents were asked, “How do you describe your gender?” and, if needed, “Gender refers to your current gender which may be different to the sex recorded at birth or on legal documents”. Throughout this report, gender identity is based on the respondents’ response to this item and is referred to for simplicity as “gender”.

NCAS results are reported for three categories of gender: men, women and non-binary people. In 2021, for the first time, the NCAS reports on the results for non-binary and gender-diverse respondents, where there were sufficient numbers for reliable reporting. The sample included 78 respondents who explicitly identified as “non-binary”. The sample also included another three respondents who identified outside the gender binary but used a term other than “non-binary”. This number of respondents (N = 3) was too small for reliable reporting on them as a separate group. Thus, all 81 respondents who identified outside the gender binary are reported on as a single group. Based on stakeholder advice and for ease of reporting, “non-binary” is used in the present report as an umbrella term to refer to all respondents who reported they were non-binary or another gender identity outside the gender binary.

Gender experience

“Gender experience” refers to how individuals experience gender, and the extent to which their gender identity matches or deviates from the sex recorded or presumed for them at birth. Examples of gender experience include “cisgender”, which refers to people who identify their gender as the same as the sex that was presumed for them at birth; and “transgender”, which is an inclusive umbrella term referring to people whose gender is different from the sex recorded or presumed for them at birth and is not contingent on how they socially, medically or legally affirm their gender (Transhub, 2021).

Following the ABS Standard, the 2021 NCAS used a two-step method to classify cis and trans experiences in the NCAS sample. This two-step method involved cross-classifying responses to demographic items on current gender (i.e. “How do you describe your gender?”) and sex recorded at birth (i.e. “What sex was recorded on your birth certificate when you were born?”). In line with stakeholder advice, we report on the number of trans and cis respondents in the 2021 sample (Section 2.4), but do not present results on the understanding or attitudes held by these groups of respondents. Thus, when reporting the NCAS results, the category of “women” includes cis and trans women, and the category of “men” includes cis and trans men. By grouping together respondents with cis and trans experiences of gender, the analyses cannot tell us whether the understanding and attitudes measured differ by cis or trans experience. However, reporting the results in this way respects respondents’ stated gender identity.

Intersex

For the first time in 2021, respondents were asked, “Were you born with a variation of sex characteristics, sometimes called ‘intersex’ or ‘DSD’?” (disorder/difference of sex development). We report on the number of intersex or DSD respondents in the 2021 sample (Section 2.4), but do not present results on the understanding or attitudes held by intersex respondents.

Sexuality

In 2021, an item on sexuality was included in the NCAS for the first time to provide more inclusive reporting. This item asked, “How would you describe your sexuality?”, with the following response options being read out to respondents:

• heterosexual/straight
• lesbian
• gay
• bisexual or pansexual
• queer
• another term [please specify]
• prefer not to say/unanswered. ^[15]

The response option “another term” allowed respondents (who did not identify with one of the sexualities read out) to specify the term that they prefer to use (e.g. “asexual”, “diverse”). As there were insufficient numbers to report separately on each sexuality identified, the results below (and throughout the report) are provided for the following five sexuality groupings:

• heterosexual
• lesbian
• gay
• bisexual or pansexual
• asexual, queer or diverse sexualities. ^[16]

Disability

The 2017 item on disability was amended to better capture the range of disabilities and long-term health conditions, including stress-related, mental health, intellectual and physical conditions. An additional item was added to capture the impact of disability on core activities. These changes brought the measurement of disability in the NCAS in line with the ABS’s PSS; the ABS’s Disability, Ageing and Carers Survey; and the Census (ABS, 2017, 2018a, 2018c).

Changes since 2017 to measurement of understanding and attitudes

As outlined below, the 2017 NCAS instrument was redeveloped for 2021, retaining many items to facilitate examination of changes in understanding and attitudes over time.

New items were added to better measure understanding and attitudes regarding forms of violence that have emerged recently or have not been a major focus of the NCAS previously.

Violence against women involving intersectional inequalities

The 2021 NCAS includes new items on forms of violence against women that are related to intersectional forms of oppression (see also Section 1.2). Specifically, new items were added to examine understanding of the range of behaviours that constitute domestic violence, including controlling, threatening or neglecting a partner in ways that target an aspect of the partner’s identity or experience, such as their migrant or disability status, gender experience, sexuality or religion (UVAWS). Two attitude items were added to examine trust in women’s reports of violence where the women had mental health issues, or where the women were lesbian or bisexual (AVAWS).

Technology-facilitated abuse, sexual harassment and stalking

Items on technology-facilitated abuse, sexual harassment and stalking were added to the NCAS to allow more detailed reporting on these forms of violence. As already noted, with the addition of new items there were sufficient items on technology-facilitated abuse to develop a psychometrically validated scale on this form of violence. Similarly, with the addition of new items it was possible to develop a Sexual Assault Scale and a Sexual Harassment Scale. ^[17]

Changes to main scales since 2017

Table 2-1 details the changes to the three main scales since 2017, including the number of retained items and the number of new items. The UVAWS retained the same name as in 2017, whereas the AGIS was previously called the Gender Equality Attitudes Scale (GEAS) and the AVAWS was called the Community Attitudes Supportive of Violence against Women Scale (CASVAWS).

Most UVAWS items present statements describing behaviours enacted against women and ask respondents whether they are forms of violence against women. A higher score represents more “yes” responses to the statements, indicating higher understanding of violence against women. The UVAWS was substantially expanded since 2017, when it comprised only six items and no subscales. The 2021 UVAWS comprises three subscales and 19 items. The 2017 UVAWS examined understanding of violence against women and understanding of domestic violence. These items were split into two subscales in 2021 – the Recognise Violence Against Women (VAW) Subscale and the Recognise Domestic Violence (DV) Subscale – and both subscales were expanded to draw on the new content on violence driven by intersecting inequalities and technology-facilitated abuse. In addition, a third subscale – the Understand Gendered Domestic Violence (DV) Subscale – was added, which comprises three (revised) items from the 2017 NCAS that were not included in the 2017 UVAWS.

The AGIS presents statements about gender inequality and asks respondents whether they agree or disagree. A higher score represents higher disagreement with the statements, indicating stronger attitudinal rejection of gender inequality. The 2021 AGIS is identical to the 2017 GEAS, except that one item was removed because of poor statistical fit. ^[18]  The name of the scale was changed to reflect that the items present statements about gender inequality (rather than gender equality). The same five subscales were retained as in 2017, although their names were also changed to better reflect the content of the items they contain. The 2021 AGIS subscales are the Reinforce Gender Roles, Undermine Leadership, Limit Autonomy, Normalise Sexism and Deny Inequality subscales. ^[19]

The AVAWS presents statements about violence against women and asks respondents if they agree or
disagree with these statements. Notably, the scoring of the AVAWS was reversed compared to 2017. Higher scores in 2021 indicate higher disagreement with the statements, indicating stronger attitudinal  rejection of violence against women. In contrast, higher scores in 2017 indicated stronger attitudinal  support  for violence against women. The 2021 AVAWS was expanded from the 2017 CASVAWS and its subscales were revised from four in 2017 to three in 2021. The name of the scale was changed as the scale measures the  level of rejection of violence against women rather than the  level of support of violence against women.

Importantly, the change to the direction of the AVAWS scoring was made so that scores on all scales in 2021 run in the same direction to aid interpretation and comparison of scales. That is, in 2021,  higher scores on all scales and subscales indicate more “positive” understanding or attitudes. ^[20]

New items were developed according to strict social science methods and involved a comprehensive scan
of existing peer-reviewed literature and validated questionnaires for relevant items, as well as cognitive testing, psychometric scale validation and pilot testing. To make room for new items, some items from the 2017 NCAS were removed. Items were removed based on their lack of clarity and precision (according to cognitive or pilot testing), their poor statistical performance or fit, or because new content was deemed to have greater policy or research relevance. ^[21]  For example, the following three constructs were removed because they were less likely to reveal new insights compared to new items on technology-facilitated violence and violence resulting from intersectional inequalities:

• factors that contribute to domestic violence
• prejudice attitudes
• general violence attitudes. ^[22]

Table 2-1: Changes to main scales, 2021

2021 scale	Key differences from 2017	Total items in 2021	Items also in 2017 scale	Items in 2017 NCAS but not in 2017 scales	New items in 2021
UVAWS	Expanded, subscales created	19	6	5	8
AGIS a	Identical to 2017 except 1 item removed, same subscales	17	17	0	0
AVAWS b	Expanded, subscales revised and scoring reversed	43	32	6	5

Note:
a This scale was called the GEAS in 2017.
b This scale was called the CASVAWS in 2017.

2.3 Sampling

The sample consisted of 19,100 Australians aged 16 years or over, who were interviewed via mobile telephone between 23 February and 18 July 2021. The sampling approach largely involved random digit dialling (RDD) of mobile telephones, which was supplemented or “topped up” with listed mobile telephones. Eighty-one per cent of the interviews were achieved via RDD. Random probability sampling, ^[23]  such as RDD mobile sampling, is widely acknowledged as the best approach for achieving a sample that best reflects the demographic profile of the population and allows for accurate reporting about the population.

It was not practicable to use RDD mobile sampling for the entire sample for two reasons. First, to support reliable reporting at the state and territory level, small population states were purposely “oversampled” so that a minimum of 1,000 interviews were achieved in each jurisdiction. ^[24]  As RDD sampling of mobile telephones cannot use location information, it was necessary to use listed mobile telephones to efficiently achieve the additional interviews required in the smaller population states.

Second, an additional 1,600 interviews with Victorians were required (i.e. in addition to the expected number based on random sampling). ^[25] Again, listed mobile telephones were used to conduct these additional interviews in Victoria. ^[26]

A response rate of 11 per cent was achieved. ^[27]  Although low in absolute terms and lower than the 2017 response rate of 17 per cent, this is consistent with the notable decline in survey response rates globally (Pickett et al., 2018). Response rates do not create bias in the sample unless reasons for response (such as incentives) or non-response are related to the outcome of interest (Groves & Peytcheva, 2008; Hendra & Hill, 2019; Pickett et al., 2018). Steps were taken to ensure a random and representative sample, including random sampling and weighting, so the low response rate is unlikely to have affected outcomes.

To maximise the range of topics that could be explored, some survey items were not asked of the full sample but were “split-sampled”. That is, the sample was randomly allocated into four subsets of approximately 4,775 respondents each. A minority of items were asked only of half the sample (two subsets) or one quarter of the sample (one subset). Items that were not asked of the full sample are noted in tables and figures.

In addition to English, interviews were available in the other 10 languages most commonly spoken in Australia, ^[28]  using translated versions of the instrument and bilingual interviewers. Of the 19,100 interviews, 116 were conducted in languages other than English.

Weighting

Random population-level surveys such as the NCAS usually produce samples with similar demographics to the population. However, some sections of the community can be somewhat under-represented in random surveys, for example, because they are less likely to own a telephone or less likely to agree to an interview. Weighting is typically used with population-level surveys to adjust for any such small differences between the sample and the population that may be due to non-coverage or non-response. By aligning the sample to population benchmarks, weighting strengthens confidence that the survey results accurately represent the population.

The following demographic benchmarks were used to align the non-Indigenous respondents in  each state or territory with the demographic profile of the population in that jurisdiction:

• gender
• age by education
• region (i.e. capital city versus rest of state)
• country of birth (i.e. main language is English versus other language).

Use of state and territory rather than national benchmarks facilitates accurate reporting at both the jurisdictional and national level.

With the exception of country of birth, the same demographic benchmarks were used to align the Aboriginal and/or Torres Strait Islander sample with the demographic profile of the  national population of Aboriginal and/or Torres Strait Islander peoples. There were insufficient numbers of Aboriginal and/or Torres Strait Islander respondents to use state/territory benchmarks rather than national benchmarks for these respondents. ^[29]

To allow accurate reporting for Australia as a whole, the weights assigned to Indigenous and non‑Indigenous respondents were combined into a single weighting variable which also adjusted for the oversampling of the smaller population states and Victoria.

Additional weighting variables were also derived based on this weighting approach to facilitate reporting on each state/territory. ^[30]

2.4 Demographics of the final sample

Table 2-2 presents the number of respondents in each demographic group.

Table 2-2: Demographics of the final sample, 2021

Demographic factor – Gender

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Men	8,860	46	9,299	49
Women	10,122	53	9,658	51
Non-binary respondents	81	<1	106	1
Total answered	19,063	100	19,063	100

Demographic factor – Sex

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Male	8,896	47	9,338	49
Female	10,174	53	9,731	51
Another term	3	<1	5	<1
Total	19,073	100	19,074	100

Demographic factor – Intersex/DSD ^a

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Yes	72	<1	99	1
No	18,437	97	18,234	96
Unsure	481	3	638	3
Total answered	18,990	100	18,971	100

Demographic factor – Gender experience ^b

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Cis respondents	18,916	99	18,882	99
Trans (including non-binary) respondents	127	1	162	1
Inadequately described	57	<1	56	0
Total	19,100	100	19,100	100

Demographic factor – Age (in years)

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
16–24 years	1,669	9	2,692	14
25–34 years	2,708	14	3,548	19
35–44 years	3,028	16	3,200	17
45–54 years	3,421	18	3,002	16
55–64 years	3,801	20	2,773	15
65–74 years	3,156	17	2,192	11
75+ years	1,317	7	1,693	9
Total	19,100	100	19,100	100

Demographic factor – Sexuality

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Heterosexual	17,504	94	17,328	93
Lesbian	158	1	126	1
Gay	262	1	251	1
Bisexual or pansexual	630	3	768	4
Asexual, queer or diverse sexualities	151	1	171	1
Total answered	18,705	100	18,643	100

Demographic factor – Disability

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Disability – moderate/profound impact	2,343	12	2,524	13
Disability – no/mild impact	3,141	17	3,072	16
No disability	13,454	71	13,321	70
Total answered	18,938	100	18,917	100

Demographic factor – Aboriginal and/or Torres Strait Islander ^c

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Yes, Aboriginal	405	2	376	2
Yes, Torres Strait Islander	20	<1	13	<1
Yes, both	17	<1	14	<1
No	18,594	98	18,623	98
Unsure	35	<1	45	<1
Total answered	19,071	100	19,070	100

Demographic factor – Country of birth and length of time in Australia ^d

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Born in Australia	13,761	73	12,664	67
MESC: 0–5 years	90	<1	91	<1
MESC: 6–10 years	162	1	172	1
MESC: >10 years	1,754	9	1,536	8
N-MESC: 0–5 years	517	3	829	4
N-MESC: 6–10 years	489	3	708	4
N-MESC: >10 years	2,167	11	2,915	15
Total answered	18,940	100	18,913	100

Demographic factor – English proficiency ^e

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
English at home	15,981	84	15,034	79
LOTE: good/very good English	2,951	15	3,743	20
LOTE: no/poor English	138	1	278	1
Total answered	19,070	100	19,055	21

Demographic factor – Formal education

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
University or higher	8,760	46	5,137	27
Trade/certificate/diploma	5,184	27	7,012	37
Secondary or below	5,040	27	6,848	36
Total answered	18,984	100	18,997	100

Demographic factor – Main labour activity ^f

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Employed	11,563	61	11,032	58
Unemployed	671	4	853	4
Home duties	1,076	6	1,228	6
Student	1,044	5	1,664	9
Retired	3,998	21	3,507	18
Unable to work	586	3	649	3
Volunteering	68	<1	67	<1
Other	41	<1	41	<1
Total answered	19,047	100	19,041	100

Demographic factor – State/Territory

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Australian Capital Territory	1,006	5	319	2
New South Wales	4,330	23	6,083	32
Northern Territory	1,000	5	177	1
Queensland	3,055	16	3,810	20
South Australia	1,110	6	1,335	7
Tasmania	1,000	5	408	2
Victoria	6,143	32	5,010	26
Western Australia	1,456	8	1,960	10
Total	19,100	100	19,100	100

Demographic factor – Socioeconomic status of area ^g

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
1 – Lowest status	2,518	13	2,904	15
2 – Second-lowest status	2,952	16	3,228	17
3 – Middle status	3,612	19	4,039	21
4 – Second-highest status	4,023	21	3,848	20
5 – Highest status	5,750	30	4,767	25
Total with valid area status	18,855	100	18,786	100

Demographic factor – Remoteness

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Major city	12,683	67	13,504	72
Regional	5,770	31	5,009	27
Remote	424	2	303	2
Total with valid remoteness by postcode	18,877	100	18,816	100

Employed men: Gender composition of respondent’s occupation

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Highly men-dominated (≥75% men)	2,540	48	2,910	53
Men-dominated (60–74% men)	843	16	848	15
Gender-balanced (<59% for each gender)	1,094	21	965	18
Women-dominated (60–74% women)	438	8	406	7
Highly women-dominated (≥75% women)	382	7	352	6
Total employed men respondents	5,297	100	5,481	100

Employed women: Gender composition of respondent’s occupation

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Highly men-dominated (≥75% men)	568	11	521	12
Men-dominated (60–74% men)	452	9	403	9
Gender-balanced (≤59% for each gender)	1,125	22	949	21
Women-dominated (60–74% women)	979	19	853	19
Highly women-dominated (≥75% women)	1,904	38	1,700	38
Total employed women respondents	5,028	100	4,426	100

Men: Gender composition of social network~

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Mainly/totally women (women-dominated)	178	8	212	9
Equally men and women (gender-balanced)	1,509	69	1,536	66
Mainly/totally men (men-dominated)	505	23	593	25
Total men respondents	2,192	100	2,341	100

Women: Gender composition of social network~

Demographic group	Unweighted N	Unweighted %	Weighted N	Weighted %
Mainly/totally women (women-dominated)	975	39	886	36
Equally men and women (gender-balanced)	1,449	57	1,430	59
Mainly/totally men (men-dominated)	105	4	122	5
Total women respondents	2,529	100	2,439	100

Note: Totals do not always add to 19,100 due to split-sampling and/or unanswered items by some respondents. Unweighted percentages reflect the proportion of respondents with that demographic characteristic in the sample, whereas weighted percentages reflect each demographic group’s share of the Australian population.

a Respondents were asked, “Were you born with a variation of sex characteristics, sometimes called ‘intersex’ or ‘DSD’?”, and provided with the following clarification if needed: “Intersex people are born with physical sex characteristics that do not fit typical definitions of male or female bodies. For example, this may include characteristics related to sexual anatomy, reproductive organs, hormonal patterns and/or chromosomal patterns. DSD stands for disorder of sex development.” Note that some people with DSD prefer the term “difference of sex development” rather than “disorder of sex development” or “intersex”. “Intersex” implies “between the sexes”, whereas many people with DSD identify themselves as strongly male or female. Information on intersex or DSD is used here to describe the sample but was not used in analysis.

b Gender experience was used to describe the sample but was not included in any analyses (Section 2.2).

c Results for Aboriginal and/or Torres Strait Islander respondents will be reported in a forthcoming paper.

d “MESC” refers to people born in a main English–speaking overseas country (ABS classification) and “N-MESC” refers to people born in a non-main English–speaking country. The number of years refers to the number of years since the respondent moved to Australia.

e “LOTE” refers to language other than English spoken at home.

f “Other” main labour activities included unpaid or overseas work, starting a business, on holiday etc.

g “Socioeconomic status of area” refers to an ABS measure of socioeconomic conditions in geographic areas in terms of people’s access to material and social resources, and their opportunity to participate in society (SEIFA quintiles).

~ Asked of one quarter of the sample.

2.5 Analysis and reporting

As discussed below, the data analysis involved multiple measures and statistical techniques to ensure that conclusions were based on a thorough investigation of the data from different angles. Data analysis was conducted both on individual items and on scale and subscale scores. [31] Further, analysis of scale and subscale scores included examination of both:

• mean scale and subscale scores for all scales
• dichotomous classifications of respondents into “advanced” and “developing” categories on each scale.

More specifically:

• Mean scale and subscale scores for all scales were used to examine respondents’ average level of understanding or average level of rejection of problematic attitudes.
• The proportion of respondents with “advanced” (i.e. strong) rather than “developing” (i.e. more limited) understanding of violence against women or rejection of problematic attitudes according to each scale is also reported to supplement the mean scores.

Mean scores are useful for summarising community understanding and attitudes at a single point in time and for determining whether there have been significant changes over time (see below). However, mean scores do not intrinsically indicate what might be considered a very “high” level of understanding or very “progressive” attitudes. Consequently, the classification of respondents into “advanced” and “developing” categories was used to provide information on how Australia is tracking against the aspiration that everyone in the community has “advanced” understanding and attitudes regarding violence against women. Respondents were classified as either “advanced” or “developing” on each scale based on their answers to the scale’s items. As the classification was meant to provide information against an aspirational goal, a strict criterion was used for each scale. To be classified in the “advanced” category on each scale, respondents had to disagree (strongly or somewhat) with all the items describing problematic attitudes or to recognise that all the problematic behaviours described by the items are (always or usually) forms of violence. Table 2-3 presents further details of the criterion used to classify respondents as “advanced” on each scale. The remaining respondents were classified as “developing” on each scale.

Please note that the 2017 NCAS report used quartiles instead of the above “advanced”/”developing” classification method to report on the proportion of respondents with the highest understanding and the most progressive attitudes. Given that the methods were different, it is not appropriate to compare the 2017 results based on quartiles with the 2021 results based on the advanced/developing classification. As detailed in the Technical report, Chapter T13, the quartile method was replaced because, unlike the 2021 classification method, it cannot be used to make comparisons over time or to assess the  absolute level of understanding or progressive attitudes held by respondents.

Table 2-3: Criterion defining “advanced” category for each scale, 2021

Scale	Description of scale items	Criterion for the “advanced” category for this scale a	Respondents in the “advanced” category have strong … Respondents in the “developing” category have more limited …
UVAWS a	Items ask if behaviours are a form of violence	Answered “yes, always” the behaviour is violence for at least 75% of items and “yes, usually” to the remaining items (or the equivalent)	understanding of violence against women
AGIS, AVAWS, DVS, SVS, SAS, SHS	Items describe problematic attitudes	“Strongly disagreed” with problematic attitudes for at least 75% of items and “somewhat disagreed” with the remaining items (or the equivalent)	rejection of gender inequality (AGIS), violence against women (AVAWS), domestic violence (DVS), sexual violence (SVS), sexual assault (SAS) and sexual harassment (SHS)
TFAS	Items either ask if behaviours are a form of violence or describe problematic attitudes	Answered “yes, always” the behaviour is violence or “strongly disagreed” with problematic attitudes for at least 75% of items, and answered the remaining items “yes, usually” or “somewhat disagree” (or the equivalent)	understanding and rejection of technology-facilitated abuse

aDue to split-sampling, there were some scales where no respondent received all the scale’s items. Thus, for each scale, the first step was to determine the  lowest scale score among the respondents who had received the  most scale items and  met the criterion. Second, this “cut-off” scale score was used to categorise all respondents as either “advanced” or “developing” based on whether their scale score was higher or lower than the cut-off. For the UVAWS, none of the three items comprising one subscale (the Understand Gendered DV Subscale) were included in the determination of the cut-off score because they were asked of only one quarter of respondents.

Univariate analyses

Univariate, bivariate and multivariate data analyses were conducted as summarised below. Univariate analysis involves one variable only. Univariate analyses were used to report on:

• the sample’s responses to each understanding, attitude and bystander item (e.g. percentage of respondents who disagreed/agreed with an item)
• the percentage of the sample categorised as having “advanced” understanding or attitudes according to each scale.

Bivariate analyses

Bivariate analysis examines the direct or straightforward relationship between two variables, such as an outcome of interest (e.g. attitudes towards violence against women) and one other variable (e.g. a demographic factor such as age). Thus, these analyses do not consider the effect of any other variables that may be related to the two variables being examined. Bivariate analyses provide a starting point for examining which variables may be associated with understanding or attitudes.

The bivariate analyses examined:

• comparisons over time, comparing the 2021 results to previous NCAS waves, for:
• • each understanding, attitude and bystander item
  • each scale and subscale (based on mean scores) [32]
• comparisons between different scales and subscales in 2021 (based on mean scores) ^[33]
• comparisons between different demographic groups in terms of “advanced” understanding or attitudes.

Statistical significance for bivariate analyses

Tests of statistical significance were used to determine whether a difference observed in the sample (e.g. over time or between demographic groups) is likely to represent a true and meaningful difference in the population. Throughout the report, statistically significant results are noted in tables and figures and are referred to as “significant” in the text. Bivariate results are reported as statistically “significant” if:

• the difference was significant at the 95 per cent confidence level ( p ≤ 0.05), after adjusting for multiple comparisons via the Benjamini and Yekutieli (2001) method, and
• the difference was of non-negligible effect size according to Cohen’s  d 0.2. ^[34]

Multivariate analyses

An outcome variable of interest (e.g. understanding of violence) can be related to multiple factors (e.g. multiple demographic factors) and these factors can also be related to one another. As bivariate analyses examine the relationship of the outcome variable to only one factor at a time, they cannot provide information on which factors are most strongly related to the outcome. For example, the demographic factors of age and education level are related such that younger people tend to have a lower level of education. If education level is associated with an outcome of interest, bivariate analyses cannot determine whether this association is due to education or age or both.

Thus, multivariate analyses were used to determine which factors were most predictive of understanding and attitudes. Multivariate analysis examines the relationship of an outcome variable of interest (e.g. understanding of violence) to  multiple factors considered together (e.g. multiple demographic factors). Two types of multivariate analyses were conducted: multiple linear regression analysis and multiple logistic regression analysis. Unlike bivariate analysis, these types of regression analyses can determine which of multiple factors:

• are  independently related to or “predict” the outcome variable, after adjusting for any relationships between the factors
• are  most important in predicting the outcome variable, after adjusting for any relationships between factors.

Multiple linear regression analyses are appropriate when the outcome variable is a continuous variable, such as a scale score that can range from 0 to 100. Multiple linear regression analyses were used to determine which input variables best predict the following scale scores as outcome variables:

• UVAWS scores – understanding of violence against women
• AGIS scores – rejection of gender inequality
• AVAWS scores – rejection of violence against women.

Multiple logistic regression analyses are appropriate when the outcome variable is a dichotomous variable, such as engaging versus not engaging in a prosocial behaviour. Multiple logistic regression analyses were used to determine the predictors of bystander responses when witnessing disrespect or abuse. These models examined the input variables that predict likelihood of:

• being bothered by a male friend telling a sexist joke (Friend sexist joke – bothered)
• being bothered by a male boss telling a sexist joke (Boss sexist joke – bothered)
• intervening if a male friend told a sexist joke (Friend sexist joke – intervene)
• intervening if a male boss told a sexist joke (Boss sexist joke – intervene)
• intervening if a male friend verbally abused his partner (Friend verbal abuse – intervene).

The regression analyses examined if each outcome variable could be predicted by both demographic factors and relevant aspects of understanding and attitudes as measured by the main scales. As summarised in Table 2-4, three regression models were generally conducted for each outcome variable to examine the predictive ability of:

1. the demographic factors only (Model 1)
2. the relevant scales only (Model 2)
3. both the demographic factors and the relevant scales together (Model 3).

Table 2-4 details the scales used as predictors in each model.

The model on how well each outcome of interest can be predicted by respondents’ demographic characteristics  alone (Model 1) was conducted to identify any key differences between demographic groups to assist policymakers and practitioners to target education and prevention initiatives more effectively to the specific needs of different demographic groups. ^[35]

The model on how well each outcome of interest can be predicted by respondents’ understanding and attitudes  alone (Model 2) was conducted to identify the key aspects of understanding and attitudes most related to the outcome of interest. As attitudes towards violence against women (AVAWS) are the key focus of the NCAS, the UVAWS and AGIS were examined as potential predictors of the AVAWS, but not vice versa. The UVAWS was also examined as a potential predictor of the AGIS.

Additional models were conducted to examine which of the UVAWS and AGIS subscales were most responsible for the relationships involving the UVAWS and AGIS as input variables (Table 2-4). The demographic factors examined as potential predictors in the models are stated in the note to Table 2-4.

The model with both demographics and scales (Model 3) was used to provide an estimate of how much of the variation in the outcome of interest can be explained by a person’s demographics and their understanding and attitudes, and how much of the variation is left unexplained by these factors.

Table 2-4: Multiple regression models, 2021

Outcome of variable interest – UVAWS

Model number	Input variables
UVAWS Model 1	Demographics

Outcome of variable interest – AGIS

Model number	Input variables
AGIS Model 1	Demographics
AGIS Model 2	UVAWS
AGIS Model 3	Demographics, UVAWS
AGIS Model 4	UVAWS subscales

Outcome of variable interest – AVAWS

Model number	Input variables
AVAWS Model 1	Demographics
AVAWS Model 2	UVAWS, AGIS
AVAWS Model 3	Demographics, UVAWS, AGIS
AVAWS Model 4	UVAWS subscales, AGIS subscales

Outcome of variable interest – Friend sexist joke – Bothered (B1 – Bothered)

Model number	Input variables
B1 – Bothered Model 1	Demographics
B1 – Bothered Model 2	UVAWS, AGIS, AVAWS, V1
B1 – Bothered Model 3	Demographics, UVAWS, AGIS, AVAWS, V1

Outcome of variable interest – Boss sexist joke – Bothered (B2 – Bothered)

Model number	Input variables
B2 – Bothered Model 1	Demographics
B2 – Bothered Model 2	UVAWS, AGIS, AVAWS, V1
B1 – Bothered Model 3	Demographics, UVAWS, AGIS, AVAWS, V1

Outcome of variable interest – Friend sexist joke – Intervene (B1 – Intervene)

Model number	Input variables
B1 – Intervene Model 1	Demographics
B1 – Intervene Model 2	UVAWS, AGIS, AVAWS, V1
B1 – Intervene Model 3	Demographics, UVAWS, AGIS, AVAWS, V1

Outcome of variable interest – Boss sexist joke – Intervene (B2 – Intervene)

Model number	Input variables
B2 – Intervene Model 1	Demographics
B2 – Intervene Model 2	UVAWS, AGIS, AVAWS, V1
B2 – Intervene Model 3	Demographics, UVAWS, AGIS, AVAWS, V1

Outcome of variable interest – Friend verbal abuse – Intervene (B3 – Intervene)

Model number	Input variables
B3 – Intervene Model 1	Demographics
B3 – Intervene Model 2	UVAWS, AGIS, AVAWS
B3 – Intervene Model 3	Demographics, UVAWS, AGIS, AVAWS

Note: The demographic factors included as input variables in the models were generally age, gender, sexuality, disability, country of birth and length of time in Australia, English proficiency, formal education, main labour activity, socioeconomic status of area and remoteness of area. Due to insufficient numbers in some sexuality groups, sexuality was not included as a demographic input variable for the bystander models. Unlike the B3 (verbal abuse) model, the B1 and B2 (sexist joke) bystander models also included gender composition of social network as a demographic input variable and included item V1 (“Do you agree or disagree that violence against women is a problem in Australia?”) together with the scale input variables. These variables could not be included in the B3 model because they were not asked of the quarter sample who were asked about the B3 scenario. The demographic groups compared for each demographic factor are shown in the tables presenting the regression results in the relevant chapters (Tables 4-4, 5-6, 6-4, 8-2 and 9-1).

