63

In order to carry a statistical survey, one must give a precise definition of the field of survey and find a method that covers that field as completely as possible. Beyond the “spontaneous” definition of the homeless linked to their visibility, French statisticians have chosen a definition that refers to the housing situation only, independantly from other factors such as the breaking-off of social relations or mental illness, which are part of the common representation of the homeless. However, in each “housing situation”, whether an emergency housing center, a long-term hostel, or the street, the structure of the population is different, especially as regards gender. Furthermore, the self-perception of the inquiree’s situation influences his or her answers and must be taken into account when the results are interpreted. But these difficulties in defining and constructing categories are also instructive: they show the limits of statistical tools and of the type of vision they offer of reality; they also increase the knowledge of the universe these statistics are based on.