Discussion

This paper demonstrates what becomes visible when the NCAS is used not only as a population monitoring instrument, but also as a supplementary diagnostic lens. NCAS scale scores are essential for tracking broad, long-term change. However, prevention activity rarely targets ‘scale scores’ directly. It targets the concrete beliefs and narratives that shape how people interpret harm, whether people who have experienced violence are believed, whether gender inequality is minimised, and whether non-physical abuse is recognised. Item-level secondary analysis helps translate scale-level monitoring into this more applied layer, while remaining grounded in the rigorous and validated NCAS measurement framework (1, 8, 9).

What ‘high impact’ means in this analysis

This analysis uses two complementary lenses that answer different prevention questions:

• High-leverage marker items (RQ1): items with stronger item–scale correlations that best distinguish respondents with higher versus lower overall scale scores. These items help identify ‘marker’ beliefs that are most closely aligned with the underlying construct each scale is designed to measure (i.e., attitudes towards violence against women, gender inequality and understanding of violence).
• High-reach contested beliefs (RQ2): items with higher endorsement, more pronounced response tails, higher non-response or clearer gender differences. These items point to beliefs that are more common, contested or sensitive in the population, and therefore likely candidates for messaging and message-testing. 

A central finding of this paper is that these sets do not perfectly overlap. The items that most strongly differentiate respondents are not always the items most commonly endorsed. Treating prevalence and item–scale alignment as interchangeable can lead to prevention work that either (a) focuses only on what is most common, missing key markers of lower attitudinal rejection; or (b) focuses only on what is most diagnostic, missing beliefs that remain ‘live’ for a substantial minority. Used together, the two lenses provide a clearer basis for prioritising where to focus prevention communications, education and monitoring (1, 8, 9).

Attitudes towards violence against women: Credibility narratives as a priority pressure point

For attitudes towards violence against women, the most discriminating items cluster around sexual violence myths and scepticism about the credibility of people who have experienced violence. These items show strong overall disagreement but nonetheless differentiate respondents strongly within the scale. This indicates that endorsement, where it persists, is concentrated among those with lower attitudinal rejection of violence and aligns with a broader pattern of violence-supportive attitudes (3, 4, 8).

The analysis also shows that some credibility and false-accusation narratives are not only diagnostic but also comparatively common. Around one-third of respondents endorsed beliefs that women exaggerate domestic violence claims in custody disputes (35%) and that sexual assault accusations are commonly used to ‘get back at men’ (30%), with men more likely to agree. This cluster warrants particular attention because it combines relatively high prevalence, gender-patterning and conceptual relevance to responses to disclosure (3, 4).

From a prevention perspective, this pattern suggests that shifting normative ‘credibility climates’ may be as important as shifting explicit approval of violence. Credibility narratives influence whether disclosures are treated as legitimate, whether people who have experienced violence are believed, whether bystanders intervene, and whether institutional responses are supported and trusted (1, 5, 6). Broad anti-violence messaging may not directly address these specific narratives, supporting more targeted message development and testing. The findings support a more specific approach of messaging that directly addresses these narratives, including risks of unintended reinforcement and the need for clear counter-framing (9).

Item-level results also show that sexual violence-related ‘responsibility’ beliefs are not uniform. Some responsibility narratives show low endorsement and little gender difference, while others show higher endorsement, clearer gender gaps and higher non-response. This matters because it indicates that division is concentrated in particular storylines (e.g., ‘led him on’/regret narratives) rather than across all victim-blaming attitudes. This supports precision in messaging, as prevention communications may benefit from testing responses to the specific narrative patterns that remain contested, rather than treating ‘rape myths’ as a single undifferentiated set (3, 4).

Finally, the presence of a highly discriminating item referencing violence against lesbian and bisexual women, strongly rejected overall but still diagnostic, reinforces that where scepticism about sexual violence extends to marginalised groups, it aligns with a wider violence-supportive attitudinal pattern. This supports the ongoing importance of inclusive prevention framing and attention to intersectional vulnerability, without implying that such work is absent in Victoria (13, 14). 

Gender inequality: Two related but distinct prevention tasks 

For attitudes towards gender inequality, the divergence between RQ1 and RQ2 analyses is particularly informative. The most discriminating items are beliefs about men’s greater suitability for authority (i.e., in the workplace, community, politics, household). These items appear to function as markers of more entrenched attitudes endorsing gender inequality and align closely with drivers-focused prevention frameworks that emphasise gendered power and rigid gender roles as foundational to violence against women (1).