Model fit and statistical significance for multivariate analyses

Model fit

Each model initially included the input variables detailed in Table 2-4. Input variables were removed from the final version of a model if their inclusion did not improve the goodness of fit of the model according to Akaike’s Information Criterion. ^[36]  The percentage of the variance explained by each model is reported. This percentage indicates how well the outcome variable can be predicted by the variables in the model – for example, how much of the difference in respondents’ understanding of violence (outcome variable) can be explained by the demographic factors in the model (input variables).

Significant predictors

The statistical significance of each input variable retained in a final model was then determined by conducting comparisons between categories or groups for that variable. Specifically, for each retained variable (e.g. gender), one chosen or “reference” group (e.g. men) was compared to each other group (e.g. women and non-binary respondents). ^[37]  Input variables retained in a final model are reported as “significant predictors” if they involved at least one “significant” comparison, where the difference was:

• significant at the 95 per cent confidence level ( p ≤ 0.05), and
• of non-negligible effect size, according to a standardised regression coefficient 0.2 for the multiple linear regressions or an odds ratio 1.44 for the multiple logistic regressions. ^[38]

Note that although some input variables were retained in some final models because they improved model fit, they are not reported as “significant predictors” because they did not involve a “significant difference” between the groups that were compared. ^[39]

The absolute importance of each significant predictor is also reported according to its “unique contribution” to the outcome variable – that is, the proportion of variance in the outcome variable that was uniquely explained by that predictor.

Weighting of analyses

All analyses (including univariate, bivariate and multivariate) were conducted on weighted data to strengthen confidence that the survey results accurately represent the population. The total number of respondents for each analysis (unweighted) is provided in the note to the table or figure presenting the findings of the analysis. Numbers lower than the total sample size of 19,100 reflect split-sampling of some items, data on a specific demographic group, missing data on some variables, or a combination of these. ^[40]

2.6 Strengths and limitations

Understanding the strengths and limitations of research is important for accurate interpretation of the results. Table 2-5 presents the strengths and limitations of the 2021 NCAS.

Table 2-5: Factors to consider when interpreting the 2021 NCAS results

Factor	Strength	Limitation
Representative of the Australian population, as well as demographic groups of interest	Results were representative of the Australian population as far as practically possible, and were strengthened by: a large, majority random sample weighting to population demographics where (small) deviations occurred use of statistical analyses to assess whether sample results accurately represent the population steps taken to minimise self-selection bias, including random sampling, call procedures to facilitate participation across the population (e.g. multiple calls, calls in and out of business hours, voicemails), interviews in 10 languages other than English, a 1800 number to answer queries and receive feedback, and careful consideration of introductory scripts and item wording measurement (for the first time in 2021) of non-binary gender, sexuality and disability severity	Minor deviations from the demographic profile of the Australian population may have occurred: if the sample differed from the population in ways not adjusted for by the weighting approach used if people’s decisions about participating in the survey were systematically influenced by another factor (e.g. whether they were interested in women’s safety, whether they answered calls from unknown numbers; see steps taken to minimise self-selection bias) if there were insufficient numbers for reliable reporting on some demographic groups (e.g. some sexuality groups) due to their small population proportions
Measurement of understanding and attitudes	Understanding and attitudes were measured robustly by: using multiple items (i.e. scales and subscales) to measure understanding of violence and attitudes towards violence and gender inequality using items drawn from existing measures improving items through cognitive and pilot testing psychometric validation of the scales and subscales expansion of scales to address previous gaps and emerging issues	Minor limitations were: scales and subscales with fewer items are less precise than those with more items some items still contained binary gender or heteronormative framing to cover a greater number of topics, some items were asked of only a half or quarter sample, reducing the power for some statistical analysis limitations associated with surveys in general, such as social desirability bias (Knoll, 2013; Larson, 2019; McMahon & Farmer, 2011)
Ability to benchmark change over time	The NCAS measures change in understanding and attitudes over time. It does this well by: maintaining a core set of items that are asked each NCAS wave using large, representative samples of the population in each wave applying revised scale calculation approaches retrospectively to previous years as needed adapting to the changing interview landscape by using emerging and innovative methodologies (e.g. piloting a method in 2021 for achieving a representative sample with online interviewing)	Results can only be used to assess associations, not causations, because the NCAS is cross-sectional and does not follow up the same respondents over time. Retrospective adjustments were made to mean scale scores from previous NCAS waves so that they could be compared to 2021. Thus, the mean scale scores for previous waves presented in this report may not match those published previously. The scores in the present report should be used for comparing 2021 with previous waves
Social desirability bias	Social desirability bias is where respondents give what they think are socially acceptable answers rather than their actual opinions (Brenner, 2017; Näher & Krumpal, 2012; Tourangeau & Yan, 2007). Social desirability effects were minimised by: maintaining anonymity and confidentiality allowing respondents to skip items they were uncomfortable answering assuring respondents that “we’re just interested in your opinion. There are no right or wrong answers” increasing respondent comfort by matching the gender of the interviewer to that of the respondent (or providing a choice of interviewer gender) including items that measure more covert forms of attitudinal support for violence against women and gender inequality (e.g. nuanced questions and the use of scenarios)	Despite efforts to minimise social desirability effects, it is possible that social desirability bias was not fully eliminated, especially as interviews over the telephone can feel less anonymous than online surveys. Thus, it is likely that the findings under-represent the extent of negative attitudes
Comparisons between groups of people	The measurement of a wide range of demographic factors and large number of respondents allowed examination of understanding and attitudes in different community groups	All groupings of people necessarily encompass diversity in identity, experience, understanding, attitudes and responses. To include sufficient numbers in each group without excluding individuals from analyses, some diverse groups were combined
Cultural and language differences	We tried to minimise differences in interpretation of the items due to cultural or language factors by: offering interviews in 10 languages other than English conducting cognitive and pilot testing of survey items with a broad range of people providing standard definitions and explanations of concepts as required	Despite efforts taken, it is still possible that some observed differences in results for different cultural and linguistical groups may partly reflect differences in interpretation of items rather than purely differences between groups in the constructs being measured

3 Findings: Benchmarking understanding and attitudes

3.1 Benchmarking broad understanding and attitudes

While respondents had high awareness that violence against women is a national problem, their awareness that violence against women transcends all communities, including their own local area, was much lower (Figure 3-1 and Box 3-1). This finding suggests a misconception that violence tends to occur generally outside one’s own networks, rather than everywhere, which may impede recognition that violence is a community-wide problem requiring action at all levels of society.

Benchmarking broad understanding and attitudes over time

The GVIS is a “megascale” that consists of all knowledge and attitude items included in the other eight NCAS scales. The GVIS provides an  overall indicator of the Australian community’s progress towards stronger understanding and attitudinal rejection of gendered violence and gendered inequality. The GVIS was also constructed to serve as a statistical “anchor” for the other NCAS scales to allow valid comparison between scales.

As Figure 3-2 shows, average scores on the GVIS were significantly higher in 2021 compared to each of the three previous waves of the NCAS, indicating a significant improvement over time in the Australian population’s overall understanding and rejection of gendered violence and inequality.

To examine which aspects of understanding and attitudes contributed to the improvement in GVIS scores over time, we also examined changes over time for each of the three main scales that make up the GVIS, namely the UVAWS, AGIS and AVAWS. As Figure 3-3 shows, according to mean UVAWS scores, the community’s broad understanding of violence against women was significantly higher in 2021 compared to 2009, 2013 and 2017. Similarly, mean AGIS scores indicated an improvement in attitudes rejecting gender inequality in 2021 compared to each previous wave. However, according to mean AVAWS scores, there was no significant improvement in attitudinal rejection of violence against women between 2017 and 2021, despite a significant improvement compared to 2009 and 2013. These findings suggest that attitudes rejecting violence have improved more slowly than understanding of violence and attitudes rejecting gender inequality.

Figure 3-2: Understanding and rejection of gendered violence and inequality (GVIS scores) over time, 2009 to 2021

Note:  Ns in 2009, 2013, 2017 and 2021 were: 10,102; 17,508; 17,540; 19,099.
* Statistically significant difference on this scale between the year indicated and 2021.

Figure 3-3: Understanding (UVAWS) and attitudes (AGIS, AVAWS) over time, 2009 to 2021

Note:  Ns in 2009, 2013, 2017 and 2021 were:

UVAWS – 10,033; 17,402; 8,606; 19,096

AGIS – 8,909; 15,178; 17,528; 19,040

AVAWS – 3,743; 5,478; 17,538; 19,097.

* Statistically significant difference on this scale between the year indicated and 2021.

Benchmarking broad understanding and attitudes in 2021

To benchmark overall levels of understanding and attitudes regarding violence in 2021, we compared mean scores on the UVAWS, AGIS and AVAWS in 2021 to one another (Figure 3-3). There were no significant differences in mean scores between the three scales, suggesting that the population’s understanding of violence and attitudes towards both gender inequality and violence against women were at a comparable level in 2021. [44]

While mean scale scores provide a sensitive measure of even small changes over time, they are not easy to interpret in an absolute sense. Thus, we also defined what “advanced” understanding of violence against women (UVAWS) and “advanced” rejection of problematic attitudes (AGIS, AVAWS) would look like. Figure 3-4 presents the percentages of respondents in the “advanced” category for each main scale in 2021. More than two fifths (44%) of respondents demonstrated “advanced” understanding of violence against women. More than one quarter of respondents (28%) demonstrated “advanced” rejection of gender inequality (AGIS) and about one third (34%) demonstrated “advanced” rejection of violence against women (AVAWS). These findings suggest that there is still substantial work to be done on improving community understanding and attitudes regarding violence against women and gender inequality in Australia.

Figure 3-4: “Advanced” understanding of violence against women (UVAWS) and “advanced” attitudinal rejection of gender inequality (AGIS) and violence against women (AVAWS), 2021

Note:  N = 19,100. “Advanced” understanding refers to answering “yes, always” the behaviour is violence to at least 75% of items and “yes, usually” to the remaining items (UVAWS). “Advanced” attitudes refer to answering “strongly disagree” to at least 75% of the items in the scale and “somewhat disagree” to the remaining items in the scale, which condoned gender inequality (AGIS) or condoned violence (AVAWS). See Section 2.5 for further details.

3.2 Types of violence in focus: Benchmarking understanding and attitudes

Types of violence in focus: Benchmarking understanding and attitudes over time

Figure 3-5 shows mean scores over time for the type of violence scales: the DVS, SVS and TFAS. For the TFAS, a mean score could only be provided for 2021 because there was insufficient data in previous years for reliable reporting. ^[45] These results indicate that the Australian population’s attitudinal rejection of sexual violence, according to mean

SVS scores, was significantly higher in 2021 compared to the three previous survey waves. However, although attitudinal rejection of domestic violence, based on mean DVS scores, was significantly higher in 2021 compared to 2009 and 2013, there was no significant improvement between 2017 and 2021.

It is noteworthy that the AVAWS is comprised almost exclusively of the items in the DVS and SVS. ^[46]  Thus, the findings suggest that the plateau between 2017 and 2021 in attitudinal rejection of violence according to the AVAWS (Figure 3-3) was largely due to a plateau in attitudinal rejection of domestic violence rather than sexual violence (Figure 3-5).

Figure 3-5: Understanding (TFAS) and attitudes (DVS, SVS, TFAS) regarding types of violence over time, 2009
to 2021

Note: “na” below means reliable data was not available. Ns in 2009, 2013, 2017 and 2021 were:

DVS – 4,970; 6,850; 17,537; 19,088

SVS – na; na; 17,419; 19,031

TFAS – na; na; na; 19,067.

There were no significant differences between scales in 2021.

* Statistically significant difference on this scale between the year indicated and 2021.

Types of violence in focus: Benchmarking understanding and attitudes in 2021

As Figure 3-5 shows, there was no significant difference in 2021 between the mean scores on the DVS, SVS and TFAS, suggesting that rejection of domestic violence, rejection of sexual violence and understanding and rejection of technology-facilitated abuse were at similar levels.

Figure 3-6 shows the percentage of respondents in the “advanced” category for the DVS, SVS and TFAS in 2021. Forty per cent of respondents demonstrated “advanced” rejection of domestic violence and, similarly, 40 per cent demonstrated “advanced” rejection of sexual violence. One third of respondents (33%) demonstrated “advanced” understanding and rejection of technology-facilitated abuse. These results suggest more effort is needed to improve community attitudes towards these types of violence.

Figure 3-6: “Advanced” attitudinal rejection of domestic violence (DVS) and sexual violence (SVS), and “advanced” understanding and rejection of technology-facilitated abuse (TFAS), 2021

Note:  N = 19,100. “Advanced” rejection of problematic attitudes towards domestic or sexual violence refers to answering “strongly disagree” to at least 75% of the items in the scale and “somewhat disagree” to the remaining items in the scale, which condoned this type of violence (SVS and DVS). The “advanced” TFAS category means that the respondent answered “yes, always” the behaviour is violence or “strongly disagreed” with problematic attitudes for at least 75% of items, and answered the remaining items “yes, usually” or “somewhat disagree”. See Section 2.5 for further details.

3.3 Conclusion: Benchmarking understanding and attitudes

Overall, understanding and attitudes have been improving slowly and significantly over time, as indicated by the GVIS “megascale”. For all NCAS scales, 2013 marked a turning point for understanding and rejection of violence against women and gender inequality. There was minimal change between 2009 and 2013, but significant changes between 2013 and 2021 on all NCAS scales. In addition, there were significant improvements since 2017 in understanding of violence against women (UVAWS), rejection of gender inequality (AGIS) and rejection of sexual violence (SVS).

However, between 2017 and 2021, there was no significant improvement in overall rejection of violence against women (AVAWS), largely reflecting a plateau in the rejection of domestic violence (DVS). ^[47]  Nonetheless, overall rejection of violence against women and rejection of domestic violence had improved over the longer term (since 2013).

While causation cannot be inferred from the improvements over time since 2013 in understanding and attitudes as measured by the NCAS scales, it is notable that these shifts occurred after the first National Plan 2010–2022 was released in 2010 and the first woman prime minister in Australia held office between 2010 and 2013 (COAG, 2010a, 2010b; National Archives of Australia, 2022).

Although there was no significant improvement between 2017 and 2021 in attitudes towards violence against women overall (AVAWS), and attitudes towards domestic violence in particular (DVS), understanding of violence against women (UVAWS), rejection of gender inequality (AGIS) and rejection of violence against women (AVAWS) were at similar levels in 2021. Despite the significant improvements in understanding and attitudes over the longer term, the results demonstrate that increasing community understanding of the nature of violence against women and shifting problematic attitudes regarding gendered violence and inequality is a slow and stubborn process. Fewer than half of all respondents demonstrated “advanced” understanding of violence against women (43%), “advanced” rejection of gender inequality (28%) and “advanced” rejection of violence against women (34%). Thus, there is considerable room to further enhance “advanced” understanding and attitudes across the Australian population.

The results suggest that continued, cohesive effort nationally is required at all levels of the social ecology to disrupt misconceptions and problematic attitudes that reflect broader norms, practices, systems and structures that are embedded throughout our society and facilitate and maintain violence against women (COAG, 2010b, 2022). Efforts need to include primary prevention and early intervention strategies because problematic attitudes are slow and difficult to shift. Violence against women needs to be recognised as a community-wide social problem that requires community-wide responsibility (see Chapter 10 for more details).

The following chapters detail the areas where Australians have good understanding and strong rejection of violence against women and gender equality, and identify the gaps in this understanding and the specific problematic attitudes that remain to be addressed.

4 Findings: Understanding of Violence against Women Scale (UVAWS)

4.1 Understanding of violence against women over time by gender

Figure 4-1 presents the change in understanding of violence against women over time by gender, according to mean UVAWS scores. We could not examine change over time in understanding for non-binary respondents as non-binary genders were not reliably captured in previous waves of the NCAS. However, we updated the gender item in 2021 to capture non-binary genders more accurately and are able to provide the mean UVAWS score for non-binary respondents in 2021. ^[48]

For all respondents, and for men and women separately, the mean UVAWS score was significantly higher in 2021 compared with 2009, 2013 and 2017. These findings indicate a significant increase since the three previous

NCAS waves in the understanding of violence against women for the Australian population overall and for both Australian men and Australian women separately.

Examining only 2021 UVAWS scores, women (70) had a significantly higher mean than men (67; Figure 4-1). Thus, women continue to have significantly higher understanding of violence against women than men, as they did in the three previous NCAS waves. Non-binary respondents had similar levels of understanding to women in 2021 according to UVAWS scores, but there was no significant difference between non-binary respondents and men. ^[49]

Figure 4-1: Understanding of violence against women (UVAWS) over time by gender, 2009 to 2021

Note: “na” below means reliable data was not available. Ns in 2009, 2013, 2017 and 2021 were:

women – 6,033; 9,631; 4,571; 10,119

men – 4,000; 7,771; 4,019; 8,859

non-binary respondents – na; na; na; 81

all – 10,033; 17,402; 8,606; 19,096.

Demographic items for gender were updated for the 2021 NCAS, in accordance with the ABS Standard (ABS, 2021h). As the gender item in previous survey waves did not include the same response options for non-binary respondents, only results for men and women can be compared over time.

* Statistically significant difference on this scale between the year indicated and 2021.

*1 Statistically significant difference compared to men in 2021.

4.2 Understanding of violence against women: UVAWS subscales

Figure 4-2 shows change over time for two of the three UVAWS subscales. The mean score for both the Recognise VAW Subscale and Recognise DV Subscale was significantly higher in 2021 compared to all previous waves of the survey. These results indicate an improvement over time, including an improvement since 2017, in the Australian population’s understanding of the different behaviours that constitute domestic violence and violence against women more broadly. Change over time for the remaining UVAWS subscale, which measures the understanding of the gendered nature of domestic violence, is not reported because one of the three items in this subscale was substantially revised in 2021. ^[50]

The mean scores on the UVAWS subscales in 2021 were also compared to one another to examine whether some aspects of understanding of violence against women are higher than others (Figure 4-2). Based on all respondents in 2021, mean scores on the Recognise DV Subscale were significantly higher than on the Understand Gendered DV Subscale, suggesting that Australians are generally better at recognising behaviours that constitute domestic violence than they are at understanding that domestic violence is disproportionately perpetrated by men against women. ^[51]

Figure 4-2: Understanding of different aspects of violence against women (UVAWS subscales) over time, 2009 to 2021

Note: “na” below means reliable data was not available.  Ns in 2009, 2013, 2017 and 2021 were:

Recognise VAW Subscale – 9,738; 16,927; 8,500; 19,055

Recognise DV Subscale – 10,068; 17,461; 17,146; 19,093

Understand Gendered DV Subscale – na; na; na; 4,758.

Items in the Understand Gendered DV Subscale were modified in 2021.

* Statistically significant difference on this subscale between the year indicated and 2021.

*1 The Recognise DV Subscale mean score was significantly higher than the Understand Gendered DV Subscale mean score in 2021.

~ Items revised and asked of one quarter of the sample in 2021.

Figure 4-3 compares the mean scores on each UVAWS subscale by gender in 2021. Compared to men, women had significantly stronger recognition of both violence against women (Recognise VAW Subscale) and domestic violence (Recognise DV Subscale). Although non-binary respondents had similar mean scores to women, there were no significant differences between non-binary respondents and men on the UVAWS subscales. ^[52]

Figure 4-3: Understanding of different aspects of violence against women (UVAWS subscales) by gender, 2021

Note:  N = 19,100 unless otherwise noted.

* Statistically significant difference compared to men on this subscale.

~ Asked of one quarter of the sample. Results for non-binary respondents are not reported for this subscale due to insufficient numbers.

The three “UVAWS in focus” sections below present the item-level results for each UVAWS subscale. These sections discuss the item-level results in the context of the theoretical concepts underlying each of the subscales. These latent constructs, namely recognition of violence against women, recognition of domestic violence and understanding the gendered nature of domestic violence, were identified based on factor analysis.

UVAWS in focus: Recognise VAW Subscale

The Recognise VAW Subscale of the UVAWS comprises four items that examine respondents’ understanding that certain behaviours are forms of violence against women. One item is about in-person stalking and three items are about technology-facilitated abuse. ^[53]

Violence against women has been defined as:

any act of gender-based violence that results in, or is likely to result in, physical, sexual or psychological harm or suffering to women, including threats of such acts, coercion or arbitrary deprivation of liberty, whether occurring in public or in private life. (WHO, 2019, p. 2)

This gender-based violence is specifically “directed against a woman because she is a woman” or it is “violence that affects women disproportionately” (Our Watch, 2021a, p. 20). Violence and abuse can manifest in many ways and can also occur in spaces that may merge elements of both public and private spaces, such as online spaces. The mode of violence perpetration can also evolve with social shifts and technological advances. A contemporary mode of violence perpetration against women is via digital technologies, including harmful, sometimes sexually based, aggressive and harassing behaviours used to control or instil fear in targets. Studies suggest that technology-facilitated abuse has become a key part of intimate partner and family violence, and violence against women more generally (C. J. Adams, 1996; Afrouz, 2021; C. Brown et al., 2021; eSafety, 2017, 2019a; Harris & Woodlock, 2021; Powell et al., 2022; Vera-Gray, 2017; Woodlock, McKenzie et al., 2020).

Each of the four items in the Recognise VAW Subscale required respondents to consider if a specific behaviour is a form of violence against women. Respondents who answered “yes” were then asked to qualify whether the behaviour is “always”, “usually” or “sometimes” violence against women. The behaviours were deliberatively framed to capture comprehension of the repeated or abusive intent of the behaviour.

As Figure 4-4 shows, most respondents recognised the four behaviours as always or usually forms of violence against women (80–89%). However, in-person stalking (V4) was more often recognised as always a form of violence against women (78%) than the three forms of technology-facilitated abuse involving image-based abuse (68%; V7) and text-based abuse (68%; V5, V6). For example, a sizeable proportion of respondents (18%) thought that a man sending an  unwanted  picture of his genitals (V7) to a woman is not, or is only sometimes, a form of violence against women.

Figure 4-4: Recognising violence against women (UVAWS subscale items), 2021

Note:  N = 19,100. Percentages in the figure do not always add to 100 or exactly correspond to percentages in the text due to rounding. Significant differences over time are based on the percentage of respondents who answered “yes” the behaviour is violence against women either “always” or “usually”.

a New item in 2021. Thus, change over time could not be examined.

* Significantly higher understanding in 2021 than 2017.

Table 4-1 shows the level of agreement with the Recognise VAW Subscale items over time. Consistent with the significant improvement over time in the subscale overall (Figure 4-2), the two Recognise VAW Subscale items with sufficient data in previous surveys waves also showed significant improvements over time. Specifically, compared to the three previous NCAS waves, there was increased recognition in 2021 that electronic harassment (V5) and in-person stalking (V4) are forms of violence against women.

Table 4-1: Recognising violence against women (UVAWS subscale items) over time, 2009 to 2021

…is this a form of violence against women? Percentage of respondents answering strong yes.

Item	Code	2009	2013	2017	2021
Stalking by repeatedly following/watching at home/work	V4	81*	78*	82*	89
A man sends an unwanted picture of his genitals to a woman a	V7	–	–	–	80
Harassment via repeated emails, text messages etc.	V5	73*	71*	76*^	84
Abusive messages or comments targeted at women on social media a	V6	–	–	–	83

Note:  Ns in 2009, 2013, 2017 and 2021 were 10,105; 17,517; 17,542; 19,100.

“Strong yes” refers to answering “yes” the behaviour is violence against women either “always” or “usually”.

a New item in 2021.

* Statistically significant difference between the year indicated and 2021.

^ Asked of half the sample in this year.

UVAWS in focus: Recognise DV Subscale

The Recognise DV Subscale of the UVAWS consists of 12 items, all of which examine the recognition of domestic violence between intimate partners, including one item about technology-facilitated abuse by a partner. ^[54]  In addition to the recognition of physical violence within intimate relationships as a form of domestic violence, this subscale investigates accurate recognition of coercive control as a form of domestic violence.

Coercive control is a pattern of behaviours used to manipulate, intimidate, isolate and control a partner and create an uneven power dynamic in the relationship (ANROWS, 2021; Boxall & Morgan, 2021a; House of Representatives Standing Committee on Social Policy and Legal Affairs, 2021; Victorian Aboriginal Legal Service, 2022; N. Ward, 2021). Coercive controlling behaviours can occur in person and via technology. Therefore, there is overlap between coercive control and technology-facilitated abuse (Dragiewicz et al., 2018; Dragiewicz et al., 2022; Meeting of Attorneys-General, 2022; Woodlock, McKenzie et al., 2022). These behaviours are intended to make a victim and survivor feel scared, isolated and dependent on the abuser. Common ways to enact coercive control include:

• isolating the victim from friends and family
• monitoring the victim’s activities
• restricting the victim’s autonomy
• controlling the victim’s choices about their body, such as about their appearance, food and medical decisions
• degrading the victim to undermine their self-worth
• financial abuse by restricting the victim’s access to money
• using intimidation and threats against the victim or others close to them
• gaslighting and other forms of psychological manipulation.

The intersection of different structural inequalities, such as sexism, ableism, racism, classism, queerphobia, transphobia and ageism, can also produce unique forms of domestic violence and abuse for specific groups of women. For example, abusers can employ controlling tactics by exploiting or targeting aspects of their partner’s identity or experience, such as chronic health conditions or disabilities, gender and sexuality, religion and migrant status (Gray et al., 2020; Kulwicki et al., 2010; Maher & Segrave, 2018; Peitzmeier et al., 2021; Sasseville et al., 2022). Recognition of these forms of abuse through an intersectional lens is important to ensure that every woman’s rights and safety are recognised and defended.

Coercive control is commonly described by victims and survivors as the most damaging form of abuse they experience, often generating more immediate and ongoing impact than physical forms of violence. Coercive control is also a predictor of severe physical violence and homicide (Boxall & Morgan, 2021a; J. Hill, 2019; House of Representatives Standing Committee on Social Policy and Legal Affairs, 2021; Meeting of Attorneys-General, 2022; Monckton Smith, 2019).

Recognition that coercive control is typically a key and serious aspect of domestic violence has led to steps in some Australian jurisdictions to criminalise coercive control rather than allow redress only under civil law. The Australian Government’s National principles to address coercive control: Consultation draft was released in September 2022 to help facilitate a consistent legal approach to coercive control across Australia (ANROWS, 2021; Meeting of Attorneys-General, 2022; see “Key events regarding violence against women since 2017” in Section 1.1).

Coercively controlling behaviours are not always easy to recognise. In the absence of clear physical signs of abuse, some people can experience ongoing abuse without recognising or correctly labelling their experience as abuse, which inhibits help-seeking behaviour.

The Recognising DV Subscale items asked respondents whether particular behaviours enacted against an intimate partner are forms of domestic violence. The behaviours were deliberately framed to capture comprehension of the controlling or abusive intent of the behaviour. Most respondents (78–96%) recognised the behaviours as always or usually forms of domestic violence (Figure 4-5). Behaviours that include actual or threatened physical harm (D1, D2) or a forced medical procedure (D12) were the most readily recognised as always or usually domestic violence (90–96%; Figure 4-5).

New items were introduced in 2021 that sought to gauge the community’s understanding of particular forms of domestic violence and abuse resulting from the intersection of multiple inequalities. The majority of respondents recognised that threatening, controlling or neglecting a partner in ways that target an aspect of the partner’s identity or experience are always forms of domestic violence, including threats to deport a partner on a temporary visa (73%; D9), threats to put a partner with disability into care or a home (69%; D8), forcing a partner to hide that they are transgender (66%; D11), forcing a partner to stop practising their religion (67%; D10), and refusing to assist with a partner’s disability care needs (67%; D7). However, a concerning minority felt that these behaviours are only sometimes or never a form of domestic violence (13–17%).

Figure 4-5: Recognising domestic violence (UVAWS subscale items), 2021

Note:  N = 19,100 unless otherwise noted. Percentages in the figure do not always exactly correspond to percentages in the text due to rounding. Significant differences over time are based on the percentage of respondents who answered “yes” the behaviour is violence against women either “always” or “usually”.

ns No significant difference between 2021 and 2017.

a New item in 2021. Thus, change over time could not be examined.

* Significantly higher understanding in 2021 than 2017.

~ Asked of one quarter of the sample.

As noted earlier, the Recognise DV Subscale showed significant improvement over time, including between 2017 and 2021 (Figure 4-2). Table 4-2 shows the level of agreement with the items in this subscale over time. Consistent with the significant improvement for the subscale overall, two of the items in the subscale showed a significant improvement since 2017 and another three showed a significant improvement since 2009 and 2013. Specifically, in 2021 compared to 2017, respondents were significantly more likely to recognise that financial abuse

(D5) and electronic monitoring (D6) are always or usually forms of domestic violence. Recognition that physical abuse (D1), restriction of social life (D4) and verbal abuse (D3) are forms of domestic violence has gradually improved across NCAS waves, with significantly higher recognition in 2021 compared to 2009 and 2013, but not compared to 2017. Recognition that threatening family members is a form of domestic violence has remained high across all NCAS waves without showing a significant increase (91–96%; D2; Table 4-2).

Table 4-2: Recognising domestic violence (UVAWS subscale items) over time, 2009 to 2021

…is this a form of violence against women? Percentage of respondents answering strong yes.

Item	Code	2009	2013	2017	2021
Scares or controls partner by threatening family members	D2	92	91	95~	96~
Slaps or pushes partner to cause harm or fear	D1	84*	83*	90	92
Forces partner to undergo an unnecessary medical procedure, such as contraception or abortion a	D12	–	–	–	90~
Controls social life by preventing partner seeing family and friends	D4	70*	73*	83^	87
Repeatedly threatens to deport partner on temporary visa a	D9	–	–	–	85
Repeatedly keeps track of partner on electronic devices	D6	–	–	74*~	83
Controls partner with disability by threatening to put them into care or a	D8	–	–	–	83~
Forces partner to stop practising their religion a	D10	–	–	–	81~
Controls partner by denying them money	D5	53*	54*	66*	81
Controls partner by refusing to assist with their disability needs a	D7	–	–	–	81
Repeatedly criticises to make partner feel bad or useless	D3	70*	71*	80^	83
Controls partner by forcing them to hide that they are trans gender a	D11	–	–	–	78

Note:  Ns in 2009, 2013, 2017 and 2021 were 10,105; 17,517; 17,542; 19,100.

“Strong yes” refers to answering “yes” the behaviour is domestic violence either “always” or “usually”.

a New item in 2021.

* Statistically significant difference between the year indicated and 2021.

~ Asked of one quarter of the sample in this year.

^ Asked of half the sample in this year.

