International Journal on Criminology Volume 3, Number 2, Fall 2015 | Page 96

International Journal on Criminology We were therefore able to draw on a set of indicators bearing both on the victimhood reported by households and on the elements that facilitated the characterization of households, their occupants, and their living environment detailed in table 1 and in appendix A4. Table 1: Summary of the characteristics of households, their accommodation, and their neighborhood. Characteristics of the household Characteristics of the accommodation Characteristics of the neighborhood Variables • Educational qualification of the reference person • Household income • Marital status • Type of household • Professions and socioprofessional categories of the reference person • Employment status • Age of the reference person • “Legal status” • Surface area • Presence of a watchperson • Number of security devices • Presence of a dog • Type of accommodation • Centrality • Size of the urban area • Type of neighborhood • Sensitive Urban Zone (SUZ) • Knowledge of burglary Source: “Cadre de Vie et Sécurité” survey, INSEE-ONDRP Statistical Analysis With regard to exploratory statistical analysis, taking household characteristics one by one allows the proportion of households corresponding to the values of each of the variables to be “measured.” Although this “univariate” approach (a single variable studied at a time) provides very complete descriptive information, it has two main limitations. The first limitation appears when there is a need to treat a large number of variables or characteristics. The completeness of the univariate analysis loses its significance in the face of the multiplicity of the results obtained, and the interpretation of these results loses clarity. In our case, we have available to us seventeen variables for a total of seventy values. The exploration of all of these variables individually would be tedious and would ultimately teach us little. The second limitation is that univariate analysis does not allow for a description of the relationships that may exist between several characteristics. This task, however, is of particular interest because it allows associations between various factors to be established, and thus allows new information to be acquired. These associations can be determined two by two through bivariate analysis. But once again, the problem of the number of intersections between all of the characteristics and of the exploitation of these results in terms of usable information arises. In order to overcome these limitations and offer an analysis of the characteristics of households that is both synthetic and informative, here we use multivariate analysis 91