A measure of compound intersectional inequality

When the dimensions of inequality reinforce each other, it is difficult to measure the influence of each. We explore a method for measuring the distance between predetermined groups of individuals (defined a priori by gender and geographical residence). The first step consists in classifying individuals by cluster analysis (without considering the a priori groups). The second step calculates the relative frequency distribution among the clusters for each of the a priori groups. Finally, the distance between two groups is defined as the Euclidean distance of the corresponding percentage frequency distributions. We assume that the more dissimilar this distribution is, the more unequal the a priori groups are. We apply this method to 23 European countries using data of the European Social Survey 2008 and 2018. We find that the average distance between men and women has decreased.