UVAWS in focus: Understand Gendered DV Subscale

As discussed in Section 1.1, population-level victimisation surveys and health data in Australia demonstrate that domestic violence is gendered in that it is disproportionately perpetrated by men against women. The three items in the Understand Gendered DV Subscale of the UVAWS examine the level of understanding that domestic violence is gendered in this way. This subscale consists entirely of items about domestic violence. ^[55]

In Australia, in addition to experiencing higher prevalence of intimate partner violence, women are also more likely to suffer adverse impacts from this violence, including fear or anxiety, physical injury and homicide (ABS, 2017; Australian Domestic and Family Violence Death Review Network & ANROWS, 2022).

The gender imbalance in the perpetration and experience of domestic violence is both a symptom and a reinforcer of gender inequality at the societal level (Our Watch, 2021a). As discussed in Section 1.2, gender inequality is a key driver of violence against women that is reinforced through formal mechanisms, such as laws, policies, systems and structures that maintain economic, social and political inequities, and through informal factors, such as social norms and gendered stereotypes (Our Watch, 2021a).

Acknowledging the gendered pattern of violence does not dismiss the experiences of male victims and survivors. However, it is imperative that we recognise that the most prevalent pattern of domestic violence in Australia is perpetrated by men against women and that we work towards addressing this violence across all levels of society. In this context, businesses, institutions, industries and all levels of government must consider policies, procedures and operational decisions that promote a safe and respectful environment underpinned by gender equality.

The items in the Understand Gendered DV Subscale asked respondents about who mainly commits domestic violence and who is more likely to experience its impacts. Respondents were asked to answer each item by choosing from the response options of “men”, “women” and “both equally”. This binary gender framing of the response options was retained for comparability with previous NCAS waves and simplicity of interpretation. However, there is emerging evidence that non-binary people may experience sexual violence proportionately more than women, although there is limited Australian data on non-binary people’s experience of domestic violence (Heywood et al., 2022; Reisner & Hughto, 2019). Thus, it is important to research, identify, appropriately respond to and prevent violence against all genders.

As Figure 4-6 shows, most respondents recognised that domestic violence is more commonly perpetrated by men (57%; D13). Similarly, Figure 4-7 shows that most respondents recognised that women are more likely to suffer physical harm (76%; D14) and experience fear from domestic violence (70%; D15). However, a substantial proportion of respondents incorrectly indicated that the perpetration (41%; Figure 4-6) and impacts (21–28%; Figure 4-7) of domestic violence were equal for men
and women.

Figure 4-6: Understanding the gendered nature of domestic violence perpetration (UVAWS subscale items), 2021

Note:  N = 4,777. “Men” is the correct answer according to empirical evidence from police and court data (Hulme et al., 2019). Asked of one quarter of the sample in 2021.

ns No significant difference between 2017 and 2021.

Figure 4-7: Understanding the gendered nature of domestic violence impacts (UVAWS subscale items), 2021

Note:  N = 4,777. “Women” is the correct answer according to empirical evidence from the Personal Safety Survey (PSS; ABS, 2017). Percentages in the figure do not always add to 100 due to rounding. Asked of one quarter of the sample in 2021.

ns No significant difference between 2017 and 2021.

a Revised item in 2021. Thus, change over time could not be examined.

Table 4-3 shows change over time for two of the three items in the Understand Gendered DV Subscale. Given that the remaining item (D15) was substantially changed in 2021, it was not possible to reliably examine changes over time for this item. ^[56]  There was a decrease in understanding in 2021 compared to 2009 and 2013 for both subscale items examined over time. Specifically, in 2021 compared to 2009 and 2013, significantly fewer respondents recognised that men are more likely to commit domestic violence (D13) and that women are more likely to experience physical harm from domestic violence (D14). Although the trend towards decreasing understanding continued in raw terms after 2013 for both items, there was no significant decline in understanding between 2017 and 2021.

Table 4-3: Understanding the gendered nature of domestic violence (UVAWS subscale items) over time, 2009 to 2021

Item:
2021: Who is domestic violence mainly committed by?
2009–2017: Do you think that it is mainly men, mainly women or both men and women that COMMIT ACTS of domestic violence?

Code: D13.

Response	2009 (%)	2013 (%)	2017~ (%)	2021~ (%)
Men	74*	71*	64	57
Both equally	23*	25*	32	41
Women	1	2	2	0

Item:
2021: Who is domestic violence mainly committed by?
2009–2017: Do you think that it is mainly men, mainly women or both men and women that COMMIT ACTS of domestic violence

Code: D14

Response	2009 (%)	2013 (%)	2017~ (%)	2021~ (%)
Men	74*	71*	64	57
Both equally	23*	25*	32	41
Women	1	2	2	0

Item:
2021: Who is more likely to experience fear as a result of domestic violence?

Response	2009 (%)	2013 (%)	2017~ (%)	2021~ (%)
Men	–	–	–	1
Both equally	–	–	–	28
Women	–	–	–	70

Note:  Ns in 2009, 2013, 2017 and 2021 were 10,105; 17,517; 17,542; 19,100.

Percentages do not add to 100 as “unsure” and “unanswered” responses are not shown in the table.

* Statistically significant difference compared to 2021.

~ Asked of one quarter of the sample in this year.

4.3 Understanding of violence against women: Assessing the importance of demographics

Contribution of demographics to understanding of violence against women

Efforts to improve community understanding of violence against women are aided by information about the factors that are associated with an individual’s understanding. Multiple linear regression models can assess how well an outcome variable can be predicted or explained by a group of factors considered together and how much remains unexplained. ^[57] A multiple regression model was conducted to examine how well we can predict respondents’ understanding of violence against women (the outcome variable) if we know only their demographic characteristics (UVAWS Model 1). Information about any key differences between demographic groups in understanding of violence against women can assist policymakers and practitioners to target education initiatives more effectively according to the needs of different demographic groups. Together the demographic factors explained 7 per cent of the variance in UVAWS scores (Figure 4-8; UVAWS Model 1). Thus, while demographic characteristics help us to predict understanding of violence against women, their total contribution is only small. Most of the difference in respondents’ understanding of violence against women (93%) cannot be explained by their demographic characteristics alone, suggesting other factors are also important in predicting or shaping understanding.

Figure 4-8: Contribution of demographics to understanding of violence against women (UVAWS scores), 2021

Note:  N = 18,876. Based on UVAWS Model 1.

Demographic characteristics related to understanding of violence against women

Table 4-4 shows the significant demographic predictors of understanding of violence against women based on the regression (UVAWS Model 1). In order of importance (as listed in the table), the significant demographic predictors of understanding of violence against women were gender, English proficiency, country of birth and length of time in Australia, and sexuality. Gender, the most important predictor, explained only about 2 per cent of the variance in understanding (first column in Table 4-4).

Table 4-4 also shows significant differences between demographic groups in understanding of violence based on the regression results (UVAWS Model 1). For each significant demographic predictor, a selected or “reference” group was compared to each other group. For example, for gender, the “comparison” groups of women and non-binary respondents were both contrasted against the “reference” group of men. ^[58]  The table shows whether each comparison group had significantly higher (>), significantly lower (<) or not significantly different (ns) understanding compared to the reference group.

Based on the regression, the demographic groups that had  significantly higher understanding of violence against women were: ^[59]

• gender: women and non-binary respondents compared to men
• English proficiency: respondents who spoke English at home compared to respondents who spoke a language other than English (LOTE) at home
• country of birth and length of time in Australia: Australian-born respondents compared to respondents born in a non-main English–speaking country (N-MESC) who had lived in Australia for less than six years
• sexuality: lesbian respondents compared to heterosexual respondents. ^[60]  In addition, for each significant demographic predictor in the regression, Table 4-4 presents bivariate results showing the percentage of each demographic group with “advanced” understanding of violence against women. ^[61]  For example, for gender, half of women (50%) and non-binary respondents (50%) were categorised as having “advanced” understanding of violence against women, compared to 38 per cent of men. ^[62]  Thus, even though some demographic groups have higher understanding of violence against women, further improvement is needed across all demographic groups to achieve a society where all people have “advanced” understanding.

Table 4-4: Significant demographic predictors of understanding of violence against women (UVAWS score), 2021

Demographic factor – Gender (2%)

Demographic group (% unique contribution to UVAWS scores)	Regression results Significantly higher (>) or lower (<) understanding of violence compared to REF a	Bivariate results % of respondents with “advanced” understanding of violence against women b
Men REF		38
Women	>	50
Non-binary respondents	>	50

Demographic factor – English proficiency ^c (2%)

Demographic group	Regression results	Bivariate results
English at home REF		48
LOTE: good English	<	31
LOTE: poor English	<	22

Demographic factor – Country of birth and length of time in Australia ^d (1%)

Demographic group	Regression results	Bivariate results
Born in Australia REF		48
MESC: 0–5 years	ns
MESC: 6–10 years	ns
MESC: >10 years	ns
N-MESC: 0–5 years	<	21
N-MESC: 6–10 years	ns
N-MESC: >10 years	ns

Demographic factor – Sexuality (0.5%)

Demographic group	Regression results	Bivariate results
Heterosexual REF		44
Lesbian	>	63
Gay	ns
Bisexual or pansexual	ns
Asexual, queer or diverse sexualities	ns

Note:  N = 18,876. Regression results are from UVAWS Model 1. Only significant predictors are shown in the table. The total contribution of the demographic predictors to UVAWS scores was 7%. Age, formal education, main labour activity and socioeconomic status of area were retained in the model because they improved model fit but they were not significant predictors. Disability and remoteness were removed from the final model because they did not improve model fit.

REF The reference group for this demographic factor. All other groups for the demographic factor were compared to the REF. The REF was chosen based on considerations of statistical power (i.e. the group with the most respondents) and ease of interpretation (e.g. comparing the group with the highest formal education to each other group).

ns No significant difference between this demographic group and the REF.

a Based on the regression results, this demographic group had significantly higher (>), significantly lower (<) or not significantly different (ns) understanding of violence against women compared with the REF. For example, for gender, the table shows that women and non-binary respondents had  significantly higher ( >) understanding compared to men (the REF). It can also be stated that men (the REF) had  significantly lower understanding compared to women and non-binary respondents, but this direction is not shown in the table.

b “Advanced” understanding of violence against women means recognising at least 75% of behaviours as always forms of violence against women or domestic violence (based on UVAWS items) and recognising the remaining UVAWS items as usually forms of violence. See Section 2.5 for further details.

c “LOTE” refers to language other than English spoken at home. “Good English” refers to good or very good self-reported English proficiency and “poor English” refers to no English or poor self-reported English proficiency.

d “MESC” refers to people born in a main English–speaking overseas country (ABS classification) and “N-MESC” refers to people born in a non-main English–speaking country. The number of years refers to the number of years since the respondent moved to Australia.

4.4 Conclusions about understanding of violence against women

Understanding the nature of violence against women, including distinguishing abusive behaviours from healthy relationship dynamics and understanding the gendered nature of violence, is important:

• for victims and survivors to promptly recognise abuse and seek help
• for family members, friends and service providers to recognise abuse, validate the victim’s and survivor’s experience and support them in a trauma-informed way
• for family members, friends, employers, businesses and service providers to call out abusive actions when safe to do so
• for government, police, men’s behaviour change services and employers to hold perpetrators responsible for their actions and call out abusive actions
• for perpetrators to recognise their abusive behaviours and the impact of those behaviours so that they can change.

The results in this chapter show that community understanding of the behaviours that constitute violence against women is significantly improving overall, and that most Australians recognise that domestic violence and violence against women can manifest as a range of violent, abusive and controlling behaviours. However, Australians are generally better at recognising behaviours that constitute domestic violence than they are at understanding that domestic violence is a gendered phenomenon disproportionately perpetrated by men against women.

There were also some significant, albeit small, differences between demographic groups in understanding of violence against women according to the regression analysis. The strongest demographic relationship was for gender, with men demonstrating significantly lower understanding of violence against women compared to both women and non-binary respondents. Understanding also differed based on English proficiency, country of birth and length of time in Australia, and sexuality. Importantly, however, the combined ability of demographic factors to predict levels of understanding was only small (7%). Thus, the results indicate that there is room to improve levels of understanding of violence against women across the Australian population.

These results have implications for violence education and prevention initiatives. Although Australians’ recognition of the forms of domestic violence and violence against women is improving, more needs to be done to address remaining gaps in understanding. Initiatives should therefore:

• Develop consistent definitions of domestic violence and coercive control across legislative and policy settings Australia-wide, recognising domestic violence as an ongoing pattern of multiple forms of  violence, abuse and  control (Carlisle et al., 2022; Meeting of Attorneys-General, 2022).
• Ensure these consistent definitions are used across education and prevention initiatives to facilitate shared understanding and competence to accurately name and respond to abusive behaviours.
• Increase recognition of the many “subtle” or non-physical forms of domestic violence and violence against women more broadly, including coercive control, to correct perceptions that violence against women equates to predominantly physical violence (Carlisle et al., 2022; Meeting of Attorneys-General, 2022).
• Expand understanding and awareness of technology-facilitated abuse, including its interrelationship with coercive control.
• Help people distinguish between healthy and unhealthy relationship interactions.
• Raise awareness of the ways intersecting inequalities in societal systems and structures exacerbate risk of violence for marginalised groups and produce unique forms of violence against women.
• Raise awareness of perpetrator tactics of control, violence and abuse which target a partner’s identity or needs, such as spiritual abuse, migration abuse, financial abuse, carer abuse and threats to “out” a partner’s gender identity or sexuality.
• Support industries, businesses, service providers and governments to create policies to identify, appropriately respond to and prevent violence against women within their spaces.
• Increase the level of “advanced” understanding of violence against women across the population and social ecology by addressing barriers and employing enablers to understanding, including barriers and enablers that may be particularly relevant to certain demographic groups (Chapter 9). ^[63]

The present results also suggest that considerable proportions of Australians may be conceptualising domestic violence through a “gender-ignoring” or gender-neutral lens, which may be incorrectly shaping perceptions that men and women equally perpetrate domestic violence (Carlisle et al., 2022; Our Watch, 2021a). This perception focuses on the importance of being “fair” by treating everyone the same but fails to recognise the gendered norms and gendered differences within structures and systems that drive gender-based inequalities and violence. Thus, the results underscore the importance of raising awareness of the gendered norms and inequalities in structures and systems that drive violence against women and addressing this gender-ignoring lens. For example, prevention initiatives should:

• Address any scepticism and misconceptions among the community about the gendered nature of domestic violence and abuse by raising awareness of the established statistics on this issue.
• Improve community understanding about the structural gender inequalities and other inequalities that drive the conditions for men’s predominant use of violence, abuse and control.
• Adopt gender-transformative strategies to target and address the gendered norms and other drivers of violence, abuse and control at all levels of the social ecology.
• Address “backlash”, or resistance towards gender equality movements, as these attitudes may underlie perceptions that men and women equally perpetrate domestic violence.
• Employ respectful relationships education to emphasise both the importance of an equal power balance in respectful relationships and the barriers to this in the current patriarchal and heteronormative society, as well as to transform problematic gendered expectations.
• Address barriers to understanding violence against women across the population and at all levels of the social ecology. ^[64]

5 Findings:
Attitudes towards Gender Inequality Scale (AGIS)

Gender inequality is a social phenomenon in which women and men do not have equal social standing, value, power, resources or opportunities in society, providing the context for violence against women to proliferate and become entrenched (Our Watch, 2021a). Thus, systems and structures within society can either challenge or perpetuate gender inequality. Community attitudes that condone gender stereotyping or gender discrimination are an expression of gender inequality and these attitudes function to reinforce and reproduce gender inequality in society. Attitudes condoning gender inequality have been repeatedly associated with attitudes that condone violence against women (Flood, 2019b; Our Watch, 2021a, p. 36; Webster et al., 2014; Webster et al., 2018a; Webster et al., 2021).

Achieving gender equality requires changes throughout society, including via changes to individual attitudes and to practices, systems and structures at the organisational, institutional and societal levels. It has been argued that achieving gender equality may also require strategies focused on equity (or fairness) to compensate for women’s historical and social disadvantages that prevent women and men from otherwise operating on a level playing field (WHO, 2011).

This chapter presents the results for the Attitudes towards Gender Inequality Scale (AGIS), including:

• AGIS scores over time by gender (Section 5.1)
• scores for the five AGIS subscales, which examine rejection of five aspects of attitudes condoning gender inequality (Section 5.2)
• results for individual AGIS items in each subscale (Section 5.2)
• predictors of AGIS scores, including demographic factors and understanding of violence against women (Section 5.3)
• the conclusions and implications arising from these results (Section 5.4).

5.1 Attitudes towards gender inequality over time by gender

Figure 5-1 presents the change in rejection of gender inequality over time by gender, according to mean AGIS scores. We could not examine change over time in attitudes for non-binary respondents as non-binary genders were not reliably captured in previous waves of the NCAS. However, we updated the gender item in 2021 to capture non-binary genders more accurately and are able to provide the mean AGIS score for non-binary respondents in 2021. ^[65]

For all respondents, and for men and women separately, the mean AGIS score was significantly higher in 2021 compared with 2009, 2013 and 2017. These findings indicate a significant increase since all previous NCAS waves in the attitudinal rejection of gender inequality in the Australian population overall and for both men and women separately.

Examining only 2021 AGIS scores, there were significant gender differences. Specifically:

• compared to men, women continue to have significantly higher attitudinal rejection of gender inequality in 2021, as they did in previous NCAS waves
• compared to women and men, non-binary respondents had significantly higher rejection of gender inequality in 2021.

Figure 5-1: Attitudinal rejection of gender inequality over time (AGIS scores) by gender, 2009 to 2021

Note: “na” below means reliable data was not available.  Ns in 2009, 2013, 2017 and 2021 were:

women – 5,532; 8,728; 9,275; 10,095

men – 3,377; 6,450; 8,215; 8,827

non-binary respondents – na; na; na; 81

all – 8,909; 15,178; 17,528; 19,040.

Demographic items for gender were updated for the 2021 NCAS in accordance with the ABS Standard (ABS, 2021h). As the gender item in previous survey waves did not include the same response options for non-binary respondents, only results for men and women can be compared over time.

* Statistically significant difference on this scale between the year indicated and 2021.

*1 Statistically significant difference compared to women and men in 2021.

*2 Statistically significant difference compared to men in 2021.

5.2 Attitudes towards gender inequality: AGIS subscales

Figure 5-2 displays changes over time for the five AGIS subscales between 2017 and 2021. ^[66]  There were improvements over time for four of the five subscales. The mean score for rejection of the Limit Autonomy Subscale was significantly higher in 2021 compared to all three previous waves of the NCAS. The Reinforce Gender Roles Subscale, Normalise Sexism Subscale and Deny Inequality Subscale also showed significantly higher rejection of gender inequality in 2021 compared to 2017. However, the Undermine Leadership Subscale showed no significant change in 2021 compared to 2017. ^[67]

The mean scores on the different AGIS subscales in 2021 were also compared to one another to examine whether some types of problematic attitudes towards gender inequality are more likely to be rejected than others (Figure 5-2). There was little difference in the Australian population’s rejection of the different aspects of gender inequality, with only one significant difference between the AGIS subscales. Specifically, the mean score for rejection of the Normalise Sexism Subscale was significantly lower than that for the Undermine Leadership Subscale in 2021. This finding is consistent with the considerable level of acceptance of the sexist joke scenarios described in Chapter 8.

Figure 5-2: Rejection of different aspects of gender inequality (AGIS subscales) over time, 2009 to 2021

Note: “na” below means reliable data was not available.  Ns in 2009, 2013, 2017 and 2021 were:

Reinforce Gender Roles Subscale – na; na; 17,018; 9,509

Undermine Leadership Subscale – na; na; 17,398; 18,792

Limit Autonomy Subscale – 9,549; 16,396; 16,674; 18,356

Normalise Sexism Subscale – na; na; 17,248; 6,467

Deny Inequality Subscale – na; na; 16,077; 9,316.

* Statistically significant difference on this subscale between the year indicated and 2021.

*1 The Normalise Sexism Subscale had a significantly lower mean score compared to the Undermine Leadership Subscale in 2021.

Figure 5-3 displays the mean scores on each AGIS subscale in 2021 by gender. There were significant differences between genders for all AGIS subscales in 2021. Specifically:

• women demonstrated significantly higher rejection of the aspects of gender inequality measured by all five AGIS subscales compared to men
• non-binary respondents demonstrated significantly higher rejection of gender inequality on four of the five AGIS subscales, showing higher rejection on:
• • the Reinforce Gender Roles Subscale, Undermine Leadership Subscale and Deny Inequality Subscale compared to men
  • the Normalise Sexism Subscale compared to both men and women.

Figure 5-3: Rejection of different aspects of gender inequality (AGIS subscales) by gender, 2021

Note:  N = 19,100 unless otherwise noted.

* Statistically significant difference compared to men on this subscale in 2021.

*1 Statistically significant difference compared to men and women on this subscale in 2021.

~ Asked of one quarter of the sample.

AGIS in focus: Reinforce Gender Roles Subscale

The Reinforce Gender Roles Subscale of the AGIS includes five items examining attitudes to traditional, rigid, heteronormative gender roles and expectations. Gender roles and stereotypes relate to common, oversimplified assumptions about the characteristics, skills, behaviours, preferences and roles that people have or demonstrate based on their biological sex. Although stereotypes and expected gender roles are often perceived as natural or innate, they are socially constructed and learned through socialisation (Basu et al., 2017; Our Watch, 2021a; Reigeluth & Addis, 2016). Research suggests that children are often socialised into traditional gender roles by early childhood and have internalised inequitable gender attitudes by pre-adolescence (Gutierrez et al., 2020; Hammond & Cimpian, 2021; Kågesten et al., 2016; Mayeza & Bhana, 2020). Despite women’s greater participation in non-traditional domains and efforts to shift stereotypes over time, recent attitudinal research suggests that many gender role stereotypes persist, including in academia, business and the private domain (Hipp & Bünning, 2021; Marques, 2021; V. Meyer et al., 2017; Morawska et al., 2021; Stout et al., 2016).

Gender roles and expectations for men include biologically essentialist ideas about being a “real man”, defined by acting “tough”; demonstrating aggression, dominance and control; self-reliance; suppression of “feminine” emotions; and adhering to compulsory heterosexuality, hypersexuality and sexual entitlement (Mahalik et al., 2003; Our Watch, 2021a; The Men’s Project & Flood, 2018). In line with these traditional stereotypes, Australian men continue to be substantially less likely than women to take up primary caring roles or parental leave and remain less likely to seek employment in “caring” industries (WGEA, 2022a).

Additionally, stereotyped gender norms for women include expectations relating to their demeanour (e.g. passive, emotional and submissive), appearance (e.g. sexualised, pretty, thin and adhering to beauty norms of whiteness), and character (e.g. caring and maternal, but also “bitchy”, inherently deceitful or manipulative, out to “get men”; Biefeld et al., 2021; McCann, 2022; Minter et al., 2021; Our Watch, 2021a). Exemplifying the stereotyped contradictions which women are subjected to, the Madonna–whore dichotomy denotes polarised perceptions of women as either good and chaste or bad and promiscuous (Bareket et al., 2018). A recent international study found endorsement of the Madonna–whore dichotomy correlated with the endorsement of patriarchy-supporting ideologies, confirming the role of stereotypes in controlling women and limiting their sexual freedom (Kahalon et al., 2019). Similarly, stereotyped assumptions about women’s “natural” desire for children posit motherhood as a defining feature of women’s identities in Australia across cultural contexts. As a result, women who cannot or choose not to have children face stigma and are cast as “incomplete” or not “real” women (Bhambhani & Inbanathan, 2018; Gui, 2020; Iverson et al., 2020; Riggs & Bartholomaeus, 2016).

Rigid gender roles and stereotypes impact the ways people relate to each other in their relationships, as well as in organisational and institutional contexts (Our Watch, 2021a). Rigid gender norms and stereotypes can also have harmful effects. Expecting women to be passive and submissive can reduce women to sexual objects and targets for sexual exploitation (Bernstein et al., 2022a; C. Knowles, 2021; Schick, 2014; Wright & Tokunaga, 2016). Likewise, gender expectations that women must be nurturing and caring, rather than driven and ambitious, result in biases that undermine women’s independence and autonomy in workplaces and public life (Barreto et al., 2009; Hideg & Shen, 2019). Patriarchal gender roles can also be harmful for all people, including men and non-binary people. Harms for men have been evidenced across a range of indicators relating to men’s mental health, wellbeing and reluctance to seek help; suicide; their proclivity for risk-taking behaviours (such as alcohol use); and their risk of perpetrating or experiencing violence and perpetrating sexual harassment of women (Apesoa-Varano et al., 2018; Flood, 2022a; Murnen, 2015; Rice et al., 2021; The Men’s Project & Flood, 2018). Rigid gender roles also lead to hostile climates for non-binary people, who can feel pressure to express gender in a cisnormative way or face exclusion and discrimination (Francis & Monakali, 2021). Thus, adhering to and condoning rigid gender norms ultimately contributes to and reinforces gendered oppressions, thereby maintaining a context that enables gender-based violence to occur (Abrams et al., 2003; Koepke et al., 2014).

Figure 5-4 shows the level of agreement or disagreement with each item in the Reinforce Gender Roles Subscale. Disagreement with an item indicates rejection of attitudes that support rigid gender roles and stereotypes. Most respondents strongly or somewhat  disagreed with each statement (84–94%) and only a minority strongly or somewhat agreed (4–7%). These results indicate that, positively, Australians predominantly reject attitudes that reinforce rigid gender roles and stereotypes.

Nonetheless, some of the subscale items were more strongly rejected than others (Figure 5-4). Rejection was strongest for attitudes that chastise men for working in stereotypically “feminine” industries (G7) and for expressing emotion (G8), with 78–80 per cent of respondents strongly disagreeing with these items. Comparatively, only 59 per cent of respondents strongly disagreed with expectations that women should not initiate sex when a couple starts dating (G15), suggesting that there is still room to shift this traditional heteronormative gender role expectation and proscriptive sexual scripts that limit women’s sexual agency.

Figure 5-4: Reinforcing rigid gender roles (AGIS subscale items), 2021

Note: N = 19,100 unless otherwise noted. Significant differences over time are based on the percentage of respondents who answered “strongly disagree” or “somewhat disagree”.

ns No significant difference between 2017 and 2021.

^ Asked of half the sample..

It was also of interest to investigate if the generally high level of rejection of rigid gender roles and stereotypes evidenced in 2021 represents an improvement compared to previous survey waves. Table 5-1 shows the results for the Reinforce Gender Roles Subscale items over time. Although, as noted earlier, attitudes on the overall subscale improved significantly between 2017 and 2021 (Figure 5-2), this improvement did not translate to a significant improvement for any of the individual items in this subscale between 2017 and 2021 (despite small increases in the raw percentages). These results suggest that while attitudes reinforcing gender roles and stereotypes are rejected by most Australians, such attitudes are slow to change in a minority of the Australian population.

Table 5-1: Reinforcing rigid gender roles (AGIS subscale items) over time, 2009–2021

% net disagree ^a

Item	Code	2009	2013	2017	2021
I think it is embarrassing for a man to have a job that is usually held by a woman	G7	–	–	93^	94
A man should never admit when others have hurt his feelings	G8	–	–	92	93
Women need to have children to be fulfilled	G9	84	83	90	89^
If a woman earns more than her male partner, it is not good for the relationship	G14	–	–	89^	90^
When a couple start dating, the woman should not be the one to initiate sex	G15	–	–	80^	84^

Note:  Ns in 2009, 2013, 2017 and 2021 were 10,105; 17,517; 17,542; 19,100.

There were no significant differences between previous years and 2021 on any items in this subscale.

a Percentage of respondents who strongly or somewhat disagreed with the item.

^ Asked of half the sample in this year.

AGIS in focus: Undermine Leadership Subscale

The Undermine Leadership Subscale of the AGIS includes four items relating to attitudes towards women in work and leadership. Although women’s ascent to leadership positions in politics and other career areas has accelerated in recent years, the gender pay gap in Australia continues to favour men and is evident across industries, and most senior roles and governing bodies in Australia remain dominated by men (WGEA, 2022b). The recent review into Commonwealth parliamentary workplaces found that Parliament is a highly gender-segregated workplace and that women face challenges in attaining senior roles or are given tasks on a gendered basis, in line with stereotyped gender expectations (AHRC, 2021). These challenges to progression in parliamentary leadership were amplified for women, people with diverse sexualities, people with disability and people from various cultural and linguistic backgrounds (AHRC, 2021). According to the review, Commonwealth parliamentary workplaces are characterised by a lack of diversity and a “boy’s club” culture (AHRC, 2021).

Political institutions in Australia have also been found to produce unequal and unsafe working conditions for women politicians (Collier & Raney, 2018). Recent examples of the culture of hostility towards women in politics and leadership roles in Australia since the 2017 NCAS wave include:

• sexist and offensive remarks directed at Greens Senator Sarah Hanson-Young in the Senate in 2018 (McKinnell, 2019; Women’s Agenda, 2021)
• sexist and abusive social media backlash directed at journalist Leigh Sales following her interview with Prime Minister Scott Morrison on the ABC’s  7.30 television program in 2020 (Molloy, 2020)
• “growling” and dog noises directed at Senator Jacqui Lambie during Senate Question Time in 2021 (Maiden, 2021)
• allegations of sexism, bullying, harassment and sexual misconduct across political parties and at local, state and federal levels of government (AHRC, 2021; L. Knowles, 2018; Mikolajczak et al., 2021; Williams, 2020a; Worthington & Snape, 2021).

Additionally, there is extensive evidence of systemic gender bias in media coverage of women as political leaders, including against former Australian Prime Minister Julia Gillard (Jansens, 2019; Lee-Koo & Maley, 2017; Sawer, 2013; Sorrentino et al., 2018; Williams, 2017, 2020b). Hostile sexism, which involves antagonistic views of women, was also found to influence voters’ preferences in the United States for Donald Trump over Hillary Clinton in the 2016 presidential election (Ratliff et al., 2019). Despite such evidence of attitudes and cultures that undermine women’s leadership, recent research indicates that countries led by women fared better than those led by men in terms of COVID-19 outcomes, as women leaders locked their countries down more quickly and communicated more effectively (Garikipati & Kambhampati, 2021).

Senior and leadership roles in various industries continue to be dominated by men and remain imagined in masculine terms (Poorhosseinzadeh et al., 2019). Attitudes relating to the traits of authentic and influential leaders are often constructed along the lines of gendered, racial and sexuality hierarchies (Liu, 2021; Liu et al., 2015; Stephenson, 2020). Gendered and racialised barriers to leadership have been evidenced in medicine, academia, public relations, science, information technology, tourism and hospitality, among many other industries (Filiberto et al., 2019; Fitzsimmons & Callan, 2020; Hutchings et al., 2020; Khan et al., 2019; Liang et al., 2019; Mate et al., 2019; McGee, 2018; Nash et al., 2019; Parker et al., 2018; Parkinson et al., 2019; Place & Vardeman-Winter, 2018; Punshon et al., 2019; A. N. Smith et al., 2019; Wolfert et al., 2019). Moreover, research shows that race and gender often intersect to impact on the salaries, treatment (e.g. workplace bullying) and leadership opportunities of women of colour across various industries (Aaron, 2020; Bourabain, 2021; Burton et al., 2020; T. Clark et al.2021; Hollis, 2018; Levchak, 2018; Macias & Stephens, 2019; Quah, 2020). One study reported that women feel pressured to adopt more “masculine” gender performances to secure top-level managerial positions (Einarsdottir et al., 2018). Additionally, taking parental leave both interrupts women’s career advancement into leadership positions and contributes to their longer term financial insecurity due to gendered pay gaps and reduced retirement savings (Baird & Heron, 2019; Offermann et al., 2020; Volpato, 2018). Online abuse has also been linked to negative personal and career consequences for professional women, including pulling back from career and public life, a suspension of online professional activity and resignation (eSafety, 2022j).

