Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)




Throughout the past two decades of MMPI research minimal effort has been directed toward adolescent populations. Notable exceptions to this have been the pioneering works of Hathaway and Monachesi (1951, 1953, 1957, 1960, 1963), and the later prediction studies concerning subtypes of delinquents, emotionally disturbed adolescents and most recently, drug abusers. Characteristic of all this research was a focus on the study or demonstration of the MMPI itself and its ability to differentiate personality types that the test does in fact identify. The most frequently occurring profile type uncovered by these studies appears to have been the 4-8 configuration. Some clinical observations of these individuals have been offered, but with little or no empirical backing.

It was the purpose of this investigation to inquire in more depth into the personality and functioning of these adolescents, while obtaining quantifiable information. More precisely, this was a problem in actuarial prediction; the specific model being that of predicting from a taxonomic class.

Subjects were selected from the adolescent psychiatry clinic at the University of Kansas Medical Center in Kansas City, Kansas. All persons seen in the adolescent clinic between the dates of January 1, 1969, and December 31, 1970, were included in the study. This was further broken down into two one-year samples for the purposes of cross-validation. Two types of data were analyzed for each of three groups of .subjects ("pure 4-8," "mixed 4-8," and "non 4-8"). These include test (MMPI) data and non-test data (hospital charts, biographical data sheets, etc.). The latter were rated by expert judges using a specially developed checklist of clinical descriptors (criterion characteristics).

A 19 x 3 x 2 x 2 factorial design with repeated measures on one factor was used to analyze the initial one-year sample of adolescents, as well as the cross-validation sample (19 levels of MMPI scales and subscales x 3 levels of Profile Types x 2 levels of Age, Sex). The analysis of the descriptor list consisted of Chi-square tests of association with multiple, rather than dichotomous, classification categories being used.

The results of the Chi-square tests for both the Year I and Year II data indicated that a majority of the criterion characteristics did not attain any measure of cross-validation. The analysis of variance, however, indicated replicated differences (p=.01) between the "non 4-8" group and both the "pure 4-8" and "mixed 4-8" groups on 12 of the MMPI scales and subscales. No other statistically significant differences were obtained.

It appears clear that a consistent, valid pattern of MMPI scale and subscale scores emerged from this study for the "mixed 4-8" and "pure 4-8" groups when considered together. However, the general failure to predict the criterion characteristics from these test- defined classes casts a great deal of doubt on the non-validated rating methods used by other researchers. This pertains particularly to the research done by Gilberstadt and Duker (1965) and to a lesser degree to that of Marks and Seeman (1963). Although there appears to be substantial agreement between these authors' results, both in terms of the attributes and means found among similar code groups it must be re-emphasized that until cross-validation is provided for these studies their use should be viewed critically.