• Schizophrenia subtypes are defined predominantly by manifest symptoms and behavior. This report, based on sign and symptom data from the International Pilot Study of Schizophrenia, addresses three questions: (1) Are traditional subtype diagnoses applied similarly across cultures? (2) Are the various traditional subtypes symptomatically distinguishable from one another? (3) Can cluster analytic techniques define a more distinctive set of schizophrenic subgroups? Present State Examination data were reduced to 27 psychopathologic signs and symptoms. Profile analysis of variance results indicate that each subtype appears similar, regardless of center of origin. However, this is based on a lack of distinguishing features between different subtypes. On the other hand, when a cluster analytic technique was used, it showed one large and three small subgroups, each readily distinguishable from the others. These subgroups, labeled "usual," "flagrant," "insightful," and "hypochondriacal," are described clinically. If replicated or validated, such subgroups may prove meaningful in future considerations of subdivisions of the schizophrenia syndrome.

Another View of Schizophrenia Subtypes: A Report From the International Pilot Study of Schizophrenia

Multivariate classification of alcoholics by means of the MMPI.

To delineate homogenous subgroups among hospitalized opiate addicts, a multivariate correlational clustering technique was applied to the Minnesota Multiphasic Personality Inventory profiles of 1,500 addicts, subdivided into 10 subsamples (5 for each sex) representing four different categories of admission into treatment (civilly committed, volunteers, probationers, prisoners). Within each subsample, two homogeneous and replicable profile types were isolated. Type I (33% of all addicts) showed elevations on Scales 2, 4, and 8, suggesting marked subjective distress, nonconformity, and disturbed thinking. The much smaller Type II (about 7% of addicts) showed a single peak on Scale 4. Sixty percent of addicts thus were unclustered. The two basic types, however, were very effectively discriminated on a variety of other psychometric indexes and were consistent with the two major profile types found among alcoholics in prior research.

Subgroups among opiate addicts: a typological investigation.

The sensitivity of the classifications produced by cluster analysis to type of data preparation and choice of program has been demonstrated. Because of this sensitivity, it is helpful to determine at least that a given cluster program can reproduce known groups in a stable way utilizing data of the type to be analysed in the real sample. Once this has been accomplished, analysis of the real data can be performed. In the application of clustering techniques to problems of psychiatric classification, the results of the analysis of real data can be related to diagnostic categories and characteristic profiles. If these appear promising, the next step is the difficult task of validation of the groups using criteria different from the kind of data from which the patients were originally classified.

The Use of Clustering Techniques for the Classification of Psychiatric Patients

Increasing the impact of cluster analysis research: the case of psychiatric classification.

on a sample of individuals, is to detect any mutually exclusive subgroups that may exist. A procedure is thus needed for identifying the subgroups in a non-random sample of persons studied. Formal models such as factor analysis (Stephenson, 1936) and latent profile analysis (Gibson, 1959) have been proposed as solutions. A variety of looser procedures labeled &dquo;cluster analysis&dquo; have also been suggested. Within this class fall methods proposed by Thorndike (1953), McQuitty (1964), Sawrey, Keller and Con-

Belur K. Radhakrishnan

Papers

A Comparison of two Methods of Cluster Analysis