Recent evidence suggests that the COVID-19 pandemic has strongly impacted women’s workplace participation and exacerbated pre-existing gender inequalities in the labour force (R. Cook & Grimshaw, 2021; Landivar et al., 2020). Additionally, policy roadmaps out of lockdowns and stimulus packages, both in Australia and internationally, were seen as favouring men and men-dominated industries over women and women-dominated industries, thereby perpetuating the employment disadvantages women faced through the pandemic response (Australian Unions, 2020; C. Johnson, 2022; M. Morris, 2020; Viswanath & Mullins, 2021; Wood et al., 2021). The COVID-19 pandemic thus created difficulties for women’s economic participation and career advancement. For example, compared with men in academic positions, women’s research productivity was especially impacted through the pandemic as a result of women’s increased unpaid caring responsibilities (Andersen et al., 2020; Gabster et al., 2020; Pinho-Gomes et al., 2020). Moreover, the effects of the pandemic on career advancement were not felt equally by all women. Research suggests that women with lower socioeconomic status were more likely to have experienced a decrease in their work hours, while women in higher positions or with advanced degrees were found to have experienced an increase in paid work hours (Fan & Moen, 2021). In addition, government spending in response to COVID-19 may have limited the speed and scope of reforms to areas of spending that primarily affect women, including childcare and parental leave (Wood et al., 2021). Additional consequences of the response to the COVID-19 pandemic include an increase in women experiencing online abuse and harassment as a result of working from home and shifting work to online forums (Ahuja & Padhy, 2021; Strenio & Chowdhury, 2021).

Figure 5-5 shows the level of agreement or disagreement with each item in the Undermine Leadership Subscale. Disagreement with an item indicates rejection of attitudes that undermine women’s leadership in public life. Most respondents strongly or somewhat disagreed with each subscale item (85–93%) and only 10 per cent or fewer agreed (strongly or somewhat) with the items. Positively, these results indicate that, except for a small minority, Australians overwhelmingly reject attitudes that undermine women’s leadership and decision-making in public life. Nonetheless, the level of rejection was higher for some of these items than others. While 83 per cent of respondents strongly disagreed that women are less capable of thinking logically (G11), only about two thirds strongly disagreed that men generally make better bosses (G5) and political leaders (G4) than women. Thus, further improvement could be made in attitudes towards women in leadership roles.

Figure 5-5: Undermining women’s leadership in public life (AGIS subscale items), 2021

Note: N = 19,100 unless otherwise noted.

ns No significant difference between 2017 and 2021.

^ Asked of half of the sample.

Table 5-2 shows the level of disagreement with the Undermine Leadership Subscale items over time. Consistent with the lack of significant improvement between 2017 and 2021 for the subscale overall (Figure 5-2), none of the individual items showed improvement since 2017. These findings indicate that the Australian population’s fairly high level of rejection of attitudes that undermine women’s leadership evidenced in 2021 was similar to that demonstrated in 2017. However, one item showed significant improvement in 2021 compared to 2009 and 2013. Specifically, in 2021 a significantly higher percentage of respondents disagreed (either strongly or somewhat) with the statement that men make better political leaders (G4) than in 2009 and 2013.

Table 5-2: Undermining women’s leadership in public life (AGIS subscale items) over time, 2009–2021

% net disagree ^a

Item	Code	2009	2013	2017	2021
Women are less capable than men of thinking logically	G11	–	–	92	93
Men, rather than women, should hold positions of responsibility in the community	G6	–	–	87^	88^
On the whole, men make better political leaders than women	G4	71*	67*	80	85
In the workplace, men generally make more capable bosses than women	G5	–	–	81^	85^

Note:  Ns in 2009, 2013, 2017 and 2021 were 10,105; 17,517; 17,542; 19,100.

a Percentage of respondents who strongly or somewhat disagreed with the item.

* Statistically significant difference compared to 2021.

^ Asked of half the sample in this year.

AGIS in focus: Limit Autonomy Subscale

The Limit Autonomy Subscale of the AGIS includes two items that examine attitudes to men being in charge or taking control in their intimate relationships with women. Traditional, heteronormative gender roles shape expectations about men’s and women’s roles and responsibilities within intimate relationships. Benevolent sexism further reinforces these gender expectations by positing that women are innately more nurturing and thus best suited to be primary caregivers and passive homemakers, while men are positioned in the role of protector, provider and decision-maker in relationships (Glick & Fiske, 1996, 1997; Hammond et al., 2017; Overall & Hammond, 2017; Salin et al., 2018). Reinforcement of women’s conformity to traditionally feminine attributes (such as being nurturing, gentle, demure and self-sacrificing for others) can undermine their feelings of independence, competence and ambition both within and outside relationships (Cross & Overall, 2018). Moreover, these attitudes establish and maintain a gendered hierarchy of power within intimate relationships, especially in relation to caring responsibilities and financial decision-making. Research has demonstrated how entrenched these gender inequitable attitudes remain. For example, data comparing attitudes towards sharing paid work and unpaid care responsibilities from 22 Western countries suggests that the model of the man as the main income provider remains the most widely supported (Salin et al., 2018).

Expectations that men are the decision-makers and main income providers in relationships, and that women should sacrifice themselves for their family, can create a context where men may resort to aggression or abuse if they feel that their power or status within the relationship is threatened by traditional gender roles being challenged (Cross & Overall, 2019; Cross et al., 2019; Harrington et al., 2021).

Figure 5-6 shows the results for the two items in the Limit Autonomy Subscale. In 2021, 87 per cent of respondents strongly or somewhat disagreed that men should be in a position of control in intimate relationships (G12). Although most respondents also strongly or somewhat disagreed that women prefer men to take charge in relationships (G13), this percentage was somewhat lower (74%). These results indicate that while most Australians reject the normative statement that men  should be in charge of relationships, a sizeable minority nonetheless think that women  prefer men to take control. Thus, more work is needed to challenge deep-seated attitudes that presume men’s patriarchal position in the family and intimate relationships, and to promote equality within intimate relationship dynamics.

Figure 5-6: Limiting women’s personal autonomy in relationships (AGIS subscale items), 2021

Note: N = 19,100.

ns No significant difference between 2017 and 2021.

As noted earlier, scores on the Limit Autonomy Subscale improved significantly in 2021 compared to 2017, 2013 and 2009 (Figure 5-2). Table 5-3 shows the level of disagreement with the items in the Limit Autonomy Subscale over time. Consistent with the improvement at the subscale level, both subscale items showed a significant increase between 2013 and 2021 in the rejection of attitudes that limit women’s personal autonomy in relationships. However, the raw trend towards continued improvement from 2017 to 2021 did not reach statistical significance (Table 5-3). Thus, a shift away from more traditional attitudes that normalise men’s control within relationships appears to be occurring slowly over time.

Table 5-3: Limiting women’s personal autonomy in relationships over time (AGIS subscale items), 2009–2021

% net disagree ^a

Item	Code	2009	2013	2017	2021
Women prefer a man to be in charge of the relationship	G13	65	62*	67	74
Men should take control in relationships and be the head of the household	G12	79*	78*	80	87

Note:  Ns in 2009, 2013, 2017 and 2021 were 10,105; 17,517; 17,542; 19,100.

a Percentage of respondents who strongly or somewhat disagreed with the item.

* Statistically significant difference compared to 2021.

AGIS in focus: Normalise Sexism Subscale

The Normalise Sexism Subscale of the AGIS includes three items describing attitudes that downplay or normalise sexism in specific social contexts. Expressions of everyday sexism, harassment and sexist microaggressions play a role in normalising the culture of violence against women, with the result that sexualised disrespect, discrimination and abuse come to be seen as acceptable and normal (L. Bates, 2014; Mellgren et al., 2018; Sinko et al., 2021). For example, it has been argued that universities have normalised men’s sexual aggression through poor management of sexual assault cases, which ultimately perpetuates and normalises a broader “hook-up” and rape culture (AHRC, 2017a; Heywood et al., 2022; Nisbet et al., 2022; Sinko et al., 2021). Additionally, the normalisation of everyday sexism makes gendered microaggressions, such as jokes about violence against women or questioning the reality of people’s experiences of discrimination, appear acceptable (Algner & Lorenz, 2022; V. E. Johnson et al., 2021). Women’s experiences of gendered microaggressions and everyday sexism have been widely documented, with studies suggesting these experiences are even more acute for women of colour, people with diverse sexualities, and women with disability (Arayasirikul & Wilson, 2019; L. Bates, 2014; V. E. Johnson et al., 2021; A. M. Jones, 2021; Levchak, 2018; Nadal, 2019a; Nadal et al., 2016; Nuru & Arendt, 2019; Olkin et al., 2019).

Research indicates that everyday sexism and microaggressions are especially normalised through men’s peer groups as key sites where gender inequalities and tolerance of violence against women are maintained (DeKeseredy, Hall-Sanchez, et al., 2018; Flood & Ertel, 2020, p. 194). Men’s “locker-room talk”, “banter”, fighting “for fun” and “lad culture” have all been identified as tools of cis men’s heterosexual bonding and intimacy, as well as serving as performances of hegemonic masculinity, especially among younger men (Bolton et al., 2021; Flood, 2008; Jeffries, 2020; Johansson & Odenbring, 2021; Odenbring & Johansson, 2021; Vaynman et al., 2020; Whittle et al., 2019). These homosocial practices – which can include the objectification and harassment of women – are based on normative expectations regarding how men need to act in order to attain and retain masculine status and achieve belonging among their peers (Bolton et al., 2021; Van Doorn et al., 2021). Because performances of traditional masculinity are central to social belonging in men’s peer groups, “men often refrain from intervening in other men’s sexism or violence because of concerns about loss of status among male peers” (Flood & Ertel, 2020, p. 194).

However, these practices can have harmful outcomes and implications. Men’s hegemonic and hypermasculine bonding has been linked with bullying, sexist behaviour and enactments of homophobia (Diefendorf & Bridges, 2020; C. Jackson & Sundaram, 2018; R. A. Miller et al., 2021; Rosen & Nofziger, 2019). Moreover, research has shown how hostile masculinity in men’s peer groups is associated with proclivity to perpetrate multiple types of violence (E. Miller et al., 2020; Ray & Parkhill, 2021), including “upskirting” (i.e. taking a photo up a woman’s skirt), the non-consensual sharing of sexual images and videos, dating violence and even sexual assault (Durán et al., 2016; Hall et al., 2021; Hunehäll Berndtsson & Odenbring, 2021; Ringrose et al., 2022; S. Roberts et al., 2021). Peer pressure for men to engage in “locker-room talk” has been associated with rape myth acceptance and problematic attitudes towards women (Cole et al., 2020). It has been argued that dismissing these forms of hypermasculine bonding as instances of “lad culture” or “just a laugh” can mask the problematic elements of these behaviours, ultimately normalising disrespectful attitudes towards women and gender and sexual minorities and legitimising sexual violence (C. Jackson & Sundaram, 2018; Nichols, 2018; Vaynman et al., 2020).

Figure 5-7 shows the level of disagreement with the three items in the Normalise Sexism Subscale. The majority of respondents disagreed, strongly or somewhat, with all three items (82–98%). However, there was stronger rejection of one of these items. Whereas 93 per cent of respondents strongly disagreed that jokes about violence against women are acceptable (G17), only two thirds (66%) strongly disagreed that workplace discrimination against women is no longer a problem (G10) and only 57 per cent strongly disagreed that sexist jokes are acceptable (G16). These findings suggest that although expressions of violent behaviour among friends are not tolerated, more work is needed to challenge attitudes that microaggressive and sexist behaviour, including sexist jokes among men in their peer groups, is acceptable. Acceptance of such sexist behaviour creates a context whereby gendered discrimination is no longer seen as a problem.

Figure 5-7: Normalising sexism (AGIS subscale items), 2021

Note: N = 19,100 unless otherwise noted. Percentages in the figure do not always exactly correspond to percentages in the text due to rounding. Significant differences over time are based on the percentage of respondents who answered “strongly disagree” or “somewhat disagree”.

ns No significant difference between 2017 and 2021.

* Significantly higher understanding in 2021 than 2017.

~ Asked of one quarter of the sample.

^ Asked of half of the sample.

Table 5-4 shows the level of disagreement over time with attitudes that normalise sexism according to the individual items in the Normalise Sexism Subscale. Consistent with the increased rejection of attitudes that normalise sexism at the subscale level (Figure 5-2), there was a significant increase from 2017 to 2021 in disapproval of sexist jokes within men’s peer groups (from 72% to 82%; G16). However, as already noted, this item was less strongly rejected in 2021 than the other Normalise Sexism Subscale items and thus has the greatest room for improvement. There was no significant improvement since 2017 for the other two items, although there was a significant increase between 2013 and 2021 in the percentage of respondents who disagreed that workplace discrimination against women is no longer an issue (G10). In addition, the extremely high rejection of jokes about violence against women shown in 2017 (97%) was maintained in 2021 (98%). Notwithstanding the positive shifts over time in the Normalise Sexism Subscale items, there is still room to improve community attitudes that normalise sexism, particularly through sexist jokes.

Table 5-4: Normalising sexism (AGIS subscale items) over time, 2009–2021

% net disagree ^a

Item	Code	2009	2013	2017	2021
I think it’s OK for men to joke with their male friends about being violent towards women	G17	–	–	97	98~
Discrimination against women is no longer a problem in the workplace in Australia	G10	84	81*	86	90~
I think there’s no harm in men making sexist jokes about women when they are among their male friends	G16	–	–	72^*	82~

Note:  Ns in 2009, 2013, 2017 and 2021 were 10,105; 17,517; 17,542; 19,100.

a Percentage of respondents who strongly or somewhat disagreed with the item.

* Statistically significant difference compared to 2021.

~ Asked of one quarter of the sample in this year.

^ Asked of half the sample in this year.

AGIS in focus: Deny Inequality Subscale

The Deny Inequality Subscale of the AGIS describes attitudes expressing “backlash” to gender equality. “Backlash” is defined as resistance to progressive social change and resistance to change in existing gendered power structures (Flood et al., 2020). Resistance or backlash can be informal or formal, overt or covert, and may be expressed at the individual or collective level (Flood et al., 2020; Smolović Jones et al., 2020). Backlash attitudes aim to maintain the status quo in the social order by impeding or seeking to overturn progressive changes (Flood et al., 2020).

Backlash attitudes and resistance to gender equality take various forms. One example is the denial of the need to change gendered relations of power, specifically “the rejection of the claim that women are disadvantaged and men are privileged” (Flood et al., 2020, p. 396). Another example is “disassociation”, whereby some people construct themselves as separate to the problem of violence against women in order to avoid confronting the ways that they themselves are implicated or complicit in structures of gender inequality and the culture of patriarchy (Burrell, 2020, 2021). Other examples of backlash or resistance include claims that:

• efforts to gain gender equality for women have occurred at the expense of men
• men in general have been unfairly marginalised and subordinated based on the actions of a minority of men
• gender equality advocacy is simply a campaign of “man-hating” ideology
• women are the recipients of unwarranted special treatment
• women are not adequately qualified or are unable to assume the roles that would facilitate their equality
• women deserve their subjugated position
• gender equality is a “women’s issue” and thus is not men’s responsibility (Carian, 2022; Flood & Ertel, 2020; J. Green & Shorrocks, 2021; Horwath & Diabl, 2020, p. 1123; IPSOS, 2022; Tildesley et al., 2021; VicHealth, 2018).

International evidence points to increasing resentment towards gender equality initiatives targeting workplaces and political leadership based on the perception that these initiatives discriminate against men (Elomäki & Ylöstalo, 2021; J. Green & Shorrocks, 2021; Johansson et al., 2019). Relatedly, recent Australian survey results suggest that 42 per cent of men believe gender equality initiatives in the workplace do not take men into account (Haussegger et al., 2018).

Backlash attitudes that resist gender equality are prominent among some groups within the community and are not necessarily held by the majority. Studies suggest that such attitudes are most likely to be held by “individuals who hold sexist norms, and in contexts characterised by sexism, gender segregation and male dominance” (Flood et al., 2020, p. 400), and among younger people and people from lower to middle socioeconomic backgrounds (Carian, 2022). Backlash attitudes have been linked with the acceptance of rape myths (Carian, 2022) and resistance to gender equality initiatives (J. Green & Shorrocks, 2021; Kantola & Lombardo, 2020; Kováts, 2018; Löffler et al., 2020; Pease, 2020; Toldy & Garraio, 2020). Relatedly, an international study suggests that people with politically conservative attitudes, such as backlash attitudes, are less likely to prioritise addressing violence against women (Araújo & Gatto, 2021). Resentment towards gender equality initiatives have also been linked with resentment towards progressive efforts for racial and sexuality equality and may be fuelled by structures and systems based on heteropatriarchal and settler colonial values (L. Nicholas, 2020; L. Nicholas & Agius, 2018; Pease, 2020).

The Deny Inequality Subscale comprises three items describing attitudes that deny gender inequality experiences through backlash. As Figure 5-8 shows, more than half (52–63%) of the respondents strongly or somewhat disagreed with these statements. However, the level of rejection of these backlash attitudes was not particularly high, with only 22–36 per cent of respondents strongly disagreeing with these statements. Further, substantial proportions of respondents agreed with these statements, indicating that approximately 4 in 10 Australians agree that many women mistakenly interpret innocent remarks as sexist (G2), 1 in 3 agree that women exaggerate the unequal treatment of women in Australia (G1) and almost 1 in 3 agree that women do not fully appreciate what men do for them (G3). These results indicate considerable support for backlash attitudes within the Australian community and highlight the need for continued efforts to address backlash attitudes.

Figure 5-8: Denying gender inequality experiences (AGIS subscale items), 2021

Note: N = 19,100 unless otherwise noted. Percentages in the figure do not always exactly correspond to percentages in the text due to rounding. Significant differences over time are based on the percentage of respondents who answered “strongly disagree” or “somewhat disagree”.

ns No significant difference between 2017 and 2021.

* Significantly higher understanding in 2021 than 2017.

^ Asked of half the sample.

As discussed earlier, there was significant improvement between 2017 and 2021 in rejection of attitudes that deny gender inequality at the subscale level (Figure 5-2). Table 5-5 shows the level of rejection over time of attitudes that deny gender equality experiences, according to the individual items in Deny Inequality Subscale. There was a significant positive shift for one of the three items between 2017 and 2021. Specifically, a significantly higher percentage of respondents in 2021 than in 2017 disagreed that many women interpret remarks or acts as sexist (52% versus 41%; G2). The raw trends towards improvement for the other two items did not reach statistical significance (Table 5-5).

Table 5-5: Denying gender inequality experiences (AGIS subscale items) over time, 2017–2021

% net disagree ^a

Item	Code	2017	2021
Many women don’t fully appreciate all that men do for them	G3	57^	63^
Many women exaggerate how unequally women are treated in Australia	G1	53^	58^
Many women mistakenly interpret innocent remarks or acts as being sexist	G2	41*	52

Note: These items were not asked in 2009 and 2013.  Ns in 2017 and 2021 were 17,542; 19,100.

a Percentage of respondents who strongly or somewhat disagreed with the item.

* Statistically significant difference compared to 2021.

^ Asked of half the sample in this year.

5.3 Attitudes towards gender inequality: Assessing the importance of demographics and understanding

Contribution of demographics and understanding to attitudes rejecting gender inequality

Efforts to reduce gender inequality and violence against women are aided by understanding the factors that may underlie an individual’s attitudes towards gender inequality. Four multiple regression models were conducted to examine how well we can predict
respondents’ attitudes towards gender inequality (the outcome variable) if we know their demographic characteristics and their understanding of violence against women (the input variables;  Methodology reminder 5-3  and Section 2.5).

When demographic factors were considered on their own (AGIS Model 1), they explained almost one fifth (18%) of the variance in AGIS scores (Figure 5-9). Thus, while demographic characteristics help us to predict attitudes towards gender inequality, much of the difference in these attitudes (82%) cannot be explained by demographic characteristics alone.

Similarly, when only understanding of violence as measured by the UVAWS was considered as a predictor of AGIS scores (AGIS Model 2), it was a significant predictor and explained approximately one fifth (19%) of the variance in AGIS scores. Thus, improving the community’s understanding of violence against women may assist to improve the rejection of gender inequality. However, most of the difference in respondents’ attitudes towards gender inequality (81%) could not be explained by their understanding, suggesting that other factors are also important in predicting and shaping attitudes towards gender inequality (Figure 5-9).

Another regression (AGIS Model 4) examined which UVAWS subscales were most responsible for the scale-level relationship between the UVAWS and the AGIS. The Recognise DV Subscale was the UVAWS subscale that made the largest contribution to AGIS scores and was a significant predictor of AGIS scores. ^[68] This result suggests that improving community understanding of the different forms of domestic violence, including coercive control, may be an important component of initiatives that aim to improve rejection of gender inequality by increasing understanding of violence.

Considering both demographic factors and UVAWS scores together (AGIS Model 3) improved the ability to predict AGIS scores, with almost one third (31%) of the variance in AGIS scores being explained (Figure 5-9). ^[69]  However, most of the difference in respondents’ attitudes (69%) still could not be explained by respondents’ understanding of violence and their demographic characteristics. Thus, other factors are important in predicting and shaping attitudes towards gender inequality.

Figure 5-9: Contribution of demographics and scale to attitudinal rejection of gender inequality (AGIS scores), 2021

a Based on AGIS Model 1.  N = 18,869.

b Based on AGIS Model 2.  N = 18,868.

cBased on AGIS Model 3.  N = 18,868.

Demographic characteristics related to attitudes towards gender inequality

As noted above, the regression results revealed that demographics considered alone explained 18 per cent of the variation in attitudes towards gender inequality (AGIS Model 1). Information about differences between demographic groups in attitudes towards gender inequality can assist policymakers and practitioners to target attitude change initiatives more effectively according to the needs of different demographic groups. Table 5-6 shows the significant demographic predictors of attitudes towards gender inequality based on the regression (AGIS Model 1). In order of importance (as listed in the table), the significant demographic predictors of attitudes towards gender inequality were gender, formal education, age, English proficiency, sexuality, country of birth and length of time in Australia, and socioeconomic status of area. Gender, the most important predictor, explained 5 per cent of the variance in attitudes towards gender inequality (first column in Table 5-6).

Table 5-6 also shows significant differences between demographic groups in attitudes towards gender inequality based on the regression results (AGIS Model 1). For each significant demographic predictor, a selected or “reference” group was compared to each other group. For example, for gender, the “comparison” groups of women and non-binary respondents were both contrasted against the “reference” group of men. ^[70]  The table shows whether each comparison group had significantly higher (>), significantly lower (<) or not significantly different (ns) understanding compared to the reference group.

Based on the regression, the demographic groups that had  significantly higher rejection of gender inequality were: ^[71]

• gender: women and non-binary respondents compared to men
• formal education: university graduates compared to respondents without university education
• age: all ages on average compared to respondents aged 75 or over
• English proficiency: respondents who spoke English at home compared to respondents who spoke a language other than English (LOTE) at home
• sexuality: lesbian; gay; bisexual or pansexual; and asexual, queer or sexuality-diverse respondents compared to heterosexual respondents
• country of birth and length of time in Australia: Australian-born respondents compared to respondents born in a non-main English–speaking country (N-MESC) who had lived in Australia for less than six years
• socioeconomic status of area: respondents living in areas with the highest economic status compared to those living in areas with the lowest socioeconomic status. ^[72]

In addition, for each significant demographic predictor in the regression, Table 5-6 presents bivariate results showing the percentage of each demographic group with “advanced” attitudinal rejection of gender inequality. ^[73] For example, for gender, about one third of women (35%) and more than half of the non-binary respondents (56%) were categorised as having “advanced” rejection of gender inequality, compared to about one fifth of men (21%). Thus, even though some demographic groups have higher rejection of gender inequality, further improvement is needed across all demographic groups to achieve a society where all people have “advanced” rejection of gender inequality.

Table 5-6: Significant demographic predictors of rejection of gender inequality (AGIS score), 2021

Demographic factor – English proficiency ^c (2%)

Demographic group (% unique contribution to AGIS scores)	Regression results Significantly higher (>) or lower (<) rejection of gender inequality compared to REF a	Bivariate results % of respondents with “advanced” rejection of gender inequality b
Men REF		21
Women	>	35
Non-binary respondents	>	56

Demographic factor – Formal education (3%)

Demographic group	Regression results	Bivariate results
University or higher REF		39
Trade/certificate/diploma	<	26
Secondary or below	<	22

Demographic factor – Age (in years) (2%)

Demographic group	Regression results	Bivariate results
All ages on average REF		28
16–24	ns
25–34	ns
35–44	ns
45–54	ns
55–64	ns
65–74	ns
75+	<	11

Demographic factor – English proficiency ^c (2%)

Demographic group	Regression results	Bivariate results
English at home REF		30
LOTE: good English	<	21
LOTE: poor English	<	13

Demographic factor – Sexuality (2%)

Demographic group	Regression results	Bivariate results
Heterosexual REF		27
Lesbian	>	59
Gay	>	48
Bisexual or pansexual	>	50
Asexual, queer or diverse sexualities	>	52

Demographic factor – Country of birth and length of time in Australia ^d (1%)

Demographic group	Regression results	Bivariate results
Born in Australia REF		30
MESC: 0–5 years	ns
MESC: 6–10 years	ns
MESC: >10 years	ns
N-MESC: 0–5 years	<	21
N-MESC: 6–10 years	ns
N-MESC: >10 years	ns

Demographic factor – Socioeconomic status of area ^e (1%)

Demographic group	Regression results	Bivariate results
5 – Highest status REF		35
1 – Lowest status	<	20
2 – Second-lowest status	ns
3 – Middle status	ns
4 – Second-highest status	ns

Note:  N = 18,869. Regression results are from AGIS Model 1. Only significant predictors are shown. The total contribution of the demographic predictors alone to AGIS scores was 18%. Main labour activity and remoteness of area were retained in the model because they improved model fit, but they were not significant predictors. Disability was removed from the final model because it did not improve model fit.

REF The reference group for this demographic factor. All other groups for the demographic factor were compared to the REF. The REF was chosen based on considerations of statistical power (i.e. the group with the most respondents) and ease of interpretation (e.g. comparing the group with the highest formal education to each other group).

ns No significant difference between this demographic group and the REF.

a Based on the regression results, this demographic group had significantly higher (>), significantly lower (<) or not significantly different (ns) rejection of gender inequality compared with the REF. For example, for gender, the table shows that women and non-binary respondents had  significantly higher (>) rejection compared to men (the REF). It can also be stated that men (the REF) had  significantly lower  understanding compared to women and non-binary respondents, but this direction is not shown in the table.

b “Advanced” rejection of gender inequality means strongly disagreeing with at least 75% of attitudes condoning gender inequality, and somewhat disagreeing with the remaining AGIS items. See Section 2.5 for further details.

c “LOTE” refers to language other than English spoken at home. “Good English” refers to good or very good self-reported English proficiency and “poor English” refers to no English or poor self-reported English proficiency.

d “MESC” refers to people born in a main English–speaking overseas country (ABS classification) and “N-MESC” refers to people born in a non-main English–speaking country. The number of years refers to the number of years since the respondent moved to Australia.

e “Socioeconomic status of area” refers to an ABS measure of socioeconomic conditions in geographic areas in terms of people’s access to material and social resources, and their opportunity to participate in society (SEIFA quintiles).

5.4 Conclusions about attitudes towards gender inequality

Attitudes that endorse gender inequality are both inherently problematic and problematic in practice because they perpetuate a system where some individuals in society, generally men, are valued more highly than others and where individuals are constrained in their interests, participation in society and self-expression (Our Watch, 2021a). In addition, attitudes towards gender inequality have been repeatedly associated with attitudes that condone violence against women (Chapter 6; Webster et al., 2018a).

The results in this chapter show that Australians’ attitudinal rejection of gender inequality continues to improve over time, albeit slowly. While most Australians reject gender inequality, concerning proportions still condone certain attitudes that undermine women’s leadership, reinforce rigid gender roles in specific areas, limit women’s personal autonomy, normalise sexism and deny that gender inequality is a problem.

Tolerance of gender inequality was also significantly stronger for some demographic groups, although these differences were not particularly large. Gender was the most important demographic predictor of attitudes towards gender inequality, with men demonstrating lower rejection of gender inequality than women and non-binary respondents. Notably, however, the combined ability of all demographic factors for predicting attitudes towards gender inequality was relatively small (18%). Thus, there is room to improve rejection of gender inequality across the Australian population. Understanding of violence against women, particularly recognition of the different forms of domestic violence, was also a significant predictor of attitudes towards gender inequality, although again its predictive ability was not large (19%).

These results have implications for informing policy and practice design to reduce violence against women. It appears that, despite the renewed focus on gender inequality and sexism since the previous iteration of the NCAS, some attitudes towards gender inequality persist among a minority of the community and are slow and difficult to change. Policy and prevention efforts should therefore:

• Engage with all genders and all demographic groups across the population to improve attitudes and behaviours that support gender equality.
• Ensure strategies are gender-transformative in their design; that is, ensure initiatives encourage the community to actively challenge and ultimately reject rigid or harmful gender norms, roles, expectations, relations and power imbalances (Our Watch, 2021a).
• Address “backlash” attitudes, or resistance towards gender equality movements, and correct denial of gender inequality experiences wherever they occur across the community; for example, through interventions that engage men as advocates and highlight the mutual benefits of gender equality in intimate relationships and public life (Bell & Flood, 2020; Flood, 2019b; Kingma & Vandeplas, 2022).
• Address the normalisation of everyday sexism and tolerance of sexist microaggressions across social settings, including within organisations and institutions and online.
• Promote equality within intimate relationships and challenge attitudes that presume and accept men’s patriarchal position in the family and intimate relationships.
• Promote gender equality in public life by requiring institutions, organisations and community groups to take responsibility for ensuring that both formal and informal processes provide equal opportunity.
• Challenge attitudes condoning gender inequality and sexism through points of influence, such as peer and social groups.
• Engage school-aged children in respectful relationships education.
• Incorporate knowledge of violence against women components in programs that aim to promote gender equality, including knowledge about the range of behaviours that constitute violence, such as coercive control, as well as training in appropriate responses to signs of abuse.
• Use strengths-based and gender-transformative approaches to effectively engage with men and improve their attitudes towards gender equality.
• Increase the proportion of the population with “advanced” rejection of gender inequality by breaking down barriers and facilitating enablers that are relevant for specific demographic groups (Chapter 9). [74]

6 Findings: Attitudes towards Violence against Women Scale (AVAWS)

6.1 Attitudes towards violence against women over time by gender

Figure 6-1 presents the change in rejection of violence against women over time by gender, according to mean AVAWS scores. We could not examine change over time in attitudes for non-binary respondents as non-binary genders were not reliably captured in previous waves of the NCAS. However, we updated the gender item in 2021 to capture non-binary genders more accurately and are able to provide the mean AVAWS score for non-binary respondents in 2021.

