Functional \(k\)-sample problem when data are density functions (Q964616)

Analysis of distance is a generalization of ANOVA to the case when distances between subjects are available only. The author proposes to apply this technique to the analysis of homogeneity in the \(k\)-sample problem when each subject is represented by a probability density function. The \(L_1\) distance is used for the analysis. Two versions of permutation tests are proposed to test the homogeneity of the samples. Special attention is paid to weighted data analysis. Results of simulations and application to European regional income distributions are presented.