Theoretical background

The first stage of developing a conceptual model of associations with violence-supportive attitudes was to map relevant concepts and previous research onto the NCAS scales and subscales. This section provides the rationale for including the NCAS scales in the model and sets out the research questions and the theoretically proposed directions of association.

Violence-supportive attitudes

The NCAS scale that measures violence-supportive attitudes is the Attitudes towards Violence against Women Scale (AVAWS). Widespread adoption of attitudes that accept, endorse or condone violence against women have been associated with higher perpetration of violence (4). One review found that communities with higher levels of justification of intimate partner violence also have higher rates of prevalence. The review authors conclude that potential perpetrators are more likely to feel justified in committing violence when it is socially condoned (5).  

Extant evidence suggests that while gender norms and inequality do not solely determine the prevalence of violence, they ‘operate through long causal chains, interacting dynamically and in complex ways across the social ecology’ (6). To the extent that attitudes are representations that maintain gendered relations, when individuals internalise and enact their own attitudes and those of their community (7), challenging harmful attitudes is a key strategy in the prevention of violence against women.  

Primary prevention initiatives can target attitudes that normalise or justify violence to dismantle the social acceptability of such behaviour. Public education campaigns, community-based interventions and school programs that challenge these attitudes operate through a theory of change in which reducing social tolerance for violence contributes to prevention alongside changes to social norms, structures, behaviours and material conditions. Tracking attitudes over time with surveys such as the NCAS enables policymakers to monitor population-level change and make data-informed decisions about how to strengthen interventions (1, 2).

The established relationship between violence-supportive attitudes and perpetration supports the inclusion of AVAWS as the attitudinal outcome in the conceptual model. AVAWS does not measure violence perpetration, victimisation or prevalence. Examining relationships among the measured attitudes may shed light on potential pathways for holistic attitude change.

Attitudes towards gender inequality

The NCAS scale that measures attitudes endorsing gender inequality is the Attitudes towards Gender Inequality Scale (AGIS). Previous studies, including those using NCAS data, have found a strong association between attitudes endorsing gender inequality and attitudes towards violence against women (i.e., AVAWS), above demographic factors such as age or education (2, 8). This supports the inclusion of the AGIS as the main predictor of AVAWS scores in the current statistical model and is consistent with primary prevention frameworks that position gender inequality as a key driver of violence against women. The present analysis tests associations with violence-supportive attitudes; it does not test whether changing AGIS scores will reduce the prevalence of violence. Efforts to shift beliefs around gender roles and inequality may nevertheless contribute to reducing attitudes that endorse violence.

Over the past four decades, beliefs that overtly blame women for violence have become less publicly acceptable, as societal awareness of the issue has grown (4). However, more subtle forms of gender bias persist, often manifesting through myths about violence and ambivalent sexist attitudes. These myths are closely related to the endorsement of rigid gender stereotypes, reinforcing harmful beliefs about the roles of men and women in society, and legitimising violence against women in certain circumstances (6). Primary prevention efforts therefore must not only target violence-supportive attitudes but also address underlying (and sometimes less blatant) attitudes that justify violence.  

• Research Question 1: Do attitudes towards gender inequality (AGIS) predict attitudes towards violence against women (AVAWS)?

Dimensions of attitudes towards gender inequality

Previous research has pinpointed specific dimensions of gender inequality that contribute to persistent violence-supportive attitudes (7). Regression analysis of NCAS data at the national level has found those who support limiting women’s autonomy in private are more likely to endorse the use of violence against women (7). Attitudes that support men’s dominance within intimate relationships, including the belief that men have the right to enforce their dominance, legitimise violence against women and thereby perpetuate it (9).  

This previous research highlights the need for separate analyses of attitudes toward gender inequality in (a) public and (b) private domains. Doing so enables research to better understand the distinct influences from each domain on attitudes supportive of violence against women. Examining how subscales of the AGIS differentially predict violence-supportive attitudes can identify specific attitudes that underpin support for violence in the Victorian context.  

While some individuals may support gender equality in the public domain – such as endorsing women’s participation in politics or the workforce – they may simultaneously reject gender equality within the private domain, believing that men should maintain control over household and intimate relationships. Understanding this discrepancy requires a multidimensional approach to gender equality that distinguishes between public and private domains.  

Research suggests that individuals who endorse traditional gender roles in the private sphere are more likely to support violence against women, even if they support gender equality in public life (9). For instance, some studies indicate that women with higher economic status relative to their partners may be at even greater risk of intimate partner violence, particularly if their partners hold traditional beliefs about gender roles (9). This backlash effect underscores the complexity of gender relations, where increased gender equality can provoke negative responses. Therefore, this analysis includes attitudes towards gender inequality in both the public and private spheres.  