For all respondents, and for men and women separately, the mean AVAWS score was significantly higher in 2021 compared with both 2009 and 2013. However, there were no significant differences in mean AVAWS scores between 2017 and 2021, indicating no further significant improvement in overall community attitudes towards violence against women since 2017. As detailed in Section 3.2, the plateau in rejection of violence against women since 2017 reflects a plateau in rejection of domestic violence, as rejection of sexual violence significantly improved between 2017 and 2021. ^[75]

Examining only 2021 AVAWS scores, there were significant gender differences. Specifically, both non-binary respondents and women showed significantly higher rejection of violence against women in 2021 compared to men. ^[76]

Figure 6-1: Attitudinal rejection of violence against women over time (AVAWS scores) by gender, 2009 to 2021

Note: “na” below means reliable data was not available.  Ns in 2009, 2013, 2017 and 2021 were:

women – 2,200; 2,997; 9,276; 10,121

men – 1,543; 2,481; 8,224; 8,858

non-binary respondents – na; na; na; 81

all – 3,743; 5,478; 17,538; 19,097.

Demographic items for gender were updated for the 2021 NCAS, in accordance with the ABS Standard (ABS, 2021h). As the gender item in previous survey waves did not include the same response options for non-binary respondents, only results for men and women can be compared over time.

* Statistically significant difference on this scale between the year indicated and 2021.

*1 Statistically significant difference compared to men in 2021.

6.2 Attitudes towards violence against women: AVAWS subscales

Figure 6-2 displays changes over time for the three AVAWS subscales. ^[77]  There were improvements over time for all three subscales. The mean scores for the Mistrust Women Subscale and Objectify Women Subscale were significantly higher in 2021 compared to 2017, indicating stronger rejection of these attitudes. While the Minimise Violence Subscale was significantly higher in 2021 than in 2009 and 2013, improvement on this subscale slowed, with no significant difference between 2017 and 2021.

The mean scores on the different AVAWS subscales in 2021 were also compared to one another to examine whether some types of problematic attitudes towards violence against women are more likely to be rejected than others (Figure 6-2). Based on all respondents in 2021, there were no significant differences between the mean scores on the three AVAWS subscales. This finding suggests that the Australian population has a similar level of rejection of attitudes that minimise violence against women and shift blame, attitudes that mistrust women’s reports of violence and attitudes that objectify women or disregard consent.

Figure 6-2: Rejection of different aspects of violence against women (AVAWS subscales) over time, 2009 to 2021

Note: “na” means reliable data was not available.  Ns in 2009, 2013, 2017 and 2021 were:

Minimise below Violence Subscale – 5,045; 16,852; 17,538; 19,093

Mistrust Women Subscale – na; na; 17,269; 18,968

Objectify Women Subscale – na; na; 17,480; 18,780.

There was no significant difference between AVAWS subscales in 2021.

* Statistically significant difference on this subscale between the year indicated and 2021.

The mean scores on each AVAWS subscale in 2021 were also compared by gender (Figure 6-3). There were significant differences between genders for each AVAWS subscale in 2021. Specifically:

• Women demonstrated significantly higher rejection of violence against women on two of the three AVAWS subscales, showing higher rejection on the Minimise Violence Subscale and Mistrust Women Subscale compared to men.
• Non-binary respondents demonstrated significantly higher rejection of violence against women on two of the three AVAWS subscales, showing higher rejection on:
• • the Mistrust Women Subscale compared to men
  • the Objectify Women Subscale compared to both men and women.

Figure 6-3: Rejection of different aspects of violence against women (AVAWS subscales) by gender, 2021

Note:  N = 19,100.

* Statistically significant difference compared to men on this subscale in 2021.

*1 Statistically significant difference compared to men and women on this subscale in 2021.

The three ”AVAWS in focus” sections below present the item-level results for each AVAWS subscale in turn. In these sections, the item-level results are discussed in the context of the existing literature on the concepts underlying the subscales of mistrusting women, minimising violence against women and objectifying women. These latent constructs measured by the subscales were identified based on factor analysis of the AVAWS items, which examine attitudes to  multiple types of violence . Each AVAWS subscale similarly relates to attitudes about multiple types of violence. Chapter 7 discusses the results for each type of violence scale separately and provides further conceptual insights on  attitudes related to each type of violence. Chapter 7 notes where these additional conceptual insights about specific types of violence link to the concepts underlying the AVAWS subscales of mistrusting women, minimising violence against women and objectifying women.

AVAWS in focus: Minimise Violence Subscale

The Minimise Violence Subscale of the AVAWS comprises 15 items examining the attitudinal concept of  minimising violence against women and  shifting blame from the perpetrator to the victim or survivor. This subscale consists almost entirely of items about domestic violence (12 items), but also includes two items about sexual violence and one item about violence against women more generally.

The attitudinal concept underlying the Minimise Violence Subscale suggests that violence against women is not serious and that the women who experience this violence may be responsible for causing or triggering the violence. Attitudes minimising violence can involve excusing the perpetrator, positioning women as responsible for the violence occurring and continuing, and dismissing or underplaying the adverse impacts or harms of violence. Sometimes victims’ and survivors’ experiences are minimised based on the type of abuse to which they are subjected. Physical and sexual violence continue to be perceived as the most extreme and serious forms of violence, while the impacts of other forms of domestic abuse are downplayed in comparison (Mwatsiya & Rasool, 2021).

At their core, minimising and blame-shifting attitudes are characterised by over-identification with the abuser’s perspective (Bongiorno et al., 2020; Gilmore, 2019). These attitudes can involve failing to recognise the serious impacts of violence, as well as perceiving violence as unavoidable or acceptable, given particular circumstances or perpetrator vulnerabilities. For example, the perpetrator’s alcohol or drug use, mental health issues or experience of life stressors such as unemployment are sometimes used to excuse violence (Keilholtz et al., 2022; Mwatsiya & Rasool, 2021; Pugh et al., 2021). Similarly, minimising attitudes can result in apportioning blame to the victim or survivor by suggesting, for example, that she “triggered” the violence by her clothing choices, general antagonism, infidelity or consumption of alcohol, or by “leading” the man on (Hockett et al., 2016; Minter et al., 2021; Persson & Dhingra, 2022; Suarez & Gadalla, 2010). These excuses thereby shift responsibility away from perpetrators and ultimately reinforce a culture that downplays the seriousness, prevalence and impacts of violence against women (Bongiorno et al., 2020; Mwatsiya & Rasool, 2021).

Minimising and blame-shifting attitudes have real-world implications. Research has shown how the media continues to frame domestic violence, sexual assault and sexual harassment through a lens that implies mutual responsibility for violence or that women “drive” men to behave violently (Easteal, Holland, et al., 2018; Easteal et al., 2015; Sutherland et al., 2016; Sutherland et al., 2019). Minimising and blame-shifting attitudes have been evidenced in both formal and informal support systems (J. M. Gray & Horvath, 2018; Hine & Murphy, 2017; A. Murphy & Hine, 2019; Mwatsiya & Rasool, 2021; Temkin et al., 2018). As discussed further in Section 7.2, downplaying the seriousness of violence and abuse, and shifting focus away from the perpetrator’s responsibility, creates a context where it is difficult for victims and survivors to leave abusive relationships (Ahmad et al., 2009; Bongiorno et al., 2020; Capezza & Arriaga, 2008; Heron et al., 2022).

Figure 6-4 presents the findings for the 2021 NCAS items in the Minimise Violence Subscale. The vast majority of respondents disagreed, either strongly or somewhat, with each item in this subscale (74–97%). These results indicate that Australians generally reject attitudes that minimise violence and shift blame from perpetrators to victims and survivors. Nonetheless, the results suggest that further positive shifts could be made in some of these attitudes that minimise violence, particularly attitudes that position violence as simply a reaction to day-to-day stress (D17) and attitudes that women are responsible for their own victimisation because they make their partner angry (D25). Around one in five respondents agreed (strongly or somewhat) with these statements (23% and 19%, respectively).

Figure 6-4: Minimising violence against women and shifting blame (AVAWS subscale items), 2021

Note:  N = 19,100 unless otherwise noted. Percentages in the figure do not always add to 100 or exactly correspond to percentages in the text due to rounding.

ns No significant difference between 2017 and 2021.

a New item in 2021. Thus, change over time could not be examined.

^ Asked of half of the sample.

Table 6-1 shows the level of disagreement with the items in the Minimise Violence Subscale over time. Consistent with the lack of significant improvement at the subscale level between 2017 and 2021 (Figure 6-2), there was also no significant improvement in any of the subscale items (Table 6-1). It is worth noting that this lack of improvement since 2017 may partly reflect the reasonably high level of rejection of minimising attitudes in 2017. At least 9 in 10 respondents disagreed with the minimising attitudes measured by six of the 10 subscale items that were present in the 2017 NCAS (Table 6-1). ^[78]

However, in keeping with the significantly higher Minimise Violence Subscale mean score in 2021 compared to 2009 and 2013 (Figure 6-2), four items improved significantly in 2021 compared to either 2013 or 2009 or both. Specifically, in 2021 compared to 2009 or 2013, a higher proportion of respondents strongly or somewhat disagreed that it is a woman’s duty to stay in a violent relationship to keep the family together (D24), women who are sexually harassed should deal with it themselves rather than report it (S9), and domestic violence can be excused on the basis of outbursts of anger (D18) or because the person regrets their actions afterwards (D19).

Table 6-1: Minimising violence against women and shifting blame (AVAWS subscale items) over time, 2009 to 2021

% net disagree ^a

Item	Code	2009	2013	2017	2021
Women should keep quiet about domestic violence to protect their family’s reputation	D30	–	–	–	97
It’s a woman’s duty to stay in a violent relationship to keep the family together	D24	91^	89*	96	95^
A man is less responsible for rape if he is drunk or affected by drugs at the time	S19	90^	90	90~	93^
It’s only really stalking if it’s by a stranger	V8	–	–	–	95
Domestic violence can be excused if the offender is heavily affected by alcohol	D20	91^	90	94	94
Women who are sexually harassed should deal with it themselves rather than report it	S9	84^*	85*	91~	93^
Women who stay in abusive relationships deserve less help from counselling and support services than women who leave their abusive partner	D31	–	–	87	91^
Domestic violence can be excused if the victim is heavily affected by alcohol	D21	90^	88	93	93^
It’s acceptable for police to give lower priority to domestic violence cases they’ve attended many times before	D32	–	–	86	89^
Domestic violence can be excused if the violent person was themselves abused as a child	D22	–	85	90	90
Domestic violence can be excused if it results from people getting so angry that they temporarily lose control	D18	79^	76*	87	84
Domestic violence can be excused if, afterwards, the violent person genuinely regrets what they have done	D19	71^*	74*	84	85
Sometimes a woman can make a man so angry that he hits her when he didn’t mean to	D25	–	–	75	78^
Domestic violence is a private matter that should be handled in the family	D16	83^	80^	85	87^
A lot of what is called domestic violence is really just a normal reaction to day-to-day stress and frustration	D17	–	–	76	74

Note:  Ns in 2009, 2013, 2017 and 2021 were 10,105; 17,517; 17,542; 19,100.

a Percentage of respondents who strongly or somewhat disagreed with the item.

* Statistically significant difference compared to 2021.

~ Asked of one of the sample in this year.

^ Asked of half the sample in this year.

AVAWS in focus: Mistrust Women Subscale

The Mistrust Women Subscale of the AVAWS comprises 13 items focusing on the attitudinal concept of mistrusting women’s reports of violence victimisation. This subscale comprises eight items about sexual violence, four about domestic violence and one about violence against women more generally. Further conceptual insights regarding attitudes related to specific types of violence against women are presented in Chapter 7.

The concept of mistrust involves attitudes that women’s reports of violence victimisation are suspicious,
exaggerated or false. Recent ANROWS research investigating community perceptions of women’s reports of sexual assault found that mistrusting these reports was the default position for almost all participants (Minter et al., 2021). Both the ANROWS study and a recent English investigation similarly revealed how people perceive disclosures of violence through this default lens of mistrust and draw on rape myths and gendered stereotypes to “fill in” the gaps and make sense of reports of victimisation (Minter et al., 2021; Mulder & Bosma, 2022). Mistrusting women’s reports of sexual assault victimisation based on myths about “genuine victims” of sexual assault is discussed further in Section 7.3. Hostile gender stereotypes that have been linked to mistrust include that women are “untrustworthy”, deceitful, vindictive, motivated by greed and “willing to use their sexuality to harm men” (Rees & White, 2012, p. 428). Perceived ulterior motives for reporting sexual assault – such as to gain some advantage, as a way of “getting back at men” or to cover for embarrassment or regret – were highly influential on community mistrust in women’s reports of violence (Minter et al., 2021). Perceptions of ulterior motives have also been shown to increase levels of mistrust among police towards women’s reports of sexual assault (Jordan, 2004b; Kelly, 2010; Lievore, 2004; McMillan, 2018; Rumney, 2006; Saunders, 2012; Wall & Tarczon, 2013). Similarly, studies suggest a perceived delay in reporting violence victimisation is also linked to suspicions that the woman disclosing violence may be lying or have an ulterior motive (Ellison & Munro, 2009a, 2009b; Jordan, 2004a, 2004b; McMillan, 2018; Minter et al., 2021; O. Smith & Skinner, 2017; Temkin et al., 2018; von Sikorski & Saumer, 2021).

Mistrusting attitudes have real-world impacts. Many victims and survivors do not disclose their experiences of violence to informal support networks or report to police based on fears that they will not be taken seriously or will not be believed (K. J. Holland et al., 2021; H. Johnson, 2017; Lorenz et al., 2019; MacLeod, 2016; O’Donohue, 2019; Reich et al., 2021; Wamboldt et al., 2019; Weiss, 2013; Whiting et al., 2020; Wilson et al., 2021). This fear of not being believed can also prevent victims and survivors from leaving abusive relationships (E. A. Bates, 2020). These fears of not being believed are arguably warranted: extensive Australian and international literature has demonstrated how beliefs that women lie about or exaggerate claims of violence remain pervasive in the community, the justice system and the media (Beshers & DiVita, 2019; Dellinger Page, 2010; Dinos, 2014; Epstein & Goodman, 2018; Fakunmoju, 2022; Gilmore, 2019; Gunby et al., 2013; Harmer & Lewis, 2022; McMillan, 2018; Minter et al., 2021; Navarro & Tewksbury, 2017; O’Brien, 2016; O’Neal & Hayes, 2020; Webster et al., 2018b). For example, in the United States, a recent study of Californian police officers revealed that nearly three quarters (73%) of participants claimed that teenagers lie about sexual assault (O’Neal & Hayes, 2020). Relatedly, studies in Australia and overseas indicate that police often vastly overestimate the prevalence of false allegations of sexual assault, and many assume that most women who do report being assaulted are lying, contrary to the the evidence indicating that false allegations are rare (Daly & Bouhours, 2010; Dewald & Lorenz, 2021; C. E. Ferguson & Malouff, 2016; Jordan, 2004b; Kelly, 2010; McMillan, 2018; O’Neal & Hayes, 2020; Venema, 2014; Waterhouse et al., 2016). Mistrust can lead to further traumatisation when women’s reports of violence are not taken seriously or not believed, or when women are mis-identified as the perpetrators of violence (C. E. Ferguson & Malouff, 2016; Heenan & Murray, 2006; Laing, 2016; Nancarrow et al., 2020; Ullman, 2021; Untied et al., 2018).

Figure 6-5 shows the level of agreement or disagreement with the 13 items in the Mistrust Women Subscale in 2021. The subscale items present attitudes that mistrust women’s reports of violence and respondents were asked if they agreed or disagreed with these mistrusting attitudes. Respondents predominantly rejected attitudes that mistrust women’s reports of violence. With the exception of one item (D23), the majority of respondents disagreed, either strongly or somewhat, with the Mistrust Women Subscale items (57–93%). Levels of disagreement were highest for attitudes that women’s claims of violence should not be taken seriously (S2, S22, D27) and attitudes that women who delay reporting are lying (S10, S25). Nonetheless, the proportions of respondents who agreed with most Mistrust Women Subscale items was concerning, indicating considerable mistrust of women’s reports of violence among the Australian population. For example, about one quarter to one third (23–37%) of respondents strongly or somewhat agreed that women lie about domestic violence to gain an advantage in a custody battle (D23); women lie about sexual assault as a way of “getting back at men” (S23) or due to regretting consensual sex (S24); and women exaggerate the extent of men’s violence (V3). Similarly, 14 per cent of respondents agreed that many sexual assault allegations are false (S18), contrary to the Australian and international evidence that false allegations of sexual assault are rare (Heenan & Murray, 2006; Kelly, 2010; Lisak et al., 2010; Spohn et al., 2014; Wall & Tarczon, 2013; Weiser, 2017). These findings highlight that much work is still needed to challenge these deep-seated mistrusting attitudes, particularly hostile attitudes that women have malicious agendas and ulterior motives when disclosing their stories of violence victimisation.

Attitudes reflecting mistrust of women and hostile sexism also intersect with other structural inequalities and discriminatory attitudes. Racism results in white women being constructed as the “ideal”, and thus the most believable, victim and survivor compared with Aboriginal and/or Torres Strait Islander or Black women, for example (Cripps, 2021; Hamad, 2019; Slakoff & Brennan, 2020). Similarly, ableist attitudes have been shown to inform perceptions among police that women with intellectual, mental health or psychosocial disabilities are “less credible” (Antaki et al., 2015; Benedet & Grant, 2007; Ellison et al., 2015; Heenan & Murray, 2006). New items in the 2021 NCAS gauged the extent to which Australians reject attitudes that mistrust reports of sexual violence victimisation made by women with mental health issues (S1) and lesbian and bisexual women (S2). Importantly, the vast majority of respondents strongly or somewhat disagreed with such mistrusting attitudes (93% for S2 and 86% for S1). These results align with other recent American research on attributions of blame towards bisexual and lesbian sexual assault victims and survivors (K. E. Morrison & Pedersen, 2020). Nonetheless, more work is needed to address suspicions held by 1 in 16 Australians (6%) that women with mental health concerns may be lying when they say they have been sexually assaulted (S1).

Figure 6-5: Mistrusting women’s reports of violence (AVAWS subscale items), 2021

Note:  N = 19,100 unless otherwise noted. Percentages in the figure do not always add to 100 or exactly correspond to percentages in the text due to rounding. Significant differences over time are based on the percentage of respondents who answered “strongly disagree” or “somewhat disagree”.

ns No significant difference between 2017 and 2021.

a New item in 2021. Thus, change over time could not be examined.

* Significantly higher understanding in 2021 than 2017.

~ Asked of one quarter of the sample.

^ Asked of half of the sample.

Table 6-2 shows the level of disagreement with the Mistrust Women Subscale items over time. Consistent with the significant increase in rejection of attitudes that mistrust women at the subscale level between 2017 and 2021 (Figure 6-2), two items showed significant improvement over time. Specifically, a significantly higher proportion of respondents in 2021 strongly or somewhat disagreed that women falsify or exaggerate domestic violence to improve their custody claims (D23) compared to previous years (47% versus 25–36%). Similarly, a significantly higher proportion of respondents strongly or somewhat disagreed that women lie to cover for regretful sex (S24) in 2021 than in 2013 and 2017 (66% versus 47–55%). It is notable that these two items showing improvement over time were among those evidencing the highest levels of mistrust historically and still in 2021. Also noteworthy is that there was no significant improvement over time in the rejection of the other 11 Mistrust Women Subscale items, despite a trend in this direction in the raw percentages. Thus, the findings in Table 6-2 and Figure 6-5 indicate that further work is needed to address community mistrust of women’s reports of violence, despite a few promising shifts over time in these mistrusting attitudes.

Table 6-2: Mistrusting women’s reports of violence (AVAWS subscale items) over time, 2009 to 2021

% net disagree ^a

Item	Code	2009	2013	2017	2021
When lesbian or bisexual women claim to have been sexually assaulted by their partner, they probably shouldn’t be taken too seriously	S2	–	–	–	93
If a woman claims to have been sexually assaulted but has no other physical injuries, she probably shouldn’t be taken too seriously	S22	–	–	91	93
Women who wait weeks or months to report sexual harassment are probably lying	S10	–	–	87~	90
Women who wait weeks or months to report sexual assault are probably lying	S25	–	–	85	90
If a woman keeps going back to her abusive partner, then the violence can’t be very serious	D27	–	–	84	88^
It’s easy for a woman to leave an abusive relationship	D28	–	–	–	88^
Women with mental health issues who report being sexually assaulted are probably lying	S1	–	–	–	86
A female victim who does not leave an abusive partner is partly responsible for the abuse continuing	D29	–	–	65	72
Many allegations of sexual assault made by women are false	S18	–	–	72~	78~
A lot of times, women who say they were raped had led the man on and then had regrets	S24	–	47^*	55*	66
Many women exaggerate the extent of men’s violence against women	V3	–	–	65	67^
It is common for sexual assault accusations to be used as a way of getting back at men	S23	–	–	47	57
Women going through custody battles often make up or exaggerate claims of domestic violence in order to improve their case	D23	26^*	25^*	36~*	47^

Note:  Ns in 2009, 2013, 2017 and 2021 were 10,105; 17,517; 17,542; 19,100.

a Percentage of respondents who strongly or somewhat disagreed with the item.

* Statistically significant difference compared to 2021.

~ Asked of one quarter of the sample in this year.

^ Asked of half the sample in this year.

AVAWS in focus: Objectify Women Subscale

The Objectify Women Subscale of the AVAWS comprises 15 items, including 11 standalone items (Figure 6-6) and four items concerning two scenarios about sexual consent (Figure 6-7 and Figure 6-8). All the items and scenarios in this subscale examine sexual violence, except for one item about domestic violence. The discussion below focuses on the attitudinal concept of  objectifying women and disregarding sexual consent, which underlies the subscale based on psychometric analysis. Further theoretical insights regarding attitudes related to sexual harassment and sexual assault are presented in Section 7.3.

Sexual objectification is a type of gender-based discrimination where women’s experiences – from everyday street harassment to sexual assault – can result in feeling simply like a “body that exists for the use and pleasure of others” (Wesselmann et al., 2021, p. 841; see also Miles-McLean et al., 2014). Sexual objectification reduces people to sexual objects by prioritising or separating a person into their sexual features and disregarding their emotional, social or intellectual worth, and their individual agency. As a result, the person is dehumanised and cast as an “object” for others’ use or abuse (Bernard et al., 2018; Bernstein et al., 2022a; Bevens & Loughnan, 2019; Hollett et al., 2022; Loughnan et al., 2013; K. L. Morris et al., 2018; Sáez et al., 2022). The sexual objectification of women is heteronormative, being both  gendered and  heterosexual, as these attitudes and social practices reduce women “to their physical attributes and heterosexual attractiveness” as objects for men’s sexual gratification (Paasonen et al., 2020, p. 7). Objectifying attitudes, for example, underlie claims that women behave in such a way that invites men’s actions, or which imply women should always be readily available for men’s sexual pleasure. Importantly, the intersections of race, class and gender minority experiences also shape particular forms of sexual objectification and the narratives of blame associated with it (J. R. Anderson et al., 2018; Kiebler & Stewart, 2022; Ussher et al., 2022).

Sexually objectifying attitudes are reproduced and reinforced through many social institutions. For example, research regularly suggests that media, including television shows, newspaper texts, internet pornography, video games and social media, are key sites of sexual objectification in society (Bernstein et al., 2022a; Karsay et al., 2018; Skowronski et al., 2020). Analyses of such media content highlights that women are often defined by their bodies, posed in submissive or exploitative postures, and even depicted as deserving or enjoying their own abuse (Dines, 2010; Galdi & Guizzo, 2021; Paasonen et al., 2020). Other sites of sexual objectification include, for example, requirements that women wear sexualised attire in certain jobs (such as in some hospitality jobs), which may result in women becoming targets for sexual harassment by clientele who interpret such attire as permission to harass women (Easteal, O’Neill, et al., 2018; Klein et al., 2021).

Adherence to these sexually objectifying attitudes can have dangerous consequences. Research has evidenced links between sexually objectifying attitudes and sexual or physical aggression, as well as perpetration of emotional and physical intimate partner violence (Bevens & Loughnan, 2019; Blake et al., 2018; Sáez et al., 2022; Vasquez et al., 2018). Exposure to sexualised media has been associated with the development of sexist attitudes, dehumanisation and even sexual violence perpetration (Awasthi, 2017; Maes et al., 2019). Similarly, a sense of entitlement to sex has been identified as a factor in intimate partner sexual violence (Tarzia, 2020). Additionally, a systematic qualitative review found that sexual violence perpetration was significantly associated with the acceptance of violent sexual behaviours, experiences and attitudes (Tharp et al., 2013). More recent research has similarly associated sexually objectifying attitudes with the acceptance of rape myths, hostile sexism and other negative attitudes towards women (Harsey & Zurbriggen, 2021; Methot-Jones et al., 2019; Poerwandari et al., 2021; Samji & Vasquez, 2020). Thus, attitudes play a role in promoting a culture of sexual violence by normalising the sexual objectification of women as objects for men’s sexual attention and gratification.

Sexual objectification also impacts the way people see and treat themselves and others (Bernard et al., 2020). Research has associated the consumption of sexualised media with a greater use of alcohol to feel sexual and the lower likelihood of condom use, as well as negative sexual effects and negative impacts on wellbeing (L. M. Ward et al., 2018). Sexual objectification has also been associated with body image concerns, while feeling sexually objectified by an intimate partner has been linked with lower relationship satisfaction (Sáez et al., 2019; Skowronski et al., 2020). Additionally, research suggests that objectification is linked to victim-blaming attitudes, where objectified women are perceived to be more responsible for being sexually assaulted, more responsible for forms of image-based abuse (colloquially referred to as “revenge porn”) and less worthy of help or support from others (E. Holland & Haslam, 2016; Loughnan et al., 2013; Serpe & Brown, 2022; Spaccatini et al., 2022).

Attitudes that promote disregard for consent reinforce heterosexual scripts that promote women as sexual gatekeepers, and men as the aggressive pursuers of sex (Benoit & Ronis, 2022; Jozkowski et al., 2018). These heterosexual scripts normalise and legitimise men applying pressure and coercion for sex (Bernstein et al., 2022b; Fahs & Gonzalez, 2014; E. M. Morgan & Zurbriggen, 2016). This perspective also focuses attention on whether women adequately resist rather than on whether perpetrators lawfully seek and receive consent (Brady et al., 2018; Minter et al., 2021; O’Byrne et al., 2006; O’Byrne et al., 2008). Positioning women as “sexual gatekeepers” can result in blaming victims if they are unable to consent or resist; for example, due to intoxication, or if they “give in” to the repeated pressure for sex (Hills et al., 2020; Sims et al., 2007). When women do refuse consent, it can be disregarded with harmful suggestions that women play hard to get, say “no” when they mean “yes” or have already provided ongoing consent by showing interest or pleasure at some earlier point (Hills et al., 2020; Jozkowski et al., 2018). Despite recent updates to sexual consent laws in many Australian states and territories requiring affirmative consent, the Australian media continues to endorse these problematic gender roles and sexual scripts and fails to acknowledge sexual pressure and coercion as harmful (Hindes & Fileborn, 2019).

Figure 6-6 shows that the majority of respondents either strongly disagreed (56–86%) or somewhat disagreed (6–25%) with each of the 11 standalone items in the Objectify Women Subscale. These results indicate that most Australians reject attitudes that reduce women to sexual objects or show an indifference to gaining active consent. In particular, the highest level of rejection was for the items relating to rape or forced sexual touching, with around 9 in 10 respondents strongly or somewhat disagreeing with these attitudes (S21, S17, S7, S20, S4).

Nonetheless, concerning proportions of respondents  agreed with several Objectify Women Subscale items. For example, 25 per cent of respondents strongly or somewhat agreed that a man may not realise a woman does not want to have sex if he is very sexually aroused (S8). This result reveals concerning community support for gender role attitudes relating to men’s entitlement to sex, sexual dominance, and insatiable and uncontrollable sex drives – attitudes that ultimately minimise men’s responsibility for sexual violence perpetration and their failure to respect women’s consent and bodily autonomy (Jeffrey & Barata, 2017, 2019, 2020; Ray & Parkhill, 2021; R. M. Smith et al., 2015). Additionally, 21 per cent of respondents agreed (strongly or somewhat) that a woman who sends her partner a naked picture of herself is partly responsible if the partner then shares the image without her consent (S6), while around 1 in 8 respondents agreed that women should be flattered by receiving catcalls in public (S3) or being repeatedly pursued by someone they are not interested in (S11).

Figure 6-6: Objectifying women and disregarding consent (AVAWS subscale items), 2021

Note:  N = 19,100 unless otherwise noted. Percentages in the figure do not always add to 100 or exactly correspond to percentages in the text due to rounding. Significant differences over time are based on the percentage of respondents who answered “strongly disagree” or “somewhat disagree”.

ns No significant difference between 2017 and 2021.

a New item in 2021. Thus, change over time could not be examined.

* Significantly higher understanding in 2021 than 2017.

~ Asked of one quarter of the sample.

^ Asked of half of the sample.

In addition to the standalone items in Figure 6-6, the Objectify Women Subscale also included two scenarios about sexual consent, one about a married couple and the other about a couple who had just met at a party. Both scenarios asked respondents whether sexual assault was justified if 1) the man had initiated kissing before the woman pushed him away and 2) the woman had initiated kissing before pushing him away. Figure 6-7 shows the results for the married couple scenario, while Figure 6-8 shows the results for the acquaintance scenario.

The overwhelming majority of respondents strongly or somewhat disagreed that the man in each scenario was justified in forcing sex when he had initiated intimacy (94% for the married couple scenario; 96% for the acquaintance scenario).

However, for both scenarios, fewer respondents strongly or somewhat disagreed that forced sex was justified when the woman had initiated intimacy (83% for the married couple scenario; 88% for the acquaintance scenario). Additionally, around 1 in 10 respondents strongly or somewhat  agreed that the man would be justified in forcing sex if the woman had initiated intimacy (11% for the married couple scenario; 8% for the acquaintance scenario).

Figure 6-7: Sexual consent scenario (AVAWS Objectify Women Subscale items), married couple variation, 2021

Note:  N = 4,640. Percentages in the figure do not always add to 100 or exactly correspond to percentages in the text due to rounding. Asked of one quarter of the sample in 2021.

ns No significant difference between 2017 and 2021.

