A Comparison of Four Clustering Methods Using MMPI Monte Carlo Data

Abstract

Monte carlo procedures were used to generate data sets that resembled MMPI psychotic (8-6), neurotic (1-3-2), and personality disorder (4-9) pat terns. Lorr's clumping method, inverse factor analy sis, average linkage, and Ward's method were the clustering methods compared. The solutions were found to vary in terms of misclassifications and coverage. The clustering solutions also varied as a function of different optional parameters associated with each method.