• Research Question 1a: Will subscales of AGIS (Reinforce Gender Roles; Undermine Leadership; Limit Autonomy; Normalise Sexism; Deny Inequality) differentially predict AVAWS?

Gender difference in violence-supportive attitudes

Gender has been found to predict violence-supportive attitudes, with men more likely than women to agree with myths and beliefs that minimise or justify violence against women (3). This includes perceiving a narrower range of behaviours as violent, showing less empathy for people who have experienced violence, blaming them more readily, and downplaying the seriousness of physical and sexual assault. Men are also less likely to see behaviours constituting violence against women as serious, inappropriate or damaging, reflecting how traditional gender-role attitudes shape perceptions of violence (3).  

Other studies challenge the assumption that gender alone is the strongest predictor of violence-supportive attitudes. For example, the correlation between being male and scoring highly on the VSA-18 Scale, which measures attitudes supportive of violence against women, was found in one study to be weaker than the correlation with other factors, such as education, age and occupation (10).  

However, these studies analyse the unique statistical association of gender with violence-supportive attitudes, rather than whether gender changes the strength of the relationship between patriarchal attitudes and violence-supportive attitudes. This paper therefore tests gender as a moderator of the AGIS–AVAWS association.

Gender is included as a moderating variable in the conceptual model, proposing that it may change the strength of the relationship between attitudes towards gender inequality and attitudes towards violence against women. Understanding whether this association differs by gender may inform hypotheses for cohort-based interventions.

• Research Question 1b: Does gender (identifying as a man compared to identifying as a woman or non-binary person) moderate the relationship between AGIS and AVAWS?

Community understanding and recognition of violence against women

On the basis of this theoretical argument, and because community education to increase recognition of violence is a core component of primary prevention, the Understanding of Violence against Women Scale (UVAWS) is included as a central variable in the conceptual model. Specifically, the model tests whether community understanding accounts for part of the association between patriarchal attitudes and violence-supportive attitudes.

Recognising intimate partner violence as a form of abuse is crucial, not only for people who have experienced violence to seek help and perpetrators to engage in treatment, but also for creating a community capable of effectively responding to and preventing further violence (8). ‘Community understanding’ can constitute recognition of specific behaviours that constitute domestic violence, sexual violence and technology-facilitated abuse, along with understanding of how violence is shaped by contextual factors such as gender (2).

There is limited research directly examining how community understanding relates to violence-supportive attitudes, but a strong theoretical rationale for including it. We hypothesise that lower understanding will be associated with a stronger relationship between patriarchal attitudes and violence-supportive attitudes, while higher understanding will be associated with attenuation of that relationship. Put another way, recognition and understanding may influence whether violence is condemned, minimised or excused.

• Research Question 2: Does understanding of violence against women and its gendered nature (UVAWS) mediate the relationship between AGIS and AVAWS?

Denial of violence against women in one’s own community (‘distancing’)

The 2021 NCAS found that 25.6% of Victorians deny that violence is a problem in the suburb or town where they live, but only 5.4% deny that violence is a problem in Australia (2). This gap is consistent with the possibility that some Victorians see violence as a national problem but as less relevant to their own communities.

When violence occurs within their own communities, individuals are more likely to explain it through external factors, such as personal circumstances, rather than acknowledging it as a widespread or systemic issue. In contrast, violence occurring in outgroups is more likely to be attributed to internal, cultural flaws (11). This dynamic is evident when individuals from privileged, often white or economically stable communities, may believe that ‘people like us’ do not perpetrate violence, and thus, it is not a problem in their community.

Psychological distance from violence against women may be associated with violence-supportive attitudes. When individuals view violence as a problem that occurs elsewhere but not in their own communities, the problem is externalised and creates distance from both violence and action against it. One study found that when people view a person experiencing intimate partner violence as less ‘human’ (manipulated experimentally by describing the person who has experienced violence as unfaithful), and measured on a scale of perceived humanness, they are less likely to offer help and support to that person (12). Similarly, perpetrators often employ psychological processes such as denial of responsibility and victim-blaming, which are reinforced by cultural discourses that shift blame away from men, particularly within patriarchal frameworks (13).

