RQ1: Which specific attitude items most strongly differentiate overall scale scores in Victoria?
This section identifies the items within each NCAS scale that are most strongly associated with overall scale scores. Item–scale correlations (r) indicate how strongly responses to a specific item align with the total scale score. Higher absolute correlations indicate items that are more discriminating within the scale.
Attitudes towards violence against women scale (AVAWS)
Table 1 presents the 10 AVAWS items with the strongest item–scale correlations (r = 0.85–0.90). These items cluster around sexual violence myths, victim-blaming, minimisation, and mistrust of victim-survivors’ credibility.
Table 1. Ten AVAWS items most strongly correlated with overall scale scores
| AVAWS subscale | Item (1 = strongly agree; 5 = strongly disagree) | r-value | Mean | SD |
|---|---|---|---|---|
| Mistrust Women | Women who wait weeks or months to report sexual assault are probably lying. | 0.90 | 4.69 | 0.77 |
| Objectify Women | 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. | 0.89 | 4.76 | 0.80 |
| Mistrust Women | If a woman claims to have been sexually assaulted but has no other physical injuries, she probably shouldn’t be taken too seriously. | 0.89 | 4.81 | 0.62 |
| Objectify Women | Women who flirt all the time are somewhat to blame if their partner gets jealous and hits them. | 0.89 | 4.60 | 0.95 |
| Mistrust Women | When lesbian or bisexual women claim to have been sexually assaulted by their partner, they probably shouldn’t be taken too seriously. | 0.88 | 4.85 | 0.50 |
| Mistrust Women | Women who wait weeks or months to report sexual harassment are probably lying. | 0.88 | 4.69 | 0.79 |
| Excuse and Minimise | It’s only really stalking if it’s by a stranger. | 0.87 | 4.83 | 0.65 |
| Objectify Women | If a woman is raped while drunk or affected by drugs, she is at least partly responsible. | 0.87 | 4.63 | 0.95 |
| Mistrust Women | A lot of times, women who say they were raped had led the man on and then had regrets. | 0.87 | 4.12 | 1.23 |
| Excuse and Minimise | Women who stay in abusive relationships deserve less help from counselling and support services than women who leave their abusive partner. | 0.85 | 4.76 | 0.79 |
Across these items, average responses indicate strong disagreement overall (i.e., means above 4.0), alongside meaningful dispersion for a small subset (e.g., ‘led the man on …’ SD = 1.23). The correlation results indicate that endorsement of sexual violence myths and credibility-related beliefs, where it persists, is concentrated among respondents with lower overall AVAWS scores (lower attitudinal rejection) in the Victorian sample.
One item referencing lesbian and bisexual women shows strong rejection (M = 4.85) while also being strongly associated with the overall scale score (r = 0.88). This suggests that scepticism about sexual violence affecting marginalised groups, where it occurs, aligns with broader patterns of more violence-supportive attitudes.
Attitudes towards Gender Inequality Scale (AGIS)
Table 2 presents the five AGIS items with the strongest item–scale correlations (r = 0.86–0.88). These items centre on beliefs about leadership, competence and control in relationships.
Table 2. Five AGIS items most strongly correlated with overall scale scores
| AGIS subscale | Item (1 = strongly agree; 5 = strongly disagree) | r-value | Mean | SD |
|---|---|---|---|---|
| Undermine Leadership | In the workplace, men generally make more capable bosses than women. | 0.88 | 4.54 | 0.96 |
| Undermine Leadership | Men, rather than women, should hold positions of responsibility in the community. | 0.87 | 4.62 | 0.90 |
| Undermine Leadership | On the whole, men make better political leaders than women. | 0.87 | 4.57 | 0.93 |
| Undermine Leadership | Women are less capable than men of thinking logically. | 0.86 | 4.79 | 0.70 |
| Limit Autonomy | Men should take control in relationships and be the head of the household. | 0.86 | 4.60 | 0.91 |
As with the AVAWS, mean scores indicate strong disagreement overall. The item–scale pattern indicates that beliefs about men’s greater suitability for authority and decision-making are among the most discriminating indicators of higher gender-inequality attitudes within the Victorian sample, particularly in workplace leadership and relationship decision-making.
