Validation of groups derived by cluster analysis of psychiatric inpatients' MMPI item responses and scales

Abstract

The central focus of this study was the behavioral validation of subgroups of psychiatric inpatients formed by cluster analysis. The clusters were formed on the basis of Minnesota Multiphasic Personality Inventory (MMPI) responses. Case files of 174 (170 males, 4 females) psychiatric inpatients were randomly selected from a neuropsychiatric Veterans Administration Medical center;The Behavior Ratings Scale (BRS) was developed to derive from case histories 58 behaviors commonly observed among the V.A. population. Behaviors detected from case summaries were rated by three trained raters who completed a BRS for each of 174 subjects. Inter-rater reliabilities for each pair of raters produced statistically significant and adequate reliability. The obtained reliabilities indicated the efficacy of the BRS to accurately draw pertinent behavioral information from case records;The factor analyses conducted for this research utilized the 58 items of the BRS. A striking result was the similarity between the first four factors found in this study and the parallel factors found in two published rating scales which are included in the BRS;The MMPI profiles for the 174 subjects were grouped by Ward's hierarchical cluster analysis. Clustering was done for the 13 clinical MMPI scales and for the 399 MMPI item responses, resulting in nine clusters for each technique. Mean profiles of each group and published descriptors of behaviors and thought patterns typical of the cluster groups were presented;It was posited that behaviors drawn from case records would validate the groups derived from item based and T-score based cluster analyses of the MMPI. In general the validity of the clusters was supported by both univariate and multivariate analyses (discriminant function analysis) showing significant behavioral differences among the MMPI clusters. Comparisons of BRS behavioral descriptors of MMPI clusters and the MMPI profiles revealed clinically meaningful linkages in the majority of cases. The use of both types of cluster analyses showed potential for classifying subgroups of this population of psychiatric inpatients. The usefulness and versatility of cluster analysis as a classifying technique has thus been indicated.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 0 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.