Figure 6-8: Sexual consent scenario (AVAWS Objectify Women Subscale items), acquaintance variation, 2021

Note:  N = 4,661. Asked of one quarter of the sample in 2021.

ns No significant difference between 2017 and 2021.

As noted earlier, disagreement with the Objectify Women Subscale improved significantly between 2017 and 2021 (Figure 6-2). Table 6-3 shows the level of disagreement with the items in the Objectify Women Subscale over time, including both the standalone items and the two scenarios about sexual consent. Consistent with the improvement at the subscale level, most items showed a raw increase in rejection between 2017 and 2021 and this difference was significant for three of the subscale items. Specifically, in 2021 compared to 2017, higher proportions of respondents strongly or somewhat disagreed that women find it flattering to be persistently pursued (S11), that it is understandable if men think they can touch women without consent because “women are so sexual in public” (S7), and that a woman who sends her partner a naked picture of herself is partly responsible if he shares it without her permission (S6). Additionally, two of the standalone items that didn’t show an improvement since 2017 showed an improvement since earlier years. Specifically, significantly higher proportions of respondents in 2021 compared with 2009 and 2013 disagreed (strongly or somewhat) that women often say “no” when they mean “yes” (S5) and that sexual assault victims and survivors are to blame for their assault if they are intoxicated (S20).

However, there was no significant improvement in the two scenarios about sexual consent. Specifically, the high level of rejection in 2021 of forced sex when intimacy is initiated by the man was identical to that in 2017. Further, although there was a raw tendency towards an improvement between 2017 and 2021 in the proportion of respondents disagreeing that forced sex is justified when the woman initiates intimacy, this finding did not reach statistical significance for either scenario (80–83% for the married couple scenario; 83–88% for the acquaintance scenario; Table 6-3).

While the significant improvements over time for some of the Objectify Women Subscale items are promising, more work remains to challenge attitudes that normalise the sexual objectification of women and that normalise disregarding women’s consent.

Table 6-3: Objectifying women and disregarding consent (AVAWS subscale items) over time, 2009 to 2021

Standalone items (% net disagree ^a)

Item	Code	2009	2013	2017	2021
If a woman meets up with a man she met on a mobile dating app, she’s partly responsible if he forces sex on her	S21	–	–	–	92
If a woman doesn’t physically resist – even if protesting verbally – then it isn’t really rape	S17	–	85^	88~	91~
Since some women are so sexual in public, it’s understandable that some men think they can touch women without permission	S7	–	–	76*	89
If a woman is raped while drunk or affected by drugs, she is at least partly responsible	S20	80^*	78^*	85	88
If a woman is drunk and starts having sex with a man, but then falls asleep, it is understandable if he continues having sex with her anyway	S4	–	–	82	89
Women often say “no” when they mean “yes”	S5	78^*	74^*	82	86
Women who flirt all the time are somewhat to blame if their partner gets jealous and hits them	D26	–	–	83	87
If a woman sends a naked picture to her partner, then she is partly responsible if he shares it without her permission	S6	–	–	67~*	77
When a man is very sexually aroused, he may not even realise that the woman doesn’t want to have sex	S8	–	–	66	69
Women find it flattering to be persistently pursued, even if they are not interested	S11	–	–	68~*	81^
A woman should be flattered if she gets wolf-whistles or catcalls when walking past a group of men in public	S3	–	–	–	82^

Scenario items (% net disagree ^a)

Item	Code	2009	2013	2017	2021
Married: Man initiates After coming home from a party, a man kisses his wife and tries to have sex with her. She pushes him away, but he has sex with her anyway. Do you agree or disagree that the man is justified in his behaviour?	S12	–	–	94^	94~
Married: Woman initiates What if she had taken him into the bedroom and started kissing him before pushing him away. Do you agree or disagree that the man would have been justified in having sex with her anyway?	S13	–	–	80^	83~
Just met: Man initiates A man and woman have just met at a party and get on well. They go back to the woman’s home where he kisses her and tries to have sex with her. She pushes him away, but he has sex with her anyway. Do you agree or disagree that the man is justified in his behaviour?	S14	–	–	96^	96~
Just met: Woman initiates What if she had taken him into the bedroom and started kissing him before pushing him away. Do you agree or disagree that the man would have been justified in having sex with her anyway?	S15	–	–	83^	88~

Note:  Ns in 2009, 2013, 2017 and 2021 were 10,105; 17,517; 17,542; 19,100.

a Percentage of respondents who strongly or somewhat disagreed with the item.

* Statistically significant difference compared to 2021.

~ Asked of one quarter of the sample in this year.

^ Asked of half the sample in this year.

6.3 Attitudes towards violence against women: Assessing the importance of demographics, understanding and attitudes

Contribution of demographics, understanding and gender inequality attitudes to attitudes rejecting violence against women

Efforts to prevent violence against women are aided by understanding the factors that are associated with, or contribute to, an individual’s attitudes towards such violence. Four multiple regression models were conducted to examine how well we can predict respondents’ attitudes towards violence against women (the outcome variable) if we know their demographic characteristics, their understanding of the nature of this violence and their attitudes towards gender inequality (the input variables; see Methodology reminder 6-3 and Section 3.5).

When demographic factors were considered on their own (AVAWS Model 1), they explained one fifth (20%) of the variance in AVAWS scores (Figure 6-9). Thus, while demographic characteristics help us to predict attitudes towards violence against women, much of the difference in these attitudes (80%) cannot be explained by demographic characteristics alone.

AVAWS Model 2 examined how well attitudes towards violence against women (as measured by the AVAWS) could be predicted by only examining understanding of violence against women and attitudes towards gender inequality (as measured by the UVAWS and AGIS) as possible predictors. This model found that the UVAWS and AGIS were significant predictors and explained one half (50%) of the variance in AVAWS scores. The AGIS explained 37 per cent of the variance in the AVAWS scores, while the UVAWS explained 13 per cent. This finding is consistent with other literature and research that highlights the important relationship between attitudes towards gender inequality and violence against women (COAG, 2010b, 2022; Webster et al., 2018a). Thus, improving the community’s attitudes towards gender inequality, as well as understanding of violence against women, may improve the rejection of violence against women. However, half of the difference in respondents’ attitudes towards violence against women (50%) could not be explained by these factors, suggesting that other factors are also important in predicting and shaping attitudes towards violence against women (Figure 6-9).

Another regression (AVAWS Model 4) examined which UVAWS and AGIS subscales were most responsible for the scale-level relationships between the UVAWS, AGIS and AVAWS. The AGIS Deny Inequality Subscale was the subscale that made the largest contribution to AVAWS scores and was a significant predictor of AVAWS scores. ^[79]  This result suggests that shifting attitudes that deny gender inequality experiences may be an important component of initiatives that aim to improve rejection of violence against women by increasing rejection of gender inequality.

Considering demographic factors, UVAWS scores and AGIS scores together (AVAWS Model 3) improved the ability to predict AVAWS scores, with more than half (54%) of the variance in AVAWS scores being explained (Figure 6-9). ^[80]  However, some of the difference in respondents’ attitudes (46%) still could not be explained by respondents’ understanding of violence, attitudes towards gender inequality and their demographic characteristics. Thus, other factors are also important in predicting and shaping attitudes towards violence against women.

Figure 6-9: Contribution of demographics and scales to attitudinal rejection of violence against women
(AVAWS scores), 2021

Note:

a Based on AVAWS Model 1.  N = 18,876.

b Based on AVAWS Model 2.  N = 18,868.

c Based on AVAWS Model 3.  N = 18,868.

Demographic characteristics related to attitudes towards violence against women

As noted above, the regression results revealed that demographics considered alone explained 20 per cent of the variation in attitudes towards violence against women (Figure 6-9; AVAWS Model 1). Information about differences between demographic groups in attitudes towards violence against women can assist policymakers and practitioners to target attitude change initiatives more effectively according to the needs of different demographic groups. Table 6-4 shows the significant demographic predictors of attitudes towards violence against women based on the regression (AVAWS Model 1). In order of importance (as listed in the table), the significant demographic predictors of attitudes towards violence against women were age, English proficiency, country of birth and length of time in Australia, education, gender, main labour activity, sexuality and socioeconomic status of area. Age, the most important predictor, explained almost 4 per cent of the variance in attitudes towards violence against women (first column in Table 6-4).

Table 6-4 also shows significant differences between demographic groups in attitudes towards violence against women (AVAWS Model 1). For each significant demographic predictor, a selected or “reference” group was compared to each other group. For example, for age, each age group was treated as a “comparison” group that was contrasted against the “reference” group of all ages on average. ^[81]  The table shows whether each comparison group had significantly higher (>), significantly lower (<) or not significantly different (ns) rejection of violence against women compared to the reference group.

Based on the regression, the demographic groups that had  significantly higher rejection of violence against women were: ^[82]

• age: 25- to 34-year-old respondents compared to all ages on average and all ages on average compared to respondents aged 75 or over
• English proficiency: respondents who spoke English at home compared to respondents who spoke a language other than English (LOTE) at home
• country of birth and length of time in Australia: Australian-born respondents compared to respondents born in a non-main English–speaking country (N-MESC) who had lived in Australia for less than 11 years
• formal education: university graduates compared to respondents without university education
• gender: women compared to men ^[83]
• main labour activity: employed respondents compared to unemployed respondents
• sexuality: lesbian; gay; bisexual or pansexual; and asexual, queer or sexuality-diverse respondents compared to heterosexual respondents
• socioeconomic status of area: respondents living in areas with the highest socioeconomic status compared to those living in areas with the lowest socioeconomic status. ^[84]

In addition, for each significant demographic predictor in the regression, Table 6-4 presents bivariate results showing the percentage of each demographic group with “advanced” attitudinal rejection of violence against women. ^[85]  For example, for age, 43 per cent of 25- to 34-year-olds and 10 per cent of respondents aged 75 years or older were in the “advanced” rejection of violence against women category compared to 34 per cent of all ages. Thus, even though some demographic groups have higher rejection of violence against women, further improvement is needed across all demographic groups to achieve a society where all people have “advanced” rejection.

Table 6-4: Significant demographic predictors of rejection of violence against women (AVAWS score), 2021

Demographic factor – Age (in years) (4%)

Demographic group (% unique contribution to UVAWS scores)	Regression results Significantly higher (>) or lower (<) understanding of violence compared to REF a	Bivariate results % of respondents with “advanced” understanding of violence against women b
All ages on average REF		34
16–24	ns
25–34	>	43
35–44	ns
45–54	ns
55–64	ns
65–74	ns
75+	<	10

Demographic factor – Country of birth and length of time in Australia ^d (3%)

Demographic group	Regression results	Bivariate results
Born in Australia REF		38
MESC: 0–5 years	ns
MESC: 6–10 years	ns
MESC: >10 years	ns
N-MESC: 0–5 years	<	13
N-MESC: 6–10 years	<	21
N-MESC: >10 years	ns

Demographic factor – Formal education (3%)

Demographic group	Regression results	Bivariate results
University or higher REF		44
Trade/certificate/diploma	<	33
Secondary or below	<	27

Demographic factor – Gender (2%)

Demographic group	Regression results	Bivariate results
Men REF		27
Women	>	41
Non-binary respondents	ns	54e

Demographic factor – Main labour activity (2%)

Demographic group	Regression results	Bivariate results
Employed REF		39
Unemployed	<	30
Home duties	ns
Student	ns
Retired	ns
Unable to work	ns

Demographic factor – Sexuality (2%)

Demographic group	Regression results	Bivariate results
Heterosexual REF		33
Lesbian	>	69
Gay	>	53
Bisexual or pansexual	>	57
Asexual, queer or diverse sexualities	>	55

Demographic factor – Socioeconomic status of area ^f (1%)

Demographic group	Regression results	Bivariate results
5 – Highest status REF		41
1 – Lowest status	<	25
2 – Second-lowest status	ns
3 – Middle status	ns
4 – Second-highest status	ns

Note:  N = 18,876. Regression results are from AVAWS Model 1. Only significant predictors are shown. The total contribution of the demographic predictors alone to AVAWS scores was 20%. Remoteness of area was retained in the model because it improved model fit, but it was not a significant predictor. Disability was removed from the final model because it did not improve model fit.

REF The reference group for this demographic factor. All other groups for the demographic factor were compared to the REF. The REF was chosen based on considerations of statistical power (i.e. the group with the most respondents) and ease of interpretation (e.g. comparing the group with the highest formal education to each other group).

ns No significant difference between this demographic group and the REF.

a Based on the regression results, this demographic group had significantly higher (>), significantly lower (<) or not significantly different (ns) rejection of violence against women compared with the REF. For example, for age, the table shows that respondents aged 25 to 34 years had  significantly higher  (>) rejection compared to all ages on average (the REF). It can also be stated that all ages on average (the REF) had  significantly lower rejection compared to 25- to 34-year-old respondents, but this direction is not shown in the table.

b “Advanced” rejection of violence against women means strongly disagreeing with at least 75% of attitudes condoning violence against women, and somewhat disagreeing with the remaining AVAWS items. See Section 2.5 for further details.

c “LOTE” refers to language other than English spoken at home. “Good English” refers to good or very good self-reported English proficiency and “poor English” refers to no English or poor self-reported English proficiency.

d “MESC” refers to people born in a main English–speaking overseas country (ABS classification), “N-MESC” refers to people born in a non-main English–speaking country. The number of years refers to the number of years since the respondent moved to Australia.

e Regression results can differ from bivariate results because regressions provide the unique contribution of each predictor variable after accounting for associations between predictor variables. Non-binary respondents had significantly higher mean AVAWS scores than men in bivariate analyses, but not multivariate analyses.

f “Socioeconomic status of area” refers to an ABS measure of socioeconomic conditions in geographic areas in terms of people’s access to material and social resources, and their opportunity to participate in society (SEIFA quintiles).

6.4 Conclusions about attitudes towards violence against women

Shifting individuals’ attitudes that condone violence against women is a key aspect of breaking down the broader societal culture that allows violence against women to perpetuate. Attitudes that tolerate, minimise or condone violence interact with a broad range of other factors, systems and structures at multiple levels within society to facilitate violence against women (Our Watch, 2021a).

Positively, the results indicate that most Australians hold attitudes that reject violence against women and that this attitudinal rejection of violence against women has generally improved over the longer term. However, there are still concerning levels of endorsement of some attitudes that condone violence against women. These problematic attitudes include attitudes that minimise the seriousness of violence against women and shift blame to victims and survivors, attitudes that mistrust women’s reports of victimisation based on hostile gendered stereotypes that women often lie to gain some advantage over men, and attitudes that objectify women and disregard the need to gain consent. Further, progress in shifting attitudes appears to have stalled somewhat in recent years, largely reflecting a plateauing of attitudinal rejection of domestic violence despite an improvement in attitudinal rejection of sexual violence since 2017.

The results also reaffirm the 2017 NCAS finding that high rejection of gender inequality was the strongest predictor of high rejection of violence against women. Thus, problematic attitudes towards violence against women need to be addressed together with problematic attitudes towards gender inequality, particularly attitudes that deny gender inequality experiences. The findings also confirm that better understanding of the nature of violence against women is linked, albeit less strongly, to higher attitudinal rejection of violence against women. While demographic factors also significantly predicted attitudes towards violence, much of the difference in respondents’ attitudes towards violence (80%) could not be explained from their demographic characteristics alone. Thus, there is room for improvement in the rejection of violence against women across demographic groups within the Australian community.

These results have implications for policy and prevention initiatives and indicate that increasing the community’s attitudinal rejection of gender inequality is key to reducing cultures of support for violence against women.
To accelerate progress in Australia’s rejection of violence against women, initiatives should:

• Raise awareness that problematic attitudes towards violence against women normalise and perpetuate this violence.
• Address attitudes that support violence against women simultaneously with attitudes that condone gender inequality, given the continued evidence that these attitudes are closely linked.
• Foster a culture of trust and support in women’s reports of violence victimisation. Correct attitudes mistrusting women’s reports of violence and instead emphasise the barriers and difficulties women face when reporting violence (Minter et al., 2021).
• Promote appropriate reporting of perpetrators and violence against women in the media.
• Challenge attitudes that normalise the sexual objectification of women and shift problematic heterosexual sex scripts that normalise disregarding women’s active and affirmative consent (Minter et al., 2021).
• Affirm the seriousness of violence against women and place responsibility on the perpetrator to avoid minimising and blame-shifting scripts.
• Address legislative, policy and service barriers to reporting of violence and to recovery of victims and survivors. For example, it is important to reform legislation and legal processes to facilitate reporting and access to justice; upskill police, justice officers and support services in best-practice victim-centred, trauma-informed and culturally safe practices; ensure institutions (including schools and universities), industries and businesses have policies that treat violence and abuse seriously; and ensure action to support victims and survivors and prevent further perpetration.
• Increase the level of “advanced” rejection of violence against women by:
• • increasing rejection of gender inequality through gender-transformative interventions, including addressing attitudes that deny gender inequality experiences (Section 5.4)
  • increasing understanding of violence against women by improving recognition of repeated “subtle” or non-physical forms of domestic violence and violence against women more broadly, and distinguishing between healthy and unhealthy relationship conflict (Section 4.4).
• Increase the level of “advanced” attitudinal rejection of violence against women by breaking down barriers and facilitating enablers relevant to specific demographic groups (Chapter 9). ^[86]

7 Findings: Specific types of violence against women

7.1 The AVAWS and type of violence scales

The results for the Attitudes towards Violence against Women Scale (AVAWS; Chapter 6) revealed attitudes that underpin the social context that normalises and reinforces violence against women in general. These attitudes were grouped and discussed according to the constructs underlying the AVAWS subscales, which were empirically derived via factor and Rasch analyses, namely attitudes that minimise violence against women, attitudes that mistrust women’s reports of violence and attitudes that objectify women and disregard the need for consent. However, each AVAWS subscale comprises items describing different types of violence, including domestic violence, sexual violence and technology-facilitated abuse. It is important to acknowledge that these types of violence can often overlap. For example, sexual violence can occur within or outside domestic relationships, and technology-facilitated abuse can include domestic abuse, sexual abuse or abuse that is neither of a domestic nor sexual nature. Despite such overlaps, policymakers and practitioners may nonetheless be interested in the more specific attitudes that may relate to each type or form of violence against women. Thus, the AVAWS items were subdivided according to the type of violence they describe and were used as the basis for creating five type of violence scales:

• the Domestic Violence Scale (DVS; Section 7.2)
• the Sexual Violence Scale (SVS), which is a composite of two scales:
• • the Sexual Assault Scale (SAS; Section 7.3)
  • the Sexual Harassment Scale (SHS; Section 7.3)
• the Technology-Facilitated Abuse Scale (TFAS; Section 7.4).

Apart from the TFAS, all the type of violence scales consist of items drawn entirely from the AVAWS, and thus examine  attitudes towards these types of violence. The TFAS comprises two attitude items from the AVAWS and four understanding items from the Understanding of Violence against Women Scale (UVAWS).

By examining the items in each type of violence scale, the present chapter uses thematic examination to provide conceptual insights about the specific myths and misconceptions that underlie each type of violence. This thematic examination of the attitudes related to each violence type supplements the insights from the analysis of the AVAWS subscales. Thus, the analysis in this chapter may help to further inform prevention initiatives related to specific types of violence.

The 2021 NCAS also included three items on  stalking, one on  technology-facilitated stalking and two on  in-person stalking. The item on technology-facilitated stalking was included in the TFAS. There were insufficient stalking items to form a reliable psychometric scale on stalking. However, the present chapter also examines misconceptions underlying stalking based on the three items included in the 2021 NCAS.

To maximise the utility of this chapter for policymakers and practitioners, each section on each violence type begins with an examination of the changes in understanding or attitudes over time and between genders. Each section then presents item-level analysis which groups thematically linked items together to explore the underlying attitudes, myths and misconceptions associated with each type of violence. Exploring these underlying myths and misconceptions within each distinctive type of violence allows us to deconstruct problematic attitudes that may be particularly relevant to that type of violence. This analysis, coupled with the AVAWS subscale analysis, provides rich and nuanced information about the challenging and toxic attitudes that encourage violence against women and useful insights for informing education and prevention strategies.

7.2 Domestic violence

Domestic violence involves harmful, violent, abusive, coercive or bullying behaviour towards an intimate partner. Research consistently demonstrates that while men are most likely to experience violence perpetrated by a male assailant in a public place, women are most likely to experience violence perpetrated by a male partner in their homes (ABS, 2017). Women are also more likely to be afraid of, be hospitalised by or be killed by an intimate partner (ABS, 2017; Cussen & Bryant, 2015). Domestic violence can occur at home, outside the home and online, and arises across different communities, cultures, socioeconomic groups, age groups and occupations, and among people of any education level (WHO, 2013). Domestic violence can include many different types of abusive and violent behaviours, which can be considered coercive control when used in a pattern over time to create and maintain power and control over someone (Meeting of Attorneys-General, 2022; NSW Department of Communities and Justice, 2022). Examples of domestic violence include the following behaviours enacted against an intimate partner:

• physical violence or abuse (e.g. pushing, kicking, punching, slapping, strangulation)
• use of weapons or objects
• denial of food
• the destruction of property
• non-physical forms of psychological manipulation or emotional abuse, including verbal abuse, social abuse and spiritual abuse
• financial and economic abuse
• stalking
• technology-facilitated abuse, including image-based abuse (NSW Department of Communities and Justice, 2022).

As discussed in Section 1.2, attitude change often occurs slowly, with research indicating this process may be delayed or inhibited by several person- and context-related factors.

Rejection of domestic violence over time and between genders

For all respondents, and for men and women separately, the mean score on the Domestic Violence Scale (DVS) was significantly higher in 2021 compared with both 2009 and 2013 (Figure 7-1). These findings indicate a stronger attitudinal rejection of domestic violence in 2021 compared to 2009 and 2013 for the Australian population overall and for both Australian men and Australian women separately. However, for all respondents, and for men and women separately, there were no significant differences in mean DVS scores between 2017 and 2021. This finding indicates that despite efforts to educate the community and a series of high-profile domestic violence cases between 2017 and 2021 (Section 1.1), community attitudes towards domestic violence have not improved since 2017.

Mean DVS scores in 2021 were also compared by gender (Figure 7-1). There were significant differences between genders, with both women and non-binary respondents having significantly higher rejection of domestic violence in 2021 compared to men. ^[87]

Figure 7-1: Rejection of domestic violence (DVS) over time and by gender, 2009 to 2021

Note: “na” below means reliable data was not available.  Ns in 2009, 2013, 2017 and 2021 were:

women – 2,986; 3,802; 9,276; 10,116

men – 1,984; 3,048; 8,223; 8,854

non-binary respondents – na; na; na; 81

all – 4,970; 6,850; 17,537; 19,088.

Demographic items for gender were updated for the 2021 NCAS in accordance with the ABS Standard (ABS, 2021h). As the gender item in previous survey waves did not include the same response options for non-binary respondents, only results for men and women can be compared over time.

* Statistically significant difference on this scale between the year indicated and 2021.

*1 Statistically significant difference for this gender compared to men in 2021.

Domestic violence: Thematic item examination

Chapter 4 discussed the importance of developing nationally consistent definitions for domestic violence and coercive control and increasing community  understanding of both the range of domestic violence behaviours and the gendered nature of domestic violence. The DVS is comprised entirely of items drawn from the AVAWS measuring  attitudes towards domestic violence. Chapter 6 discussed how community attitudes endorsing violence against women are underpinned by empirically confirmed constructs that correspond to the AVAWS subscales, namely minimising violence and shifting blame, mistrusting women, and objectifying women and disregarding consent. Many of these problematic attitudes are evident for domestic violence more specifically. To further examine the community attitudes related to domestic violence, we examine the results for thematically grouped items from the DVS. This thematic examination is used to highlight prevailing myths or misconceptions regarding domestic violence, thereby providing guidance to policymakers on both barriers to change and opportunities for intervention.

Domestic violence myths and misconceptions: “It’s a family matter”

The AVAWS Minimise Violence Subscale (Section 6.2) examined the way dismissing the impact of violence against women, and displacing blame onto victims and survivors, affords perpetrators dangerous justifications that allow violence to continue unchecked (Bongiorno et al., 2020). This minimisation is achieved by downgrading the adverse impacts of violence against women and by holding victims and survivors accountable for the violence perpetrated against them. Minimising and blame-shifting attitudes are directly relevant to domestic violence. Several NCAS items describe attitudes involving misconceptions that domestic violence can and should be handled within the family, in the same way that a minor family disagreement might be privately addressed (Table 7-1). Such minimising attitudes suggest that domestic violence is not serious enough to warrant external assistance or support, including from services, the police and criminal prosecution.

Table 7-1 Thematic item grouping: Domestic violence myths and misconceptions: “It’s a family matter”, 2021

Item	Code	AVAWS subscale	% net disagree a	% net agree b
It’s a woman’s duty to stay in a violent relationship to keep the family together^	D24	Minimise Violence	95	5
Domestic violence is a private matter that should be handled in the family^	D16	Minimise Violence	87	12
Women should keep quiet about domestic violence to protect their family’s reputation	D30	Minimise Violence	97	2
Women going through custody battles often make up or exaggerate claims of domestic violence in order to improve their case^	D23	Mistrust Women	47	37

Note:  N = 19,100 unless otherwise noted. Percentages do not always add to 100 because undecided and unanswered categories are not shown in the table.

a Percentage of respondents who strongly or somewhat disagreed with the item.

b Percentage of respondents who strongly or somewhat agreed with the item.

^ Asked of half the sample.

The idea that women should prioritise family sanctity or reputation over their own safety minimises the gravity of domestic violence. This notion also shifts the burden onto victims and survivors to endure violent acts rather than holding perpetrators accountable for their behaviour (Croucher, 2014; Douglas & Stark, 2010). As Table 7-1 demonstrates, beliefs that domestic violence is a private relationship or family issue persist among a minority of respondents (2–12%; D16, D24, D30). These perceptions are often informed by myths and misconceptions that domestic violence incidents are too minor to report to police, by a lack of awareness that these acts constitute a criminal offence, and by a desire to “keep it private” and deal with domestic violence incidents without outside assistance (ABS, 2013; Carmody, 2009; J. Taylor, 2020). Ethnicity factors may also confer unique challenges for some women that compound their feelings of isolation and powerlessness and prevent them from accessing appropriate support services (Femi-Ajao et al., 2020). These factors include cultural proscriptions against tarnishing family names, valuing family cohesion above all else, fears that reporting will stigmatise one’s cultural group or community, and institutional racism and immigration laws deterring help-seeking (Arce et al., 2020; Dhunna et al., 2021; Fontes & McCloskey, 2011; Hulley et al., 2021; Sawrikar, 2019).

Attitudes that minimise domestic violence by suggesting that this violence should be handled privately are not only a barrier to victims and survivors obtaining assistance but can also lead to mistrust of victims and survivors who choose to seek legal remedies or fight for custody of their children. Such attitudes can contribute to women suffering “secondary victimisation” if they decide to navigate the legal system, whereby they feel silenced, controlled and undermined by the family law system and its agents (Laing, 2016). Thus, addressing attitudes that minimise the seriousness of domestic violence is crucial. As Table 7-1 shows, 37 per cent of respondents agreed that women often fabricate or embellish domestic violence claims for tactical advantage in custody proceedings (D23), contrary to the empirical evidence that this contention is unsubstantiated (Gutowski & Goodman, 2020; Kaspiew & Carson, 2016). This belief among some sections of the community, including some lawyers, perpetuates violence minimisation and blame-shifting attitudes (Tosto & Bonnes, 2022). An evaluation of the 2012 Family Violence Amendments to the Commonwealth  Family Law Act 1975 found that false denials of true violence allegations were actually more common than false reports. Similarly, despite amendments to the Act aimed at improving the family law system’s responsiveness to family violence, there was a decrease in the percentage of mothers experiencing domestic violence since separation who sought a protection order (Kaspiew & Carson, 2016).

Domestic violence myths and misconceptions: “Why does she stay?”

The Mistrust Women Subscale results highlight that when confronted with disclosures of violence victimisation by women, some members of our community are likely to doubt the veracity of the claim or question the severity of the violence (Section 6.2). Shared attitudes that mistrust women can become entrenched as toxic and misogynistic social norms (Section 1.2). Mistrust of women’s reports of violence is also demonstrated by the simplistic assumptions and misconceptions that persist regarding why women remain in violent relationships (S. Murray, 2007; Pugh et al., 2021). For example, there are misconceptions that leaving an abusive relationship is straightforward and that staying in a relationship indicates that the reported violent behaviour is benign. As Table 7-2 indicates, while most respondents disagreed that domestic violence victims already known to police and counselling services deserved less support (D32 and D31), 9–25 per cent of respondents agreed that it is easy to leave an abusive relationship (D28) and that women who don’t leave either are partly responsible for the continuing abuse (D29) or are exaggerating its gravity (D27).

Table 7-2: Thematic item grouping: Domestic violence myths and misconceptions: “Why does she stay?”, 2021

Item	Code	AVAWS subscale	% net disagree a	% net agree b
It’s acceptable for police to give lower priority to domestic violence cases they’ve attended many times before^	D32	Minimise Violence	89	9
Women who stay in abusive relationships deserve less help from counselling and support services than women who leave their abusive partner^	D31	Minimise Violence	91	6
A female victim who does not leave an abusive partner is partly responsible for the abuse continuing	D29	Mistrust Women	72	25
It’s easy for a woman to leave an abusive relationship^	D28	Mistrust Women	88	10
If a woman keeps going back to her abusive partner, then the violence can’t be very serious^	D27	Mistrust Women	88	9

Note: N = 19,100 unless otherwise noted. Percentages do not always add to 100 because the undecided and unanswered categories are not shown in the table.

a Percentage of respondents who strongly or somewhat disagreed with the item.

b Percentage of respondents who strongly or somewhat agreed with the item.

^ Asked of half the sample.

There are many complex reasons why women don’t leave abusive and violent situations, including fears about partner reprisals, the presence of children (or pets) in the home, a lack of financial independence, a lack of knowledge about or access to support services or informal support networks, and many other unique factors (Box 7-1; ABS, 2013; Baly, 2010; Carmody, 2009; Hayes, 2017; Meeting of Attorneys-General, 2022; S. Meyer, 2016; S. Murray, 2007). Two in five NCAS respondents indicated they would not know where to go to access support for domestic violence (Box 7-1 and Figure 7-2).

Summers (2022) brought attention to the devasting choice that many abused women face: remain in a violent relationship or leave and face poverty, often with their children in tow. To safely leave violent relationships, victims and survivors may need organisational, institutional and broad societal support, including financial, housing, legal and emotional support. To assist victims and survivors to leave violent relationships, Commonwealth legislation in 2022 amended the National Employment Standards to provide 10 days’ paid family and domestic violence leave under the Fair Work Amendment (Paid Family and Domestic Violence Leave) Act 2022 (Cth). The availability of safe and stable housing is also required, as is advocacy for sole parents who rent, given the growing rental crisis (Meacham, 2022; Rowley & James, 2018; Wakatama, 2022). Evidence suggests that single mothers particularly struggle to secure housing, even if they have the financial means to pay commercial rents, due to landlord and real estate agent bias (Short et al., 2008; Talbot, 2021).