Understanding of Violence Against Women Scale (UVAWS)
UVAWS items are scored in the opposite direction to the attitudinal scales; correlations are therefore negative in this output (higher recognition aligns with higher overall understanding). Table 3 presents the five UVAWS items most strongly correlated with overall scale scores (r = −0.79 to −0.81).
Table 3. Five UVAWS items most strongly correlated with overall scale scores
| UVAWS subscale | UVAWS item | r-value | Mean | SD |
|---|---|---|---|---|
| Recognise Domestic Violence | Forces partner to stop practising their religion | −0.81 | 1.53 | 0.93 |
| Recognise Domestic Violence | Controls partner by denying them access to shared money | −0.80 | 1.49 | 0.89 |
| Recognise Domestic Violence | Controls partner’s social life by preventing contact with family/friends | −0.79 | 1.37 | 0.79 |
| Recognise Domestic Violence | Repeatedly tracks location, calls or activities through devices without consent | −0.80 | 1.46 | 0.88 |
| Understand Gendered Violence | Tries to control partner by forcing them to hide that they are transgender | −0.81 | 1.49 | 0.90 |
These items relate primarily to non-physical control and surveillance behaviours, including economic restriction, social isolation, religious control and technology-facilitated monitoring. Low mean values (M = 1.37–1.53) indicate most respondents recognised these behaviours as violence (i.e., ‘yes, always’/’yes, usually’), while the correlations indicate these items still differentiate respondents with higher versus lower overall understanding. These behaviours align with contemporary definitions of non-physical domestic and family violence, including coercive control and technology-facilitated abuse (10–12).
RQ1 results synthesis
Across scales, the strongest item–scale associations cluster around three domains:
- sexual violence myths and scepticism about credibility of people who have experienced violence (AVAWS)
- beliefs about men’s greater suitability for leadership and authority (AGIS)
- recognition of non-physical control and surveillance behaviours (UVAWS).
These patterns identify which specific attitudinal statements most strongly differentiate higher versus lower overall scale scores in Victoria. However, items that are most discriminating within a scale are not necessarily the most widely endorsed. The next section examines how these beliefs are distributed across the Victorian population, and where gender differences and contested response patterns are concentrated.
RQ2: How are Victorians divided on specific attitude items, and which items show the strongest gender differences?
This section examines item-level distributions to identify where attitudes are: (a) more commonly endorsed, (b) more contested (e.g., visible in ‘tails’ such as ‘strongly agree’/’strongly disagree’ and in non-response), and (c) more gender-patterned. For AVAWS and AGIS, ‘agree’ is reported as weighted ‘strongly agree’ + ‘somewhat agree’ (SA + A). For UVAWS, endorsement is reported as recognition (‘yes, always’ + ‘yes, usually’).
Where tables report ‘women/other’, this combines women and respondents of other genders due to small base sizes of non-cis or non-binary gender identities. Interpret differences within non-binary genders with appropriate caution (8).
Attitudes towards Violence against Women (AVAWS)
Table 4 presents the AVAWS items with the highest agreement rates in Victoria, alongside gender differences and response ‘tails’ (strong agreement and strong disagreement). For the most endorsed items, endorsement is concentrated in credibility and false accusation narratives. These items also show the largest gender gaps, with men more likely to agree. By contrast, several victim-blaming items show low endorsement and relatively little gender difference, indicating cross-gender convergence on those statements.
