Hundreds of studies have been published on fear of crime, but few models offer a unified framework for this social phenomenon. This is the ambition of the model of experiential and expressive fear of crime (EEF) developed in the mid-2000s by a team of British researchers. However, despite its numerous contributions, this original model faces two limitations. First, the different versions of the model present certain discrepancies. Second, the statistical methods used consistently rely on hypothetico-deductive reasoning, which may overlook the existence of certain combinations of variables. The objective of this article, based on a French piece of research, is to revisit this model by combining two different families of statistical methods: multivariate configuration analysis and logistic regression analysis. The first analysis attempts to overcome these two limitations by adopting an inductive approach, which involves studying the multidimensional structure of the data without imposing predefined structures on them. It identifies four classes of respondents, each associated with a specific relationship to experiential fear and expressive fear. If the ‘unworried’ and the ‘worried-dysfunctional’ associate these two dimensions of fear of crime, the ‘anxious’ and the ‘worried-functional’ clearly separate them. By adopting an inferential approach, the second analysis aims to determine the sociodemographic factors of these different groups. It shows that the predictors vary significantly from one group to another and that no variable (not even gender) is a predictor for all groups. These results encourage moving beyond the dichotomous conceptualization (‘worried’/ ‘unworried’), which is still widely used in the study of fear of crime. Systematically identifying these different groups could also help combat fear of crime more effectively by implementing targeted and adapted public policies.
Noble et al. (Tue,) studied this question.