A coherent procedure for analyzing Segregation (in its Equality dimension) in a certain population is proposed, together with an application to the study of school Segregation. Two new segregation indices for non-binary variables are defined, based on the non-parametric Chi-Square independence/homogeneity test statistic, which we call the Chi-Dissimilarity Index. In addition, a procedure based on OWA operators is proposed to generate an index of segregation in a group of organizational units. With this approach it is possible to use any segregation index considered appropriate according to the object of study, where the OWA operator weights each organizational unit with respect to its contribution to the level of segregation in relation to a certain social condition of the subjects that compose it.