Table 4. AVAWS items with highest agreement: Endorsement, tails and gender gaps (weighted %, SA + A)
| AVAWS item | Agree Overall % (SA + A) | Men % | Women/ other % | Gap pp (men − women) | Strongly agree % | Strongly disagree % | Not stated % |
|---|---|---|---|---|---|---|---|
| Women going through custody battles often make up or exaggerate claims of domestic violence. | 35.2 | 39.2 | 31.4 | 7.8 | 8.8 | 23.4 | 11.2 |
| It is common for sexual assault accusations to be used as a way of getting back at men. | 30.3 | 34.9 | 26.0 | 8.9 | 6.2 | 37.9 | 7.3 |
| When a man is very sexually aroused, he may not even realise that the woman doesn’t want to have sex. | 24.3 | 24.8 | 23.9 | 0.9 | 10.5 | 60.3 | 6.2 |
| Women who wait weeks or months to report sexual harassment are probably lying. | 6.3 | 8.4 | 4.4 | 4.0 | 2.0 | 75.9 | 2.4 |
| 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. | 7.0 | 7.1 | 6.9 | 0.2 | 3.0 | 85.6 | 1.1 |
| If a woman claims to have been sexually assaulted but has no physical injuries, she probably shouldn’t be taken too seriously. | 3.8 | 4.6 | 3.1 | 1.4 | 1.3 | 84.7 | 1.9 |
| Note: Percentages are weighted. SA = strongly agree; A = agree; pp = percentage points. Gap = men − women/non-binary respondents. Women and non-binary respondents were combined due to small sample sizes. | |||||||
Table 5 illustrates how two ‘responsibility’ narratives in sexual violence differ in prevalence, gender patterning and non-response, even though both sit within the broader cluster of sexual violence myths identified in RQ1.
Table 5. Gendered agreement with two items on responsibility for sexual violence (weighted %, SA + A)
| AVAWS item | Agree Overall (%) | Men (%) | Women /other (%) | Gender gap (pp) |
|---|---|---|---|---|
| If a woman is raped while drunk or affected by drugs, she is at least partly responsible. (SV3k) | 10.4 | 9.9 | 10.8 | −0.9 |
| A lot of times, women who say they were raped had led the man on and then had regrets. (SV3l) | 21.1 | 23.5 | 18.7 | 4.8 |
| Note: Percentages are weighted. SA = strongly agree; A = agree; pp = percentage points. Gap = men − women/non-binary respondents. Women and non-binary respondents were combined due to small sample sizes. | ||||
These items show different profiles. The intoxication-responsibility item has lower endorsement and little gender difference. The ‘led him on’/regret item has higher endorsement and a moderate gender gap, with men more likely to agree. A relatively high ‘not stated’ rate for SV3l indicates higher non-response and should be noted when interpreting results.
Attitudes towards Gender Inequality (AGIS)
Table 6 presents the AGIS items with the highest agreement rates in Victoria. These items show majority disagreement overall, alongside a sizeable minority endorsement, with men consistently higher than women/other genders.
Table 6. AGIS items with highest endorsement, tails and gender gaps (weighted %,SA + A)
| AGIS item | Agree Overall % (SA + A) | Men % | Women/ other % | Gap pp (men − women/ other) | Strongly agree % | Strongly disagree % | Not stated % |
|---|---|---|---|---|---|---|---|
| Many women don’t fully appreciate all that men do for them. | 29.6 | 33.6 | 25.8 | 7.8 | 6.9 | 38.6 | 3.9 |
| Many women exaggerate how unequally women are treated in Australia. | 33.4 | 39.1 | 27.9 | 11.2 | 9.6 | 34.5 | 4.9 |
| Many women mistakenly interpret innocent remarks or acts as being sexist. | 38.1 | 42.7 | 33.8 | 8.9 | 7.7 | 24.0 | 6.4 |
| Note: Percentages are weighted. SA = strongly agree; A = agree; pp = percentage points. Gap = men − women/non-binary respondents. Women and non-binary respondents were combined due to small sample sizes. | |||||||
These items reflect a pattern of minimisation and dismissal of women’s experiences of inequality. While most respondents disagree, endorsement remains material (30%–38%), and gender gaps suggest these beliefs are more prevalent among men. This aligns with broader evidence that resistance to gender equality can take the form of minimisation, denial and backlash, even where explicit sexism is less socially acceptable (9).