Research also consistently demonstrates that women are most at risk of acute injury or being killed during the period when they are preparing to leave or leaving a violent relationship (Boxall et al., 2022; Femicide Census, 2022). Indeed, to successfully leave violent relationships, victims and survivors may often need broad, coordinated legal and human services as they are often at crisis point and face elevated rates of a wide range of often severe legal problems with adverse impacts on broad life circumstances (Coumarelos, 2019). Thus, initiatives that provide coordinated, wraparound services across the legal and human services systems are essential (Coumarelos, 2019). In addition to financial assistance and safe housing, victims and survivors may require, for example, access to free legal advice, assistance with navigating the criminal and family court systems, trauma counselling, employment services (such as career coaching to facilitate returning to the workforce) and technological support to assist with managing the impact of technology-facilitated abuse. Examples of coordination between some services for domestic violence include family violence units run by legal services, domestic violence court assistance schemes and health–justice partnerships (Coumarelos, 2018, 2019; Forell & Nagy, 2021).

In addition to fundamental needs such as economic, housing, legal and safety needs, there are also often key emotional reasons why women may stay with an abuser, including biased optimism that the relationship may improve and the psychological impacts of chronic abuse (Martin et al., 2000; Meeting of Attorneys-General, 2022; S. Murray, 2007; Pugh et al., 2021; Sweet, 2019). The Cycle of Violence theory argues that a perpetrator’s changing behaviour from one day to the next can leave women traumatised and depleted of the emotional resources required to leave (Walker, 1979). This theory describes a pattern of violence and abuse involving a tension-building phase when the abuse increases and the victim and survivor tries to defuse the situation; a severe period when the violence is explosive and acute; and a honeymoon phase when the perpetrator expresses remorse and tries to justify or mitigate their behaviour, before tension again begins to build (Walker, 1979). This dysfunctional cyclical pattern has been argued to keep victims and survivors locked into an emotional rollercoaster that may make them question their own assessment of the situation and can delay them leaving the relationship (Both et al., 2019).

Although the Cycle of Violence theory usefully describes some of the emotional reasons why women may stay in abusive relationships, it has been criticised for providing a simplistic and not always accurate view of abusive relationship dynamics and the reasons victims and survivors may stay in abusive relationships (ANROWS, 2019b; Tarrant et al., 2019). Social Entrapment theory provides a more comprehensive analysis of the challenges to leaving abusive relationships by recognising:

• the coercive, strategic and retaliatory nature of perpetrator behaviour, which can entrap victims and survivors and leave them with few practical options for leaving the relationship
• the broader social circumstances and structural inequities which can constrain victims’ and survivors’ ability to leave abusive relationships, including disbelief and lack of trauma-informed, victim-centred responses from family, friends, police, the courts and service providers (ANROWS, 2019b; Tarrant et al., 2019).

Social Entrapment theory has also been used to identify key social and structural constraints when investigating, charging, prosecuting, defending or trying a woman who has killed her violent or abusive intimate partner (Tarrant et al., 2019). Further, it has been argued that education on the Social Entrapment framework should be provided to all those involved in the criminal justice process (ANROWS, 2019b; Tarrant et al., 2019). Similarly, it is important to improve community empathy for victims and survivors and understanding regarding the barriers to leaving violent relationships, as this may help shift attitudes that attribute blame to women who stay in abusive relationships.

Domestic violence myths and misconceptions: “He must have had a reason”

Even if women’s reports of violence victimisation are believed, which they are often not (Section 6.2), erroneous misconceptions about shared responsibility for violence persist. Many of these beliefs are driven by a desire to identify plausible explanations for men’s violence against women. As Table 7-3 shows, 23 per cent of respondents attributed domestic violence to “day-to-day stress” (D17). Not only is this a simplistic explanation for a complex and systemic social problem, but this misconception also grossly minimises and underestimates the significant individual and social impact of domestic violence (see also Section 1.1 and Section 1.2).

Minimising violence against women (Section 6.2) is similarly embodied in prevailing misconceptions about the causes of domestic violence, including that perpetrators are “provoked” or “temporarily lose control” as a result of being goaded, shamed or insulted in some way (Esqueda & Harrison, 2005; J. Hill, 2019). Around 1 in 5 respondents agreed that women can make men so angry that men “accidentally” hit them (D25), and around 1 in 10 respondents agreed that flirting by a woman can trigger an assault by her partner (D26; Table 7-3). These misconceptions shift blame from the perpetrator to the victim and survivor, whose “misconduct” is seen as having triggered or invited the assault (Hockett et al., 2016; Persson & Dhingra, 2022; Suarez & Gadalla, 2010).

Table 7-3 also shows that almost 1 in 10 respondents agreed that violence might be excusable if the offender experienced childhood abuse (D22), and more than 1 in 10 agreed that domestic violence might be excusable if a man temporarily loses control (D18) or if he “regrets” his actions (D19). Prior research confirms that perpetration risk factors such as childhood trauma or substance use are sometimes offered as an explanation to mitigate and deflect perpetrator responsibility (McCloskey et al., 2016; McMurran & Gilchrist, 2008; Mwatsiya & Rasool, 2021; Pugh et al., 2021). Alcohol intoxication is likewise simultaneously identified as both an aggravating and a mitigating factor in violence against women (Balfour et al., 2018; Cafferky et al., 2016; Carline et al., 2018; Gunby et al., 2013; Spencer et al., 2020). A minority of respondents agreed that perpetrator or victim intoxication might excuse violence perpetration (6%; D20 and D21).

Examined together, these minimising and blame-shifting attitudes privilege the abuser’s perspective by offering possible “reasons” for domestic violence. These minimising attitudes obscure perpetrator responsibility and deny the victim’s and survivor’s right to personal safety because of their intoxication or “provocative” behaviour (Bongiorno et al., 2020; Thapar-Björkert & Morgan, 2010). These attitudes also hinder recognition of the power and control disparity that underlies violence against women (Our Watch, 2021a). Thus, it is important to challenge community misperceptions that violence under any circumstances is excusable and to assist perpetrators to accept responsibility for their violent behaviours rather than viewing them as “out-of-character” incidents. Accurate media reporting should also be promoted to facilitate community understanding that domestic violence is a community-wide social problem rather than isolated incidents of aberrant violence where a perpetrator “snapped”.

Table 7-3: Thematic item grouping: Domestic violence myths and misconceptions: “He must have had a reason”, 2021

Item	Code	AVAWS subscale	% net disagree a	% net agree b
A lot of what is called domestic violence is really just a normal reaction to day-to-day stress and frustration	D17	Minimise Violence	74	23
Domestic violence can be excused if it results from people getting so angry that they temporarily lose control	D18	Minimise Violence	84	15
Domestic violence can be excused if, afterwards, the violent person genuinely regrets what they have done	D19	Minimise Violence	85	13
Sometimes a woman can make a man so angry that he hits her when he didn’t mean to^	D25	Minimise Violence	78	19
Women who flirt all the time are somewhat to blame if their partner gets jealous and hits them	D26	Objectify Women	87	11
Domestic violence can be excused if the violent person was themselves abused as a child	D22	Minimise Violence	90	8
Domestic violence can be excused if the offender is heavily affected by alcohol	D20	Minimise Violence	94	6
Domestic violence can be excused if the victim is heavily affected by alcohol^	D21	Minimise Violence	93	6

Note: N = 19,100 unless otherwise noted. Percentages do not always add to 100 because undecided and unanswered categories are not shown in the table.

a Percentage of respondents who strongly or somewhat disagreed with the item.

b Percentage of respondents who strongly or somewhat agreed with the item.

^ Asked of half the sample.

7.3 Sexual violence

The concept of “sexual violence” emerged as a result of feminist discourse that sought to rectify the silencing of women who experience unwanted sexual activity that does not fit with the notion of “stranger rape”. Such silencing serves to limit what counts as sexual violence (Brownmiller, 1975). Consistent with the empirical evidence, the feminist approach emphasised that stranger rape (typically in a “dark alleyway”) is neither the most prevalent form of sexual violence nor the only form of serious or “real” rape. Stranger rape is less common than sexual violence perpetrated in everyday settings such as homes and workplaces by a person known to the victim and survivor, such as a partner, relative, friend, colleague or acquaintance (ABS, 2017; Friis-Rødel et al., 2021; Kelly & Radford, 1990; Waterhouse et al., 2016). Addressing the silencing of victims and survivors of violence requires the ability to identify violent behaviours. One feminist author therefore proposed the concept of a “continuum of sexual violence” to ensure that the full range of non-consensual sexual acts are recognised as sexual violence, instead of only those that were criminalised at the time (Kelly, 1987). Contemporary discourse, research and policy similarly conceptualise “sexual violence” as covering a wide range of criminal and non-criminal sexual activity enacted without consent, and recognise emerging forms of sexual violence that may occur via new or more recent electronic means. For example, it has been recognised that violent, abusive and dehumanising depictions of women in some internet pornography and in the use of sex robots can serve to normalise the objectification of women and to undermine the importance of gaining sexual consent (Bernstein, 2022a; B. Cook et al., 2001; DeKeseredy, 2020).

Sexual violence includes all forms of sexual assault and sexual harassment. Examples of sexual violence include intimidation, unwanted sexual touching, coerced sexual activity, forcing someone to watch and enact pornography, attaining participation in sexual acts through trickery or pressure, reproductive coercion, and many other forms of sexual abuse (Baldwin-White, 2019; Bernstein et al., 2022b; Fahs & Gonzalez, 2014; Henry et al., 2020; Henry & Powell, 2016; Stanley et al., 2018; Tarzia et al., 2020).

Currently there is no prevalence measure that comprehensively captures all types of sexual violence, but it is widely accepted that sexual violence is an evolving and complex social problem that must be examined using a multilayered, multilevel approach (AIHW, 2020; Banyard, 2014).

People of any age or gender can experience sexual violence and perpetrators of sexual violence may be acquaintances, family members, trusted individuals or strangers (National Sexual Violence Resource Center, 2010). Data based on the 2016 PSS, which uses a more circumscribed definition of sexual violence, ^[88] indicates that 23 per cent of Australian women and 8 per cent of Australian men aged 18 years and over have experienced sexual violence at some point in their lifetime, including childhood sexual abuse or sexual assault since the age of 15 years (ABS, 2017). Similarly, the Australian Longitudinal Study on Women’s Health reported high lifetime prevalence of sexual violence experienced during adulthood, ranging from 39 per cent for women respondents aged in their twenties to 12 per cent for women respondents aged 68 to 73 (Townsend et al., 2022). In 2018, the rate of police-recorded sexual assault was almost seven times as high for females as males, and one in three hospitalised sexual assault cases in 2017–18 identified a spouse or domestic partner as the perpetrator (AIHW, 2020).

Victims and survivors of sexual violence report numerous adverse psychological and physical outcomes, including physical injuries, disruption to everyday functioning such as eating and sleeping habits, post-traumatic stress disorder, depression and suicidal ideation (Ades, 2020; Balfour et al., 2018; B. Cook et al., 2001; Hailes et al., 2019). Perpetration of sexual violence is unlikely to be attributable to one layer of influence, but to a convergence of risk factors (and an absence of protective factors) found at individual, relationship, community, organisational and societal levels (Tharp et al., 2013).

On 12 August 2022, the Australian Government’s  Work Plan to Strengthen Criminal Justice Responses to Sexual Assault 2022–27 was endorsed. Under this Work Plan, jurisdictions will seek to take collective and individual action to improve the experiences of victims and survivors of sexual assault in the criminal justice system, focusing on the following priority areas:

• strengthening legal frameworks to ensure victims and survivors have improved justice outcomes and protections, wherever necessary and appropriate, across Australia
• building justice sector capability to better support and protect victims and survivors
• supporting research and greater collaboration to identify best practices, and to ensure actions are supported by a sound and robust evidence base.

The Work Plan will operate alongside ongoing and prospective initiatives that seek to improve responses to sexual violence that are being progressed at both the national and state and territory level. ^[89]

The 2021 NCAS includes the Sexual Violence Scale (SVS), which is further split into a Sexual Assault Scale (SAS) and a Sexual Harassment Scale (SHS). Figure 7-3 shows changes in attitudinal rejection of sexual violence over time by gender according to mean scores on the SVS. For all respondents, and for men and women separately, the mean SVS score was significantly higher in 2021 compared with 2017. These findings indicate a significant increase in the attitudinal rejection of sexual violence by the Australian population overall, and by both men and women separately.

SVS scores in 2021 were also compared by gender (Figure 7-3). There were significant differences in mean SVS scores between genders in 2021. Specifically:

• compared to men, women had significantly higher attitudinal rejection of sexual violence in 2021
• compared to women and men, non-binary respondents had significantly higher rejection of sexual violence in 2021. ^[90]

Figure 7-3: Rejection of sexual violence (SVS) over time by gender, 2009 to 2021

Note: Demographic items for gender were updated for the 2021 NCAS in accordance with the ABS Standard (ABS, 2021h). As the gender item in previous survey waves did not include the same response options for non-binary respondents, only results for men and women can be compared over time. “na” below means reliable data was not available.  Ns for respondents in 2017 and 2021 were:

women – 9,214; 10,091

men – 8,169; 8,822

non-binary respondents – na; na; na; 81

all – 17,419; 19,031.

* Statistically significant difference on this scale between the year indicated and 2021.

*1 Statistically significant difference compared to women and men in 2021.

*2 Statistically significant difference compared to men in 2021.

Figure 7-4 displays change over time for the SAS and the SHS separately. The mean scores for both scales improved in 2021 compared to the 2017 NCAS wave, indicating a positive improvement in Australians’ attitudes towards sexual assault and sexual harassment. Based on all respondents in 2021, there was no significant difference between the mean scores on the SAS and SHS, indicating a similar level of rejection of attitudes condoning sexual assault and attitudes condoning sexual harassment.

Although causation cannot be inferred, the promising improvement in community attitudes towards sexual assault and harassment since 2017 coincides with a period defined by an amplification of awareness, visibility and advocacy regarding sexual violence against women. The #MeToo movement, high-profile criminal trials and accusations of sexual assault and workplace sexual harassment have all served to focus the public discourse (Section 1.1). These movements and events have likewise exposed the pervasive and systemic nature of sexual violence against women. In Australia, the Set the Standard review into Commonwealth parliamentary workplaces provided recommendations for creating Commonwealth parliamentary workplaces that are safe and respectful and reflect best practice in their prevention of and response to bullying, sexual harassment and sexual assault (AHRC, 2021).

Figure 7-4: Rejection of sexual assault (SAS) and sexual harassment (SHS) over time, 2009 to 2021

Note:  Ns in 2017 and 2021 were:

SAS – 17,429; 18,839

SHS – 4,352; 18,764.

Due to insufficient items in earlier NCAS waves, changes over time for SHS can only be reported for 2017 compared to 2021. There was no significant difference between the SAS and SHS scales in 2021.

* Statistically significant difference on this scale between the year indicated and 2021.

The mean scores on the SAS and SHS in 2021 were also compared by gender (Figure 7-5). There were significant differences between genders for both scales in 2021. Specifically:

• Compared to men, women had significantly higher rejection of sexual assault, but were similar in their rejection of sexual harassment.
• Compared to men, non-binary respondents had significantly higher rejection of sexual assault.
• Compared to both men and women, non-binary respondents had significantly higher rejection of sexual harassment.

Figure 7-5: Rejection of different types of sexual violence (SAS and SHS) by gender, 2021

Note:  N = 19,100.

* Statistically significant difference compared to men on this subscale in 2021.

*1 Statistically significant difference compared to women and men on this subscale in 2021.

Section 6.2 discussed how community attitudes condoning violence against women are underpinned by the empirically confirmed constructs underlying the AVAWS subscales, namely minimising violence, mistrusting women and objectifying women. To further examine community attitudes related to sexual violence, we examine thematically grouped items from the SAS and SHS separately. This thematic examination highlights prevailing myths or misconceptions regarding these types of sexual violence and the characteristics of perpetrators and victims and survivors.

Sexual assault: Thematic item examination

The Australian Government’s  Work Plan to Strengthen Criminal Justice Responses to Sexual Assault 2022–27 defines sexual assault as any “type of criminalised sexual violence or harm that involves any physical contact, threat, or intent of contact, of a sexual nature against a person’s will” (Meeting of Attorneys-General, 2022). Legal definitions and interpretations of sexual assault are included in Commonwealth and state and territory law and vary across the jurisdictions.

Sexual assault is a major health and welfare issue in Australia and across the world. The effects of a sexual assault can be extensive and chronic across a person’s lifetime. Victims and survivors may experience physical injuries, long-term mental health effects and disruption to their day-to-day functioning (AIHW, 2020; Balfour et al., 2018; Hailes et al., 2019). Between 2010 and 2018, rates of sexual assault victimisation recorded by police for Australians aged 15 and over rose by more than 30 per cent. Recent data further revealed an increase of 2 per cent in sexual assault victims from 2019 to 2020, representing the highest number of victims recorded since the commencement of the 28-year time series (ABS, 2019, 2021e).

Women are more likely than men to be victims of sexual assault, with 84 per cent of the recorded sexual assault victims in 2020 being women (ABS, 2021g). In 2018–19, almost all sexual assault offenders recorded by police were male (97%), with males aged 15 to 19 demonstrating the highest offender rates (ABS, 2020). Most (95%) sexual assaults against women are committed without a weapon, exemplifying the clear power imbalance between male perpetrators and female victims, whereby fear alone is enough to facilitate an assault (ABS, 2021g). Notably, around 9 out of 10 Australian women did not report their most recent sexual assault victimisation to police because they felt ashamed or embarrassed, or assessed the incident as not being serious enough to report (ABS, 2017).

Sexual assault myths and misconceptions: “She’s not a genuine victim”

Section 6.2 explored how mistrust of women reporting victimisation is based on gendered and hostile stereotypes of women as malicious liars who routinely serve an agenda to harm and vilify men (Emmers-Sommer, 2017; Harmer & Lewis, 2022; Rees & White, 2012). Recent ANROWS research similarly found that participants defaulted to a position of doubt and suspicion when asked to appraise a woman’s allegation of sexual assault, engendering a range of unrealistic standards and conditions to be met for the allegation to be believed (Minter et al., 2021). Rape myths regarding the characteristics of “genuine” sexual assault victims, coupled with assumptions that women frequently lie about sexual assault, promote hostile scepticism about sexual assault disclosures from the outset (Boux & Daum, 2015; Edwards et al., 2011; Rumney, 2006).

As Table 7-4 shows, misconceptions about why women delay reporting a sexual assault or their motives for reporting still prevail. More than one in three respondents (34%) agreed that sexual assault is commonly used to get back at men (S23) and almost one quarter (24%) agreed that sexual assault allegations could be a response to a regretted sexual encounter (S24). Such mistrustful attitudes impact whether victims and survivors report sexual violence, whether bystanders intervene and whether key stakeholders, including police and judges, believe women (G. D. Anderson & Overby, 2021; Carretta et al., 2016; K. J. Holland et al., 2021; H. Johnson, 2017; Temkin et al., 2018). Research indicates that fear of not being believed and fear of retribution are key factors in whether women disclose sexual assault to their informal support networks and formally report sexual assault to police or authorities (K. J. Holland & Cipriano, 2019; K. J. Holland et al., 2021; O’Donohue, 2019; Reich et al., 2021; Wamboldt et al., 2019; Weiss, 2013; Whiting et al., 2020; Wilson et al., 2021). Sexual assault is also one of the most difficult offences to successfully prosecute, with around 85 per cent of sexual assaults never reaching the criminal justice system (ABS, 2021g; Lievore, 2003).

Myths and misconceptions regarding victimisation are also likely to contribute to the low reporting and prosecution numbers. The results in Table 7-4 suggest that victims and survivors need to meet certain parameters or characteristics to be seen as a “genuine victim” and be believed. Although most respondents disagreed, 3–6 per cent of respondents agreed that lesbian or bisexual women, women with mental health issues and women who can’t demonstrate physical resistance or injury are not “genuine” victims of sexual assault (S2, S1, S17, S22). Prior research confirms that women with disability, people with diverse genders and sexualities, women of colour and women from various cultural backgrounds face additional challenges when reporting a sexual assault to authorities, driven by racism, ableism and heteronormative assumptions about sexual violence victims (Palmer & St. Vil, 2018; Slatton & Richard, 2020).

Misunderstandings of the law, including that sexual assault evidence is predicated on physical injury and resistance, have also been noted in previous studies (Haugen et al., 2018; Kassing & Prieto, 2003; Rodríguez-Madera et al., 2017; Wirtz et al., 2018).

Thus, it is important to correct myths and misconceptions about the nature of sexual assault and “genuine” victims within the community and justice and service systems, including by correcting hostile gendered stereotypes of women as malicious, vindictive and untrustworthy; addressing persistent myths that false allegations of sexual assault are common; and increasing recognition of the diversity of ways that sexual assault can be experienced and responded to by victims and survivors.

Table 7-4: Thematic item grouping: Sexual assault myths and misconceptions: “She’s not a genuine victim”, 2021

Item	Code	AVAWS subscale	% net disagree a	% net agree b
Women with mental health issues who report being sexually assaulted are probably lying	S1	Mistrust Women	86	6
When lesbian or bisexual women claim to have been sexually assaulted by their partner, they probably shouldn’t be taken too seriously	S2	Mistrust Women	93	3
If a woman doesn’t physically resist – even if protesting verbally – then it isn’t really rape~	S17	Objectify Women	91	6
Many allegations of sexual assault made by women are false~	S18	Mistrust Women	78	14
If a woman claims to have been sexually assaulted but has no other physical injuries, she probably shouldn’t be taken too seriously	S22	Mistrust Women	93	5
It is common for sexual assault accusations to be used as a way of getting back at men	S23	Mistrust Women	57	34
A lot of times, women who say they were raped had led the man on and then had regrets	S24	Mistrust Women	66	24
Women who wait weeks or months to report sexual assault are probably lying	S25	Mistrust Women	90	7

Note:  N = 19,100 unless otherwise noted. Percentages do not always add to 100 because undecided and unanswered categories are not shown in the table.

a Percentage of respondents who strongly or somewhat disagreed with the item.

b Percentage of respondents who strongly or somewhat agreed with the item.

~ Asked of one quarter of the sample.

Sexual assault myths and misconceptions: “We expect too much of men”

Section 6.2 examined the impact of objectifying women, including by the media, and disregarding women’s consent in sexual interactions. Objectifying attitudes are exemplified in the contradictory notions that women invite sexual assaults by their inappropriate behaviour and choices but should also always make themselves desirable and sexually available to men (Bareket et al., 2018; Carline et al., 2018; Harmer & Lewis, 2022; O’Hara, 2012). These objectifying norms and sexist double standards perpetuate notions that women are responsible for keeping themselves safe from men’s violence and relieve men from accountability (Brownhalls et al., 2020; Davey, 2018). Similarly, these attitudes reflect heteronormative beliefs about stereotypical gender roles and problematic heterosexual sex scripts that privilege men’s entitlement to sex as aggressive initiators and position women as passive “gatekeepers” who must resist men’s advances. This perspective rationalises men’s non-consensual sexual behaviour on the grounds that it is “natural” due to the perception that it is biologically difficult for men to regulate their own sexual drives (Frith, 2009; Gavey, 2018; Hirsch et al., 2019; Jeffrey & Barata, 2017).

The notion that women should protect themselves from sexual violence is not mutually exclusive from the notion that perpetrators should be accountable, but they elicit different places of intervention (Brownhalls et al., 2020). The first perspective places the onus on women as sexual gatekeepers who must stay vigilant to actively resist the insatiable and inevitable demands of men, while the latter seats responsibility with men to regulate their own sexual and moral behaviour.

The items in Table 7-5 illustrate many of these notions. While most respondents disagreed with these objectifying attitudes, 1 in 4 respondents agreed that a sexually aroused man may be “unaware” that a woman has refused consent (25%; S8) and 1 in 10 respondents agreed that women say “no” when they mean “yes” (10%; S5). Attitudes minimising violence against women due to intoxication were also evident among a minority of respondents, with 1 in 20 agreeing that an intoxicated man is less responsible for perpetrating sexual assault (6%; S19).

Objectifying and blame-shifting attitudes are also demonstrated by the gendered double standard whereby offenders are pardoned for their intoxication and sexual drives, but victims and survivors are censured for their intoxication and dating choices. Around 1 in 10 respondents (6–10%) agreed that an intoxicated woman is partly responsible if she is sexually assaulted (S4 and S20) and 7 per cent agreed that meeting up with a man she met on a dating app renders a woman partly responsible for her assault (S21). These objectifying and blame-shifting attitudes held by a minority of the community belie the need for affirmative and ongoing sexual consent.

Table 7-5: Thematic item grouping: Sexual assault myths and misconceptions: “We expect too much of men”, 2021

Item	Code	AVAWS subscale	% net disagree a	% net agree b
If a woman is drunk and starts having sex with a man, but then falls asleep, it is understandable if he continues having sex with her anyway	S4	Objectify Women	89	6
If a woman is raped while drunk or affected by drugs, she is at least partly responsible	S20	Objectify Women	88	10
If a woman meets up with a man she met on a mobile dating app, she’s partly responsible if he forces sex on her	S21	Objectify Women	92	7
Women often say “no” when they mean “yes”	S5	Objectify Women	86	10
When a man is very sexually aroused, he may not even realise that the woman doesn’t want to have sex	S8	Objectify Women	69	25
A man is less responsible for rape if he is drunk or affected by drugs at the time^	S19	Minimise Violence	93	6

Note: N = 19,100 unless otherwise noted. Percentages do not always add to 100 because undecided and unanswered categories are not shown in the table.

a Percentage of respondents who strongly or somewhat disagreed with the item.

b Percentage of respondents who strongly or somewhat agreed with the item.

^ Asked of half the sample.

Sexual assault myths and misconceptions: “Rapes are committed by strangers”

Many myths and misconceptions endure about the nature and context of sexual assaults (Adolfsson et al., 2017; Basow & Minieri, 2011; Yapp & Quayle, 2018). These myths include the stranger rape myth that women are more likely to be raped by a stranger in the public domain rather than by someone they know in a private space and that consent can be assumed as ongoing or “owing” within a dating or established relationship (also see Box 7-2), and that it cannot be revoked (Angelone et al., 2015; Baldwin-White, 2019; Basow & Minieri, 2011; Bieneck & Krahe, 2011; Carline et al., 2018; Jeffrey & Barata, 2019; Waterhouse et al., 2016; Webster et al., 2018a).

As Figure 7-6 (Box 7-2) shows, most respondents (69%) strongly or somewhat agreed that a woman is more likely to be raped by someone they know than a by stranger (S16). However, the stranger rape myth was evident among almost one third of respondents, who either disagreed (18%) or didn’t know (12%) that a woman is more likely to be raped by someone they know. Perceptions about stranger rape fuel sensationalised media reports and can serve to curtail women’s freedoms to freely move around the community under the mistaken impression that stranger-based sexual assaults are the norm rather than the exception (Merken & James, 2020). These beliefs also contribute to rape by known persons being ignored or not taken seriously by police and the justice system and being overlooked in legal and policy reform, allowing misconceptions about the law to prevail (Brooks-Hay, 2019; Dinos, 2014; Lundrigan et al., 2019; Persson & Dhingra, 2022). Misconceptions that rape is typically perpetrated by strangers who use physical force can draw attention away from the need to ensure affirmative and ongoing consent in everyday sexual relationships, including intimate partner and dating relationships.

A few NCAS items examined respondents’ understanding of sexual consent. Figure 7-7 (Box 7-2) shows that while most respondents correctly understood that sexual assault in marriage is a criminal offence (S26), 20 per cent either said this is not a criminal offence or were unsure if it is.

In addition, as discussed in Section 6.2, respondents were presented with two scenarios about sexual consent, one describing a married couple and the other describing a couple who had just met at a party. Both scenarios asked respondents whether they felt sexual assault was justified under two contexts: 1) the man had initiated kissing before the woman pushed him away, and 2) the woman had initiated kissing before pushing him away (items S12 to S15). As Figures 6-7 and 6-8 show, although only 3 per cent agreed that sexual assault was justified when the man had initiated kissing, 8–11 per cent thought sexual assault was justified if the woman had initiated kissing and then pushed the man away. This finding indicates that a minority of respondents did not appreciate the need to obtain consent at every stage of sexual activity and failed to recognise that consent can be withdrawn at any stage.

These misperceptions about the nature of sexual consent highlight the need for nationally consistent definitions that remove any ambiguity about the nature of consent by legislating for the need for affirmative and ongoing consent. Under an “affirmative consent standard”, consent must be mutually confirmed, silence or lack of resistance cannot be interpreted as consent, and consent can be withdrawn at any point during sexual activity. An affirmative consent standard shifts the emphasis from the actions of the victim and survivor to those of the accused. As discussed in Section 1.1, although most jurisdictions have recently taken steps towards improving their sexual assault laws, an affirmative consent standard is not yet implemented nationally, leading to inconsistencies across jurisdictions, with people accused of rape still able to argue in some Australian jurisdictions that they had a reasonable belief of consent but were mistaken (Bucci, 2021; Burgin, 2019). In addition, awareness and education initiatives are needed to increase community understanding of affirmative and ongoing consent, both within established relationships and in more casual contexts. It is also important to shift problematic heterosexual sex scripts that privilege men’s entitlement to sex by positioning men as dominant and aggressive sexual initiators and women as submissive sexual gatekeepers, as these place the responsibility of voicing consent and preventing sexual violence on women while absolving men from responsibility (Brady et al., 2018).

Sexual harassment: Thematic item examination

Sexual harassment is a widespread and global problem that affects individuals of all genders, but is more commonly experienced by women and girls than men and boys. It is defined as an unwelcome sexual advance, unwelcome request for sexual favours or other unwelcome conduct of a sexual nature which makes a person feel offended, humiliated and/or intimidated, where a reasonable person would anticipate that reaction in the circumstances ( Sex Discrimination Act 1984  [Cth]). While prevalence data remains limited, sexual harassment is estimated to be experienced by between 51 per cent (European Union) and 81 per cent (United States) of people during their lifetime (Lim et al., 2018).

Sexual harassment can occur within institutions and workplaces, and in public spaces and online, and it includes a range of behaviours (AHRC, 2020). Examples of sexual harassment include, but are not limited to:

• verbal harassment, such as sexually suggestive comments or jokes, intrusive questions, comments about physical appearance, repeated invitations to go on dates, or requests or pressure for sex
• sexually explicit pictures, posters or gifts
• intimidating or threatening behaviours, such as inappropriate staring or leering, sexual gestures or indecent exposure
• inappropriate physical contact, such as unwelcome touching
• harassment involving the use of technology, such as sexually explicit emails, texts or social media; indecent phone calls; repeated or inappropriate advances online; or sharing or threatening to share intimate images (AHRC, 2020; eSafety, 2022a, 2022f, 2022h).