In a broader context, cultural framings of violence against women often characterise perpetrators as ‘outsiders’ (those from foreign cultures) and nonviolent ‘insiders’ (those within the dominant cultural group). For instance, in analyses of European Union policies, violence is frequently depicted as an issue stemming from immigrant or ethnic minority communities, neglecting the role of structural gender inequality in perpetuating violence within the majority culture (14). This framing marginalises vulnerable groups, particularly ethnic minority and migrant women, and undermines the seriousness of forms of violence that occur within majority populations (14).

The process of distancing is not limited to cultural or ethnic dimensions. Discussions about rural intimate partner violence, for instance, often reinforce stereotypes of rural communities as deviant or dangerous, which can further marginalise rural people experiencing partner violence (15). Such narratives suggest that violence is an issue of ‘other’ people or places, reinforcing the belief that violence is absent from one’s own community. Media representations also contribute to this dynamic, often framing domestic violence in Western countries as isolated incidents, while attributing violence in non-Western societies to cultural causes, particularly in Islamic communities (16).

Distancing attitudes may also arise from narrow definitions of violence. Individuals who define violence strictly in terms of physical harm may be unaware of or dismiss the prevalence of other forms of abuse, such as emotional, economic or psychological violence. This could contribute to a perception that violence is not occurring locally, reinforcing complacency and inaction. The conceptual model therefore tests the association between community understanding and Distancing, and whether Distancing accounts for part of the AGIS–AVAWS association.

• Research Question 3: Does believing violence against women is not a problem where you live (Distancing) mediate the relationship between AGIS and AVAWS?  
• Research Question 4: Are UVAWS and Distancing associated with each other?  

Relationship between attitudes

This paper examines theoretically proposed pathways associated with violence-supportive attitudes. It is expected that patriarchal attitudes – a cluster of attitudes including upholding rigid gender roles and norms, undermining the public leadership of women, limiting women’s autonomy in private, normalising sexism and denying inequality – will have the strongest association with violence-supportive attitudes that minimise violence against women, mistrust women’s accounts of experiencing violence, and objectify women. It is also expected that community understanding of violence – including recognition of domestic violence, stalking and technology-facilitated abuse, and understanding of the gendered nature of violence – will account for part of the AGIS–AVAWS relationship. Higher levels of community understanding and recognition of violence are expected to be associated with a weaker relationship between patriarchal attitudes and violence-supportive attitudes. These hypotheses concern associations between attitude measures, not causal effects on violence.

• Research Question 5: How much variation in attitudes towards violence against women is accounted for by the full structural equation model?

Structural equation modelling

Structural equation modelling (SEM) is a multivariate statistical analysis technique that tests multiple structural relationships simultaneously. It allows direct associations, interactions and indirect pathways to be estimated within a single model.

SEM is used here to test a theoretically specified directional model of how attitudes supporting gender inequality relate to violence-supportive attitudes, including indirect pathways via violence understanding and local acknowledgement. Because the data are cross-sectional and observational, the model cannot establish temporal order or fully rule out unmeasured confounding, and causal interpretations are treated as provisional and consistent with, rather than definitive proof of, the proposed pathways. All models incorporated NCAS survey weights. AGIS, AVAWS and UVAWS were entered as observed composite scale scores rather than estimated as latent variables.

Variable selection  

Across the NCAS scales used here, higher scores indicate less endorsement of harmful attitudes: higher AVAWS scores indicate less violence-supportive attitudes, and higher AGIS scores indicate less endorsement of gender inequality (more gender-equal attitudes). Higher UVAWS scores indicate stronger recognition and understanding of violence against women.

There are five subscales underlying AGIS:

• Reinforce Gender Roles  
• Undermine Leadership 
• Limit Autonomy 
• Normalise Sexism 
• Deny Inequality.

The Limit Autonomy subscale pertains to the private sphere and Undermining Leadership the public sphere – these act as proxy variables for public/private inequality.

The AVAWS was selected as the dependent variable to examine which measured attitudes are associated with higher or lower AVAWS scores and identify possible pathways for further testing in prevention research. There are three main subscales underlying AVAWS:

• Minimise Violence 
• Mistrust Women 
• Objectify Women.

Within AVAWS, items can pertain to either domestic violence, sexual violence (including sexual assault and sexual harassment) or technology-facilitated abuse.

The UVAWS measures awareness and understanding of what constitutes violence against women – including physical, emotional and financial abuse – and understanding of the gendered nature of violence.  

An attitude reflecting ‘distancing’ is measured by a single item that assesses the extent to which someone believes violence is a problem where they live. The Distancing variable was reverse-coded so higher scores indicate greater acknowledgement that violence is a local problem (that is, lower psychological distance).

The existing scales, subscales and the items comprising them have been maintained in this analysis as they have been validated by ANROWS.