Notably, the most prevalent AGIS items in RQ2 relate to minimisation/dismissal of sexism, whereas the most discriminating AGIS items in RQ1 relate to beliefs about men’s greater suitability for leadership and authority.
Understanding of Violence against Women Scale (UVAWS)
This section examines how Victorians’ recognition of specific behaviours as violence is distributed across the population. For the UVAWS, endorsement is not ‘agreement’ but recognition. Results are reported as the weighted proportion responding ‘yes, always’ or ‘yes, usually’ (combined as ‘recognised’), alongside ‘yes, sometimes’, ‘no’ and ‘not stated’. Note that some DVInt items were asked of a subsample (~1,532), while most other items were asked of the full Victorian sample (~6,083). Percentages are weighted and comparable as proportions, but base sizes differ.
Table 7. UVAWS items showing the highest division and gender difference (weighted %)
| UVAWS item (short label) | Recognised Overall % | Men % | Women/other % | Gap pp | Yes, always Overall % | Yes, sometimes Overall % | No Overall % | Not stated Overall % |
|---|---|---|---|---|---|---|---|---|
| Forces partner to hide they are transgender (DVInt5) | 78.9 | 74.0 | 83.5 | 9.5 | 67.2 | 8.2 | 7.3 | 5.7 |
| Forces partner to stop practising their religion (DVInt4) | 79.8 | 76.4 | 82.8 | 6.4 | 67.0 | 9.9 | 8.2 | 2.1 |
| Controls partner by denying access to shared money (DV2m) | 80.9 | 75.0 | 86.5 | 11.5 | 67.5 | 9.9 | 7.6 | 1.6 |
| Unwanted picture of genitals sent to a woman (SVK1) | 81.3 | 79.9 | 82.6 | 2.7 | 70.2 | 8.8 | 7.7 | 2.2 |
| Tracks location/calls/activities via devices without consent (DV10) | 82.8 | 77.2 | 88.1 | 10.9 | 70.7 | 8.6 | 7.6 | 1.0 |
| Prevents contact with family/friends (DV2k) | 87.0 | 81.6 | 92.1 | 10.5 | 74.7 | 6.9 | 5.5 | 0.7 |
| Note: Percentages are weighted. SA = strongly agree; A = agree; pp = percentage points. Gap = men − women/non-binary respondents. Women and non-binary respondents were combined due to small sample sizes. Recognised = ‘yes, always’ + ‘yes, usually’. | ||||||||
Overall, recognition was high for all items presented here. However, there is still a measurable minority who either did not recognise particular behaviours as violence (responded ‘no’) or only recognised them inconsistently (‘yes, sometimes’). Items with higher ‘no’ responses in this set included denial of access to shared money, device-based monitoring without consent, forcing a partner to hide their transgender identity, and forcing a partner to stop practising their religion. The presence of both ‘no’ and ‘yes, sometimes’ responses suggests that there is uneven recognition for certain forms of economic abuse, technology-facilitated monitoring and identity-based control across Victorian communities (10–12).
Gender gaps were generally modest, but clearer differences appear for a subset of items. Women/other genders were more likely than men to recognise device-based monitoring without consent and denial of access to shared money as violence.
One item in the UVAWS set had a notably higher ‘not stated’ rate: forcing a partner to hide that they are transgender (‘not stated’ = 5.7%). This suggests a higher level of non-response for this topic and should be noted when interpreting gender differences and overall recognition for that item.
RQ2 synthesis
Across the AVAWS and the AGIS, the clearest divisions are concentrated in (a) sexual violence credibility and accusation narratives, and (b) minimisation and dismissal of gender inequality. In the UVAWS, recognition is high overall, but there is still uneven recognition for specific forms of non-physical abuse, including economic abuse, technology-facilitated monitoring and identity-based control (10–12). Across all three domains, gender differences are concentrated in a subset of items rather than evenly distributed, supporting the value of item-level analysis for identifying where population averages can mask meaningful divisions.