In Australia, the  Sex Discrimination Act 1984 (Cth) prohibits sexual harassment at work and established the role of the Sex Discrimination Commissioner. Although the elimination of sexual harassment, particularly within workplaces, has been a key focus since 1984, Australia has lagged behind many other nations in its progress towards this goal (AHRC, 2020). It is estimated that 53 per cent of Australian women experience some form of sexual harassment in their lifetime, including, for example, street harassment and workplace harassment. Aboriginal and/or Torres Strait Islander peoples are also more likely to experience workplace sexual harassment than people from non-Indigenous backgrounds (53% compared to 32%; AHRC, 2020).

Sexual harassment myths and misconceptions: “It’s a compliment”

As discussed in Section 6.2, attitudes that legitimise objectifying women as sexual objects diminish and silence women when they face harassment, aggression or violence (Loreck, 2016; L. McDonald, 2022). This sexualised way of viewing women empowers men as “the viewers” and disempowers women as mere objects of men’s desire (Wright, Arroyo et al., 2015). Attitudes that objectify women are also exemplified by the misconception that women always welcome any form of sexual attention or that they provoke sexual attention, for example, by placing themselves in a particular context (e.g. a bar or club) or by the way they dress or act either in the public, private or online domain (S. Becker & Tinkler, 2021; Gillett, 2018, 2021). Relatedly, harassers often claim to be surprised when their attempts at “humour” or “flattery“ or their sexual overtures are met with offence or insult, displaying a disregard for consent and an assumption of sexual entitlement (Bouffard, 2010).

Table 7-6 shows the results for items describing the myth that sexual attention is always welcome and should be viewed as a compliment, regardless of consent. Although most respondents disagreed with these statements that objectify women, 10–21 per cent agreed that women find it flattering to receive catcalls in public or be persistently pursued (S3, S11) and that men are entitled to touch women or share naked pictures of their partner without permission (S6, S7). These attitudes that objectify women seem to persist despite research consistently demonstrating that women find these acts troubling, threatening and violating, and that they report wide-ranging psychological and physical impacts (AHRC, 2017a; Blumell & Mulupi, 2020; Lim et al., 2018). Such attitudes disregard consent (S6, S7, S11) and apportion blame to victims and survivors rather than holding harassers accountable for their behaviour (S6).

Table 7-6: Thematic item grouping: Sexual assault harassment myths and misconceptions: “It’s a compliment”, 2021

Item	Code	AVAWS subscale	% net disagree a	% net agree b
A woman should be flattered if she gets wolf-whistles or catcalls when walking past a group of men in public^	S3	Objectify Women	82	13
Women find it flattering to be persistently pursued, even if they are not interested^	S11	Objectify Women	81	13
Since some women are so sexual in public, it’s understandable that some men think they can touch women without permission	S7	Objectify Women	89	10
If a woman sends a naked picture to her partner, then she is partly responsible if he shares it without her permission	S6	Objectify Women	77	21

Note: N = 19,100 unless otherwise noted. Percentages do not always add to 100 because “undecided” and “unanswered” categories are not shown in the table.

a Percentage of respondents who strongly or somewhat disagreed with the item.

b Percentage of respondents who strongly or somewhat agreed with the item.

^ Asked of half the sample.

Sexual harassment myths and misconceptions: “You’re making too much of it”

As discussed (Section 6.2), many reports of violence against women are met with mistrust, and research indicates that reports of sexual harassment are also frequently disbelieved (Bongiorno et al., 2020; Easteal & Judd, 2008; Harmer & Lewis, 2022). Sexual harassment at work is prevalent and pervasive, occurring in every industry, in every location and at every level of both private and government organisations (Lim et al., 2018). It is a serious form of sexual violence with serious financial, social, emotional, physical and psychological harms, whether it occurs via technology or in person (AHRC, 2020; Harmer & Lewis, 2022). In 2018, workplace sexual harassment in Australia cost an estimated $2.6 billion in lost productivity and $0.9 billion in other financial costs, with each case of harassment representing around four working days of lost output. It was estimated that employers bore 70 per cent of this financial cost, with government bearing 23 per cent and individuals bearing 7 per cent. Impacts on victims’ and survivors’ wellbeing accounted for an additional $250 million, or nearly $5,000 per victim on average (AHRC, 2020).

Despite these manifest impacts, a small percentage of NCAS respondents (5–7%) have attitudes that minimise the significance of sexual harassment and mistrust women who delay reporting of sexual harassment (Table 7-7). Research suggests that despite the prevalence and seriousness of sexual harassment, many misconceptions persist and often delay timely reporting. Victim-blaming, underestimation of the impact of sexual harassment, and disbelief of victims and survivors are common features in many domains (Berdahl & Aquino, 2009; Blumell & Mulupi, 2020; Easteal & Judd, 2008; Worthington & Snape, 2021). Given these prevailing myths and misconceptions, it is not surprising that a recent survey found that only 17 per cent of people who experienced sexual harassment at work in the previous five years made a formal report or complaint about the harassment (AHRC, 2018a). Of the small percentage of incidents reported, even fewer are formalised as complaints to the AHRC or equivalent regulatory body (AHRC, 2019). Research suggests the system and process for reporting sexual harassment encourages silence, with complainants who settle before going to court often motivated to do so because of concerns about the time and cost of litigation, problems with proving their discrimination experience and low compensation (D. Allen, 2009).

Thus, it is important to ensure that all spaces, including workplaces, educational settings and online forums, are safe and respectful through legislation and policy frameworks and through initiatives that challenge misconceptions that sexual harassment is not serious, raise awareness of the different forms of sexual harassment that can occur online and in person, and address attitudes that objectify women or disregard consent.

Table 7-7: Thematic item grouping: Sexual assault myths and misconceptions: “You’re making too much of
it”, 2021

Item	Code	AVAWS subscale	% net disagree a	% net agree b
Women who wait weeks or months to report sexual harassment are probably lying	S10	Mistrust Women	90	7
Women who are sexually harassed should deal with it themselves rather than report it	S9	Minimise Women	93^	5

Note: N = 19,100 unless otherwise noted. Percentages do not always add to 100 because “undecided” and “unanswered” categories are not shown in the table.

a Percentage of respondents who strongly or somewhat disagreed with the item.

b Percentage of respondents who strongly or somewhat agreed with the item.

^ Asked of half the sample.

7.4 Technology-facilitated abuse

Modern technology has enabled increasingly sophisticated methods for facilitating and amplifying violence against women (Afrouz, 2021; Henry et al., 2020). Representing an extension of more traditional forms of violence, abuse, coercion and harassment, technology-facilitated abuse is an umbrella term used to refer to abuse that is facilitated by technology, including digital devices and social platforms, where technology is the conduit or means of enacting or exercising abuse. A recent ANROWS study similarly confirmed the prevalence of technology-facilitated abuse and that women are more likely than men to be the targets of such abuse (Powell et al., 2022). While anonymity is not a necessary feature of technology-facilitated abuse, the online context of this violence can sometimes provide perpetrators a degree of anonymity and perceived insulation from the consequences of their behaviours, a perception that may allow the abuse to intensify and escalate (Cuenca-Piqueras et al., 2020).

There are four main forms of technology-facilitated abuse:

• harassment, including sending multiple abusive messages and frequent unwanted contact across multiple platforms and communication methods
• stalking, including electronic tracking of an individual and filming them without consent
• impersonation, such as taking over internet accounts and locking the owner out of the account
• threats, including sharing or threatening to share intimate images or videos of a person without their consent (eSafety, 2022a).

Technology-facilitated abuse is often a means of enacting domestic or sexual abuse. Evidence indicates that technology-facilitated abusive behaviours are often part of an ongoing pattern of domestic violence. A survey of domestic and family violence frontline workers found that 99 per cent of these workers reported having clients who had experienced technology-facilitated stalking and abuse (Woodlock, Bentley et al., 2020). Although technology-facilitated abuse is often a means of enacting domestic or sexual violence, it can also be a separate or primary form of abuse. For example, harassing, threatening and intimidating behaviours that do not involve sexual content can be enacted online (e.g. via a dating app) against a victim whom the perpetrator has never met in person, where the only contact with the victim has been online.

Similarly, women are also often the targets of online sexist and derogatory comments, trolling and threats, particularly when they are required to maintain an online presence as part of their job (Media Entertainment and Arts Alliance & Gender Equity Victoria, 2019; Pew Research Centre, 2017). A recent study found more than one third of women journalists had experienced online harassment, trolling and stalking during the course of their work, but only 16 per cent said that they were aware of their workplace having a policy to address online abuse (Media Entertainment and Arts Alliance & Gender Equity Victoria, 2019).

Many of these behaviours are crimes under the law in Australia and cover behaviour such as stalking, sending threatening emails and texts, using tracking apps and spyware, online bullying, and sharing intimate images or videos without consent. A range of new Commonwealth civil law penalties now exist for technology-facilitated abuse, including online stalking, and eSafety has also recently been given greater powers of enforcement for some forms of technology-facilitated abuse (eSafety, 2022h;  Online Safety Act 2021 [Cth]).

The 2021 NCAS introduced a six-item Technology-Facilitated Abuse Scale (TFAS) to examine these evolving types of violence against women. Four TFAS items are drawn from the UVAWS and examine the recognition of different forms of technology-facilitated abuse, while the other two TFAS items are drawn from the AVAWS and examine attitudes towards technology-facilitated abuse. [91]

Understanding and rejection of technology-facilitated abuse between genders

In 2021, the only significant difference on TFAS scores by gender was the significantly higher mean score on understanding and rejection of technology-facilitated abuse for non-binary respondents compared to men (Figure 7-8). Changes over time in TFAS scores are not reported, due to insufficient data in earlier NCAS waves.

Figure 7-8: Understanding and rejection of technology-facilitated abuse (TFAS) by gender, 2021

Note: N = 19,100

* Statistically significant difference compared to men in 2021.

Technology-facilitated abuse myths and misconceptions: “What happens online isn’t serious”

The AVAWS Objectify Women Subscale (Section 6.2) examines attitudes supporting the objectification of women as commodified objects for the sexual gratification of men (Loughnan et al., 2013; Wesselmann et al., 2021, p. 841). This dehumanisation of women permits men a sense of entitlement and superiority to treat women as sexual property that can be disrespected and abused, particularly online, without consequence (Bernstein et al., 2022a; Tarzia, 2020). Just as women are disproportionately the victims of domestic violence, approximately two thirds of the reports of online abuse received by eSafety are from Australian women and girls (eSafety, 2022a). Women and girls are particularly at risk of being coerced into sharing sexual imagery, receiving unwanted sexual imagery, receiving threats to share intimate imagery without their consent, and being abused and harassed online (DeKeseredy et al., 2019; Harris & Woodlock, 2022; Zhou, 2020). A recent study also indicated Aboriginal and/or Torres Strait Islander women are at increased risk of online hate and serious online harm, and they experience online harm and abuse using digital devices (e.g. phones) as part of family violence at much higher rates than the general population (Brown et al., 2021). The consequences of these behaviours can include psychological distress, fear and even suicidal ideation (eSafety, 2019b; Henry et al., 2017, 2020).

The sharing of intimate images without an individual’s consent is defined as image-based abuse (eSafety, 2022h). Image-based abuse includes images or videos that have been digitally altered and making threats to share an intimate image. Research conducted by eSafety found 11 per cent of Australians aged 18 and over have had a nude or sexual photo or video posted online or sent on without their consent, with women aged 18 to 24 most likely to be the targets of this kind of abuse (eSafety, 2017). The  Online Safety Act 2021 (Cth), which replaced the  Enhancing Online Safety Act 2015 (Cth), established a civil penalties scheme to address image-based abuse across Australia. This scheme allows victims of image-based abuse to make a report to eSafety, which may be able to get content removed or act against the person responsible. In 2018, changes made to the civil penalties scheme mean that “intimate images” relating to image-based abuse include not only nude and sexual images, such as images of genitalia, but also images of a person without the religious or cultural attire they would normally wear in public (eSafety, 2019b). The  Criminal Code Act 1995 (Cth) includes an offence of “using a carriage service to menace, harass or cause offence”, and was amended in 2018 to include an aggravated offence if the use of the carriage service involved private sexual material. In addition to federal laws, most Australian states and territories also have their own criminal laws, which specifically address image-based abuse (eSafety, 2022h).

Table 7-8 and Figure 7-9 show results for four NCAS items which tap into the misconception that image-based abuse and online abuse more broadly is not that serious or does not constitute “real life”. As Table 7-8 shows, one in five respondents felt a woman was partially to blame if an intimate image she provided to a partner was shared without her consent (21%; S6). This response exemplifies the blame shifting and objectifying attitudes discussed in Chapter 6, whereby a woman’s body is treated as sexual property to be shared at will. This attitude also demonstrates limited recognition of how this behaviour violates, humiliates and dehumanises women, with the distress women experience being amplified by blaming them for the offender’s behaviour. In addition, a smaller percentage of respondents (6–9%) did not recognise that a man sending an unwanted picture of his genitals to a woman (V7) and abusive messages and comments targeted at women on social media (V6) are forms of violence against women. Figure 7-9 shows that about 1 in 10 (11%) respondents did not recognise that sharing a sexual picture of an ex-partner on social media without their consent is a criminal offence or were uncertain if it is an offence (S27). These findings suggest that while most Australians understand that technology-facilitated abuse is harmful and can attract criminal penalties, more work could be done to increase awareness of the diverse forms that this abuse can take and to change beliefs that this abuse is not serious. In addition, safety-by-design principles could be used to enhance the safety of digital spaces and the digital literacy of the community could also be enhanced to facilitate recognition and reporting of technology-facilitated abuse and enhance skills for accessing support.

Table 7-8: Thematic item grouping: Technology-facilitated abuse (TFAS) and stalking myths and misconceptions: “What happens online isn’t serious”, 2021

Attitude item	Code	AVAWS subscale	% net disagree a	% net agree b
If a woman sends a naked picture to her partner, then she is partly responsible if he shares it without her permission	S6	Objectify Women	77	21

Understanding item	Code	Scale	% strong yes c	% no
Is this a form of violence against women … a man sends an unwanted picture of his genitals to a woman?	V7	UVAWS	80	9
Is this a form of violence against women … abusive messages or comments targeted at women on social media?	V6	UVAWS	83	6

Note: N = 19,100 unless otherwise noted. Percentages do not always add to 100 because undecided and unanswered categories are not shown in the table.

a Percentage of respondents who strongly or somewhat disagreed with the item.

b Percentage of respondents who strongly or somewhat agreed with the item.

c Percentage of respondents who answered “Yes, always” or “Yes, usually”.

Figure 7-9: Thematic item grouping: Technology-facilitated abuse (TFAS) myths and misconceptions:
“What happens online isn’t serious”, 2021

Note: N = 4,789. Asked of one quarter of the sample in 2021.

7.5 Stalking: Technology-facilitated and in person

Stalking is a form of violence against women that occurs both in person and in the form of technology-facilitated abuse. Stalking is a common feature of intimate partner violence but can also occur outside domestic relationships. Whether online or in person, stalking entails a pattern of repeated, frequently intrusive behaviours intended to maintain contact with or exercise power and control over another person. These behaviours are enacted to intimidate or cause fear, distress and loss of control in the target (Campbell, 2019; Victorian Law Reform Commission [VLRC], 2021). Stalking is a criminal offence in all Australian states and territories. ^[92]  In Australia, 1 in 6 women and 1 in 15 men reported experiencing stalking since the age of 15 (ABS, 2017). Most stalking instances reported by men and women were perpetrated by a man. Research indicates that in-person and online stalking are related, with women who are stalked in person being more likely to be subsequently stalked online (Reyns & Fisher, 2018).

As noted earlier, the 2021 NCAS included three items on stalking, one on online stalking and two on in-person stalking:

• The item on online stalking was included in the TFAS and examined recognition that online stalking by a partner is a form of domestic violence.
• The in-person stalking items were not part of the TFAS. One of these items examined recognition that stalking is a form of violence against women and the other examined attitudes to in-person stalking.

The three stalking items were insufficient to form a psychometrically valid scale examining understanding and attitudes regarding stalking. Nonetheless we examine them together here as they involve similar underlying misconceptions.

Online and in-person stalking myths and misconceptions: “I’m just checking in and looking out for her”

Table 7-9 shows that most respondents recognised technology-facilitated and in-person stalking as violence. Almost all respondents strongly or somewhat disagreed with the statement that in-person stalking “is only really stalking if it is by a stranger” (95%; V8), and most recognised that in-person stalking is “usually” or “always” a form of violence against women (89%; V4). Similarly, most respondents recognised that electronically tracking a partner is usually or always a form of domestic violence (83%; D6). Our findings indicate increased recognition between 2017 and 2021 of stalking as a form of domestic violence and violence against women, especially among men, although a minority of respondents did not see these behaviours as violence (4–7%).

Our findings also indicated that men (78%) are still significantly less likely than women (88%) to recognise electronic tracking by a partner as always or usually a form of domestic violence (D6). This finding suggests a sense of entitlement and “benevolent sexism” ^[93]  may still prevail among some men in the community (A. Becker et al., 2020; Tarzia, 2020). A recent study with young people in five European countries found that the online domain has provided new patriarchal platforms for extending the scope and regularity of monitoring, control and emotional abuse (Aghtaie et al., 2018). This abuse was normalised and perpetuated when young people equated control to love, care and protection (Aghtaie et al., 2018).

Although most respondents recognised online and in-person stalking as violence, it remains important to continue to challenge the myth that stalking behaviours are harmless or are only perpetrated by a stranger (V8). Research indicates that the misconception that stalking is only perpetrated by strangers is relatively common (McKeon et al., 2014; Sheridan et al., 2003; Sinclair, 2012). Research also suggests that, for both men and women, stalking is increasingly used as part of coercive control and that being a victim and survivor of intimate partner violence during a relationship is associated with increased likelihood of becoming a victim of stalking after the relationship has ended (Breiding et al., 2011; Campbell, 2019; Englebrecht & Reyns, 2011; Senkans et al., 2021). Peer networks have also been demonstrated to be key sites of intervention for challenging stalking behaviours (DeKeseredy et al., 2017, 2019). Thus, it is important to increase community understanding of the seriousness of stalking and the different forms it can take both in person and online.

It is also important to shift the burden away from victims and survivors of stalking and towards perpetrator accountability (VLRC, 2022). The level of evidentiary proof needed to seek recourse through the justice system can be challenging and can require victims to collect evidence, apply for an intervention order and manage their risk of harm (Jerath et al., 2022; NSW Government & NSW Police, 2022). A recent report recommended financial and practical support for victims and survivors to prevent cyberstalking, as well as support by independent advocates to guide them through the justice system from the point of reporting the offence to any court actions (VLRC, 2022).

Table 7-9: Thematic item grouping: Technology-facilitated and in-person stalking myths and misconceptions:
“Just checking in”, 2021

Form of stalking	Understanding Item	Code	Scale	% strong yes a	% no
Online	Is this a form of domestic violence … repeatedly keeps track of partner on electronic devices?	D6	TFAS and UVAWS	83	7
In person	Is this a form of violence against women … stalking by repeatedly following/watching at home/work?	V4	UVAWS	89	4

Form of stalking	Attitude Item	Code	AVAWS subscale	% net disagree b	% net agree c
In person	It’s only really stalking if it’s by a stranger^	V8	Minimise Violence	95	4

Note: N = 19,100 unless otherwise noted. Percentages do not always add to 100 because undecided and unanswered categories are not shown in the table.

a Percentage of respondents who answered “Yes, always” or “Yes, usually”.

b Percentage of respondents who strongly or somewhat disagreed with the item.

c Percentage of respondents who strongly or somewhat agreed with the item.

^ Asked of half the sample.

7.6 Conclusions about types of violence against women

The results above indicate that Australians’ attitudinal rejection of sexual violence, sexual assault, sexual harassment and technology-facilitated abuse continue to improve. Attitudes towards domestic violence have improved over the long term but have plateaued since 2017. However, various myths and misconceptions are evident in a minority of the community regarding all of these types of violence and need to be addressed. In addition, there is a need to correct gaps in community knowledge of laws about sexual consent and technology-facilitated abuse. Prevention initiatives should:

• Develop nationally consistent definitions of domestic violence and coercive control, sexual violence and sexual consent, and technology-facilitated abuse across legislative and policy settings Australia-wide and raise community awareness of these definitions.
• Educate the community about the range of behaviours that constitute different types of violence against women to build community capacity to recognise and respond appropriately to all types of violence.
• Educate the community about the seriousness of all types of violence, including sexual harassment, technology-facilitated abuse and stalking, by raising awareness of their high prevalence, harmful psychological impacts and their legal penalties and by addressing attitudes that excuse or minimise violence or shift blame to victims and survivors.
• Improve understanding of the barriers domestic violence victims and survivors may face in leaving violent relationships. Findings regarding cultural proscriptions against involving outsiders in domestic violence matters emphasise the importance of working with communities to assess their needs and the points at which intervention may be most useful.
• “Personalise” domestic violence as a community-wide problem that requires community-wide responsibility, and promote accurate media reporting of domestic violence as an ongoing pattern of abusive behaviour rather than isolated incidents of aberrant violence where a perpetrator “snapped”.
• Correct rape myths about “stranger rape” and “genuine” victims, including by correcting hostile gendered stereotypes of women as malicious, vindictive and untrustworthy; addressing persistent myths that false allegations of sexual assault are common; and increasing recognition of the diverse ways that sexual assault can be experienced and responded to by victims and survivors.
• Raise awareness of the importance of affirmative, ongoing sexual consent; shift problematic heterosexual sex scripts that privilege men’s entitlement to sex; challenge attitudes that objectify women; and address the objectification and normalisation of sexual violence in media, video games and pornography.
• Employ safety-by-design principles to enhance the safety of digital spaces and enhance the digital literacy of the community to facilitate recognition and reporting of technology-facilitated abuse and enhance skills for accessing support.
• Harness the role of peer-group support in rejecting stalking and tracking behaviours, whether in person or online, both during relationships and following their conclusion.
• Increase community awareness of trauma-informed, culturally sensitive support services available for victims and survivors. ^[94]

8 Findings: Bystander response

The bystander role is important in the prevention of violence against women. Prosocial bystanders can call out unacceptable behaviour, place social sanctions on perpetrators that discourage future perpetration, help victims and survivors to feel supported and heard, and, in some situations, prevent violence from escalating or even occurring (Bell & Flood, 2020; Orchowski et al., 2018; Palmer et al., 2020). Alternatively, when bystanders choose to do nothing, this can be interpreted by others as approval of, or at least ambivalence towards, the unacceptable behaviour or attitudes (Amar et al., 2015; Baldry & Pagliaro, 2014; Banyard, 2011, 2015; A. D. Berkowitz et al., 2022; Rebollo-Catalan & Mayor-Buzon, 2020). Therefore, bystander response to witnessing violence against women or its precursors plays an important role in either challenging or perpetuating unhelpful social norms (Baillie et al., 2022). However, it is important to note that it is not always safe to act as a prosocial bystander. Sometimes intervention can put the bystander or victim at further risk of harm, or, especially in cases of power imbalance, intervention can be ineffective or have other serious consequences, such as loss of employment. Therefore, bystanders must also assess whether it would be safe to intervene and the best method for doing so.

“Increased community-wide intention to intervene when witnessing disrespect and violence against women” is mentioned in the National Plan 2022–2032 as an early intervention key indicator (COAG, 2022, p. 31).

Prosocial bystander intervention requires the individual to:

• notice the situation as violent or condoning violence
• interpret the event as one requiring intervention or action
• assume responsibility for intervening
• decide upon the method of intervention
• have confidence in their capacity to intervene (Powell, 2014; Taket & Crisp, 2017).

In addition, the bystander also needs to assess that prosocial intervention would be safe given the context of the abuse or disrespect.

The 2021 NCAS bystander items aimed to capture as many of these dimensions as possible to provide guidance to policymakers for the development of bystander interventions.

8.1 2021 NCAS bystander scenarios

Respondents were asked whether they would be bothered by each of three scenarios and those who indicated they would be bothered were then asked how they would react (Box 8-1).

To examine how contextual factors – namely, type of disrespect and power dynamics – influence bystander responses, the scenarios were chosen to vary in
terms of:

• the type of disrespectful behaviour – sexist joke (B1 and B2) versus verbal abuse (B3)
• the relationship of the perpetrator to the bystander – male boss (B2) versus male work friend (B1 and B3).

This chapter presents the 2021 NCAS results regarding bystander response including:

• bystander response to each scenario – whether they would be bothered and whether they would intervene by showing disapproval (Section 8.2)
• the impact of anticipated support or criticism from peers (Section 8.3)
• barriers to bystander intention to intervene (Section 8.4)
• predictors of bystander response (Section 8.5)
• the conclusions and implications arising from these results (Section 8.6).

8.2 Bystander response to each scenario

Most respondents said they would be bothered by each scenario. However, there were significant differences by scenario type. Figure 8-1 shows that while virtually all respondents (99%) said they would be bothered by the verbal abuse scenario, significantly fewer respondents said they would be bothered by the sexist joke scenarios (69–86%). It is particularly notable that almost one in three respondents (31%) said they would not be bothered if a close work friend told a sexist joke (B1).

There were some gender differences in these results. ^[97]  Women were significantly more likely than men to say they would be bothered by a sexist joke told by a friend (B1; 75% versus 55%) or a boss (B2; 91% versus 74%). There was no significant difference by gender for the verbal abuse scenario (B3).

Figure 8-1: Whether respondents would be bothered by scenario, 2021

Note:  N = 4,468 (B1); 4,511 (B2); 4,655 (B3). Asked of one quarter of the sample in 2021.

Figure 8-2 shows whether those who said they would be bothered by the scenarios would intervene by showing their disapproval (immediately in public or later in private) or would not intervene. The results indicate a high level of prosocial bystander intention to intervene. That is, for all three scenarios, most respondents who reported that they would be bothered said that they would show their disapproval either publicly or privately. Specifically,  based only on respondents who would be bothered , the percentage who said they would show their disapproval was 90 per cent for the friend sexist joke scenario, 73 per cent for the boss sexist joke scenario and 94 per cent for the friend verbal abuse scenario. These percentages translate to 59, 63 and 92 per cent of  all respondents indicating they would show their disapproval (in scenario B1, B2 and B3, respectively).

There was also a gender difference in the type of prosocial behaviour for the friend sexist joke scenario (B1). Among those who said they would be bothered, women were significantly more likely than men to say they would disapprove immediately in public (65% versus 47%) and less likely than men to say they would disapprove later in private (26% of women; 42% of men). ^[98]

Figure 8-2: Bystander intention to intervene if bothered by scenario, 2021

Note:  N = 3,188 (B1); 4,113 (B2); 4,623 (B3). Bar figure includes only respondents “bothered” by the sexist joke (B1, B2) or verbal abuse (B3). Pie figures are based on all respondents. Only respondents who indicated they would be “bothered” were asked how they would react. Percentages in the figure do not always add to 100 due to rounding.

Type of disrespectful behaviour and bystander response

As noted above, the likelihood of being bothered varied depending on the type of disrespectful behaviour, with virtually all respondents saying they would be bothered by the verbal abuse scenario (99%; B3), but fewer saying they would be bothered by the two sexist joke scenarios (69–86%; B1 and B2; Figure 8-1).

This result is consistent with the literature which suggests that sexist jokes are often seen as unharmful. Sexist jokes are a type of disparaging humour, where comments are intended to “elicit amusement through the denigration, derogation, or belittlement of a given target” (M. A. Ferguson & Ford, 2008, p. 284; Katz et al., 2019). Further, people are less likely to object to sexually or racially prejudiced comments if they are framed as jokes (Katz et al., 2019). However, jokes of a sexist, racist or homophobic nature can have negative health, academic and social outcomes for the individuals targeted. Negative outcomes can include stress, diminished academic achievement, increased likelihood of dropping out of university, perpetuation of gendered or racialised power hierarchies, inequality and rape myths, and increased rape proclivity among men (M. R. Lowe et al., 2021; Ringblom, 2021; Weber et al., 2020).

Targets of sexist jokes, as well as bystanders, often stay silent because subtle or ambiguous sexist jokes are commonly dismissed as “just a joke”, and those who express offence are often characterised as overly sensitive (Katz et al., 2019; M. R. Lowe et al., 2021; Ringblom, 2021). There is a tendency for people, especially men, to perceive sexist and racist jokes as harmless (M. R. Lowe et al., 2021; Pina & Gannon, 2012). This tendency is consistent with the present result that more men (20%) than women (11%) agreed that “there is no harm in men making sexist jokes about women when they are among their male friends” (G16; Chapter 5). In contrast, very few NCAS respondents (3% of women and 7% of men) thought verbal abuse of a partner was  not a form of domestic violence (D3; Chapter 4). These findings suggest that verbal abuse is considered unacceptable by almost all Australians, whereas sexist humour may still be tolerated within particular domains.

Power imbalance and bystander response

The results in Figure 8-1 and Figure 8-2 also indicate that bystander responses vary depending on the bystander’s relationship to the perpetrator. Significantly more respondents reported that they would be bothered by a sexist joke told by a male boss than a male work friend (Figure 8-1). Further, prosocial bystander intention to show public disapproval was significantly lower when the sexist joke was told by a boss rather than a friend (Figure 8-2). Figure 8-2 shows that 23 per cent of the respondents who would be bothered by the boss sexist joke scenario would be passive bystanders and say nothing, compared to only 3–7 per cent for the two friend scenarios.

These results highlight how bystander behaviour can be affected by the expectations and the power dynamics within relationships. Managers play an important role in the occurrence or deterrence of sexual harassment in workplaces by contributing to team- or organisation-level social norms and expectations around how exclusion and disrespect are responded to, and through implementation of sexual harassment policies and procedures (Perry et al., 2020). When managers tell sexist jokes they are more likely to be interpreted as inappropriate, and can be interpreted as sexual harassment (Ringblom, 2021).

Gender composition of occupation and social network and bystander response

Research suggests that from an early age men learn “masculine” behaviours associated with stereotypical masculine identities such as aggression, competition, domination and control (Corboz et al., 2016; Flood, 2007; Kidd, 2013). Gendered socialisation offers men limited agency in the construction of their masculine identities, with research suggesting that men construct notions of what constitutes a man by learning what a man is not, creating rigid binaries in their understanding of gender in society (E. Anderson, 2008; Berdahl, Cooper, et al., 2018; Gallagher & Parrott, 2011; Nichols, 2018). Rigid masculinity norms also create a “masculine contest culture” that can take hold within institutions and organisations (Berdahl, Cooper, et al., 2018). Defined by rigid masculinity norms of aggression, competition and dominance, some men may feel compelled to behave in accordance with these norms by using defensive tactics to maintain their status in the presence of other men, such as by telling sexist jokes which undermine women (Berdahl, Cooper, et al., 2018; J. Lee, 2018). These imperatives can similarly transfer to social settings such as men-dominated sporting clubs and other social contexts where “lad culture” may be used to defend sexist behaviour and may function as a barrier to bystander intervention intentions (Corboz et al., 2016; Nichols, 2018).

In Australia, only seven of the 19 identifie