More Theft and Assault Among People Without Housing

Counting women’s exposure to violence requires examining both housing institutions and harms that can precede the loss of a home. In Davantage de victimes de vol ou d’agression parmi les sans-domicile, Lucile Jamet and Christelle Thouilleux examine theft and violence through French survey evidence. Their November 2015 Insee Focus bulletin complicates an easy opposition between a dangerous street and a protective home: some harms may precede the loss of housing, while assistance and accommodation also shape who becomes visible in statistics.

The analysis concerns French-speaking adults aged 18–75 using accommodation or meal services in metropolitan French urban areas of at least 20,000 inhabitants in 2012. It compares their reports with ordinary-household survey evidence. These are not police complaints, a census of everyone sleeping outside, or a measure of today’s situation. The article’s population restrictions matter as much as the percentages.

A night’s address is not a history of exposure

In the article’s definitions, sans-domicile includes people who spent the preceding night in a place unfit for habitation or in accommodation supplied by an assistance organization. The narrower sans-abri category covers the former. A person temporarily staying with a private individual can lack a home of their own without entering this statistical category. Counting only visibly occupied public spaces would therefore produce a different population.

The violence questions concern 2010 and 2011, regardless of the person’s housing conditions during those years. Accommodation is classified for the night before the interview. Comparing those categories cannot establish that a shelter caused or prevented earlier violence. Nor does a more stable placement at the interview mean that a woman had been protected throughout the preceding period. The mismatch in time is a substantive limit, not a minor footnote.

The gender pattern depends on the harm being counted

Table 2 reports that 31.2% of women without housing, compared with 26.8% of men, experienced aggression or violence in the two years covered. This broad measure includes physical, sexual and verbal aggression, including threats and insults; it is not a sexual-assault prevalence rate. Theft and attempted theft show the opposite gender pattern: 22.1% of women and 28.3% of men. The evidence cannot be reduced to women experiencing more of every kind of harm.

Another easily misquoted figure concerns leaving one’s previous home. Among women without housing who reported aggression, 17% said they had left because of family violence. The corresponding figure among all women without housing was 6.3%. The denominator changes the claim: 17% does not describe all women without housing, still less all women in France. Family violence also cannot automatically be narrowed to violence by a partner.

These distinctions unsettle a moral geography in which home is always a refuge and the street the sole origin of danger. They do not establish a single causal route from violence to homelessness. They do make it harder to treat getting someone indoors as sufficient evidence of lasting safety. The article also reports fear of insecurity among people avoiding collective centres. That invites scrutiny of the accommodation offered before interpreting refusal as a failure to cooperate; it does not prove the safety or danger of every centre.

Who remains outside the questionnaire?

Service-based research makes experiences countable that an ordinary-household survey can miss. It has boundaries of its own. This analysis excludes non-French speakers and people outside its service and geographic frame. Being absent from the data does not establish safety. Equally, differences between users of different services cannot by themselves reveal how institutions selected people into those services or what options each person had.

The first Virage results on sexual violence offer a productive comparison because their ordinary-household sampling frame has different limits. The studies ask different questions, cover different periods and populations, and cannot be combined into a corrected national rate. Reading them together redirects attention from asking which percentage is largest to asking whose experiences each design can register. Protection involves both places to stay and the capacity to recognize harms that do not fit a familiar picture of homelessness.