Display options
Share it on

Multivariate Behav Res. 1989 Jul 01;24(3):335-52. doi: 10.1207/s15327906mbr2403_4.

Replication Problems of Substance Abuser MMPI Cluster Types.

Multivariate behavioral research

D G Fisher, M D Anglin, C P Weisman, L Pulliam

PMID: 26750501 DOI: 10.1207/s15327906mbr2403_4

Abstract

Cluster analysis solutions of drug abuser (primarily heroin addict) Minnesota Multiphasic Personality Inventory (MMPI) profiles are critically reviewed. We followed the procedures in five published research studies, using 183 heroin addict MMPI profiles. The published claim that cluster analytic results have been replicated is not supported, primarily because of methodological differences in previous research. In particular, methods using measures of shape as input (primarily Pearson correlation) did not yield classifications that agreed with methods using Euclidean distance when all methods were used on the same data. A preferred solution using complete linkage with Pearson correlation is presented and recommendations for future cluster analysis research are made. Of the four heroin addict clusters identified, the "schizoid psychopath" (8-4) cluster was associated with a significantly greater number of arrests before first drug use than was the "hypomanic psychopath" (9-4) cluster.

Publication Types