Governments count reported violence, prosecutions and deaths. A new generation of research is asking a more difficult question: can women’s safety be measured before violence occurs?

For decades, one of the most persistent problems in violence-against-women policy has also been one of the most basic: measurement.

• How many women experience violence?
• How many report it?
• How many cases reach police?
• How many result in prosecution?
• How many result in conviction?

These figures matter.

Institutum Lex Feminae itself uses several of them when examining the distance between legal protection and enforcement.But there is a methodological problem hidden inside them.

Most become visible only after something has already gone wrong.A police report measures contact with the criminal justice system.

A prosecution measures a prosecutorial decision. A conviction measures the outcome of a case that survived numerous earlier stages.

Even prevalence surveys, while indispensable, principally tell us about violence that women have already experienced.

None, by themselves, answer a different question:How safe were women before the violence occurred?

Measuring Violence Is Not the Same as Measuring Safety

That distinction is increasingly important. In June 2026, researchers from the ARC Centre of Excellence for the Elimination of Violence Against Women published a new cross-national Women’s Safety Index in Data & Policy.

Its premise challenges the conventional measurement model. Instead of treating recorded incidents of violence as the primary measure of women’s safety, the researchers examine the structural conditions associated with women’s vulnerability to violence.

The framework organizes those foundations around three dimensions: Equity. Protection. Resources.

The significance of the approach is methodological.

Violence is an outcome. Safety is a condition. Those two concepts overlap, but they are not interchangeable.

A country can record comparatively low levels of reported violence while women encounter significant barriers to reporting.

Another can record increasing reports precisely because reporting systems, institutional trust or access to services have improved.

Recorded violence can therefore increase while certain dimensions of women’s protection improve.

Conversely, low recorded violence cannot automatically be interpreted as evidence of high safety.

That creates a serious problem for comparative policy analysis.

The Missing Women in Existing Indicators

International indices have become extraordinarily influential in public policy.Governments are ranked on gender gaps, development, peace, democracy, economic participation and institutional performance.

These instruments are valuable because they allow complex conditions to be converted into comparable indicators.But every index is also a decision about what counts.

The researchers behind the Women’s Safety Index identify a striking gap.

Some of the world’s major gender-equality indices do not directly measure violence against women or women’s safety.

Meanwhile, major conflict and peace datasets can measure warfare, deaths and political instability without adequately capturing forms of gender-based violence that occur alongside them.

A country can therefore be measured for peace without fully measuring women’s insecurity.

It can be measured for gender equality without directly measuring women’s exposure to violence.That is not merely a statistical curiosity.

It affects what policymakers see.

And what policymakers cannot see is considerably harder to govern.

The Reporting Paradox

There is another reason incident data require caution: reporting itself is a variable.Consider two hypothetical countries.

Country A records 10,000 domestic-violence complaints. Country B records 4,000.

Which country is safer for women?

The answer cannot be determined from those numbers alone.

Country A may have more violence. Or it may have better reporting mechanisms, greater institutional trust, stronger victim services and fewer social barriers preventing women from approaching authorities.

Country B may genuinely have less violence. Or thousands of incidents may simply remain outside official systems.

The same problem appears when prosecution and conviction statistics are interpreted without examining the stages preceding them.

A high conviction rate among prosecuted cases does not necessarily demonstrate that a system effectively responds to violence if only a small proportion of cases ever reach prosecution.

Likewise, an increase in police reports does not necessarily establish that violence itself has increased. The denominator matters. So does institutional context.

Safety Begins Before the Police Report

The Women’s Safety Index attempts to move the analytical starting point backwards. Rather than asking only what happened after violence occurred, it asks whether the surrounding environment contains conditions associated with protection from violence.

This approach has particularly important implications during conflict and disaster.Crises can disrupt income, housing, public services, legal institutions, transportation, community networks and access to support.

Those disruptions may increase women’s vulnerability before they appear in conventional crime statistics. Measuring the foundations of safety therefore creates the possibility of identifying deterioration earlier.

That changes the policy question from: How much violence occurred? to:

Are the conditions that protect women becoming stronger or weaker?

For prevention policy, that is a fundamentally different question.

No Single Number Can Measure Women’s Safety

Composite indices are not without limitations. Transforming complex social conditions into a single score requires choices about indicators, weighting, data availability and comparability.

Missing gender data remain a significant international problem. National averages can also conceal enormous differences between women.

Income, age, disability, migration status, geography and other circumstances can materially alter exposure to risk and access to protection within the same country.

A national safety score should therefore never become a substitute for disaggregated analysis.

Nor should structural measures replace prevalence surveys, police data, prosecution statistics or victimization research. The better approach is layered measurement.

Prevalence tells us about exposure. Reporting tells us about institutional contact.

Prosecution tells us about enforcement. Conviction tells us about adjudication. Structural indicators tell us something different: whether the environment surrounding women contains the foundations associated with safety.

The mistake is expecting any one of those measures to answer all five questions

From Counting Harm to Measuring Prevention

This distinction has implications well beyond academic methodology.

Governments frequently describe violence-against-women policy through inputs and outputs.

A law was enacted. A hotline was created.

Funding increased.Police received training.

Shelter capacity expanded.

More perpetrators were prosecuted.

Each may represent meaningful institutional action. None independently establishes that women became safer. A prevention framework ultimately requires another layer of evidence: whether the conditions associated with violence changed. That means measuring not merely government activity, but exposure, accessibility, institutional response and outcomes over time.

The emerging research on women’s safety therefore points toward a broader principle for gender policy: The absence of recorded violence is not necessarily evidence of safety.And the presence of legislation is not necessarily evidence of protection.

For research institutions, governments and policymakers, the next measurement challenge is consequently more ambitious than counting what happens after women are harmed. It is identifying whether the conditions that make that harm more or less likely are changing before it occurs. Because if public policy intends to prevent violence against women, its measurement systems must be capable of measuring more than violence. They must also measure safety.