France’s Recorded Intimate Partner Violence, 2024

When recorded violence stops rising rapidly, have women become safer? Issue 56 of Interstats Info rapide, published by France’s internal security statistical service, SSMSI, provides a case requiring careful interpretation. Béryl Matinet and Maxine Voinson analyze administrative records of intimate partner violence in 2024. Published on 23 October 2025, the report describes near stabilization in registrations, not the prevalence of all violence.

Recorded victims are not necessarily distinct people

The official report summarizes approximately 272,400 recorded victim entries in 2024, an increase of 0.4% from 2023. Its population is people aged 15 and over in France; it covers offenses committed by current or former partners, not only legally married couples.

The methodological note explains that the same victim is counted for different offenses concerning them. The total is therefore neither a deduplicated count of people nor all people experiencing violence. An administrative system can count its records accurately without directly answering how many individuals have been harmed. This does not make the numbers worthless. It makes their production rules inseparable from their meaning.

Registration time is not victimization time

Around 30% of registrations concern events preceding the year of registration. Annual changes consequently reflect violence alongside reporting behavior, reception conditions, legal classifications and recording capacity. An increase may include injuries that could not previously be disclosed. Stable registrations cannot, on their own, establish the success of prevention policy.

This distinction matters for accountability. If institutions define fewer records as better performance, improved access to reporting can appear to make matters worse. Better questions concern whether people obtain reception and protection, whether harms are recognized accurately and whether repeated victimization declines. These are research questions prompted by the administrative totals, not causal conclusions already demonstrated by the report.

Gender differences need their denominators

Women account for 84% of recorded victim entries, while men account for 85% of people identified as suspects in the report. The latter is not a conviction rate. The figures should not be joined as if they described matching pairs of cases. Suspects enter the relevant records after being heard by police or gendarmerie, and their proceedings need not correspond to the victims registered that year.

The report also uses the VRS victimization survey to explain why administrative records reveal only part of the violence. Survey and police populations, periods and behavioral definitions differ. Their figures cannot simply be divided or extrapolated into the total number of unreported victims. Gender asymmetry should not erase male victims either; it should direct attention to different harms, severity and barriers to assistance.

Returning statistics to the question of safety

Read with Japan’s violence-experience survey: French administrative registrations and Japanese questionnaire responses are not the same indicator and cannot rank which country is more violent. Their definitions, treatment of unreported experiences and excluded populations offer more useful comparisons.

Saffioti’s analysis of gender violence connects these methodological questions to power. Can the category assigned to a harm affect access to protection, resources and public recognition? Numbers can support demands for an institutional response, provided precision about records accompanies attention to people who remain unrecorded.