By contrast, the most commonly endorsed items relate to minimisation and dismissal of sexism and inequality (e.g., beliefs that women exaggerate inequality, misinterpret innocent remarks as sexist, or do not appreciate what men do). Endorsement remains material (around 30%–38%), with men consistently more likely to agree. This pattern is consistent with the broader evidence that resistance to gender equality can take the form of minimisation, denial and backlash dynamics, even where overt sexism is less socially acceptable (9).

Together, these findings suggest two distinct but connected prevention tasks to tackle attitudes towards gender inequality:

Reduce the reach and perceived ‘normality’ of minimisation and dismissal narratives that remain relatively common and can undermine prevention implementation.

Track and shift authority beliefs that are less widely endorsed but strongly diagnostic of entrenched inequality and therefore important for prevention segmentation, monitoring and longer-term norm change.

In practical terms, this generates different hypotheses for intervention design, including appropriate settings to target each theme. Minimisation/dismissal narratives may be appropriate testing targets for broader public communications and community norm-setting. Authority beliefs may be better examined through more intensive, settings-based approaches (including workplaces, leadership contexts and relationship education) that can work directly with entitlement, decision-making and expectations of gendered authority (1, 9). 

Understanding of violence against women: High recognition, less consistent at the margins

For understanding violence against women, recognition of non-physical abuse is high overall, consistent with contemporary understandings of domestic and family violence as patterns of coercive control rather than isolated physical incidents (10–12). Item-level results nonetheless show uneven recognition for specific forms of non-physical control and surveillance, particularly economic restriction, technology-facilitated monitoring, religious control and identity-based coercion, where a meaningful minority responded ‘no’ or ‘yes, sometimes’, and gender differences are clearer on a subset of items (5, 6, 10–12).

For prevention, the implication is not that Victorians ‘do not recognise coercive control’. Rather, particular forms of non-physical control appear easier to minimise, normalise or reinterpret. This indicates a continued role for settings-based reinforcement and professional education (e.g., clarifying what economic abuse looks like in practice, how technology-facilitated monitoring operates, and how identity-based coercion functions as control) (10–12).

The elevated ‘not stated’ response for the trans-identity-based coercion item warrants explicit mention as a finding in itself. Non-response may reflect discomfort, uncertainty or reluctance to engage with sensitive topics in a phone survey. Treating non-response as informative can support careful framing and inclusive prevention design, while avoiding over-interpretation about the direction or intensity of underlying attitudes (8, 13).

Implications for monitoring, messaging and future use of NCAS

A practical implication of this work is that item-level analysis can help translate NCAS results into clearer priorities for messaging and program refinement. The strongest ‘pressure points’ in this analysis sit at the level of specific narratives rather than broad constructs, particularly credibility and false-accusation narratives within violence-related attitudes and minimisation/dismissal narratives within gender inequality. These findings support structured message testing that focuses on clarity, resonance and unintended effects, consistent with evidence-informed primary prevention (1, 3, 4, 9).

Item-level analysis can also strengthen interpretation of population trends. If the most prevalent items and the most diagnostic items shift at different rates (or within different subgroups), scale averages can appear stable even when meaningful movement is occurring ‘under the surface’. Building a small, stable set of item-level indicators into routine reporting alongside scale scores could therefore complement NCAS monitoring and support more precise prevention learning and evaluation over time (8, 9). 

Limitations

This exploratory cross-sectional analysis describes responses to the 2021 NCAS and does not identify causal mechanisms, intervention effects, behavioural outcomes or change over time. The findings are a point-in-time snapshot and should not be treated as estimates of current Victorian attitudes. Data collection occurred during the COVID-19 period, which may have affected participation, social context and responses. Survey coverage, participation and item non-response, telephone administration, social desirability and topic sensitivity may also have influenced the results. Some items were administered to smaller subsamples, and gender comparisons are constrained by the categories and sample sizes available; combining women and respondents of other genders does not permit conclusions about the experiences or attitudes of individual gender-diverse groups. Item–scale correlations indicate alignment with the relevant scale, rather than causal leverage or formal evidence of intervention importance. The findings are therefore best used to guide prioritisation, message testing, monitoring and further analysis (8).