Edwin I. Megargee

The Prediction of Dangerous Behavior

Error is inevitable whenever psychologists attempt to predict future behavior. This paper examines the steps involved and the factors to be considered in the prediction of behavior in general and dangerous behavior in particular. Errors can occur in identifying the relevant personality and situational variables to sample, assessing these variables, and determining their interaction. The effects of these errors are multiplied by the fact that violence is a low base-rate phenomenon. The consequences of errors in violence prediction are examined and the ethical problems and social policy implications discussed.

The association of population density, reduced space, and uncomfortable temperatures with misconduct in a prison community

This study investigated the relation of population size, amount of personal space, and population density to the incidence of disruptive behavior over a 3-year period in a correctional institution for male youthful offenders. Population size was significantly correlated with the number but not the rate of disciplinary violations. The total amount of personal space and the index of population density were significantly correlated with both the number and the rate of violations. A post hoc analysis showed that the incidence of uncomfortably hot days had no relation to disruptive behavior. These results are contrasted with laboratory data on human crowding reporting no relation between population density and behavior pathology. It is suggested that the chronic high density explored in the present field study differs from temporarily crowded conditions produced in laboratory research on human crowding.

I: the Need for a New Classification System

Since it is generally agreed that no single etiology, explanation, or treatment will be found that is universally applicable to all forms of criminal behavior, behavioral scientists are instead delineating smaller, more homogeneous subsets or types of offenders. Criteria by which the adequacy of taxonomic systems may be judged are suggested and used to evaluate existing systems. Approaches based on personality characteristics are judged most nearly satisfactory, but current systems require well-trained personnel and/ or a comprehensive array of specially collected data. It is proposed that the MMPI be used to develop a more economical taxonomic system, and the steps involved in determining such a systems feasibility and validity are outlined. A series of studies carrying out these steps is to be reported in the succeeding series of articles.

III: Revision and refinement of the classificatory rules

The third of four studies on the derivation of an MMPI-based criminal classification system, this paper reports how the classificatory rules were refined and revised and how a computer program was derived that could classify 63% of the profiles encountered in an institution for youthful male offenders with 91 % accuracy. In the course of this procedure, a tenth group was discussed. Clinical inspection of the profiles that were unclassified or multiply classified by the program resulted in 96% of the profiles being assigned to types. Mean profiles and comprehensive operational definitions are provided for all ten groups.

MMPI-A patterns of male juvenile delinquents

The usefulness of the Minnesota Multiphasic Personality Inventory-Adolescent(MMPI-A) (J. N. Butcher et al., 1992) was examined for 162 delinquent boys in a state training school. Their base rates, patterns, and configurations on all the MMPI-A scales and subscales were determined and compared with those of the 805 nondelinquent male adolescents in the MMPI-A standardization sample and with the original MMPI patterns of 7,783 delinquents identified in a literature review. The most prominent clinical scales were 4, 6, and 9, and 49/94 was the most frequent 2-point code. The study confirmed 14 of 18 hypotheses for mean differences on the 38 MMPI-A validity, clinical, supplementary, and content scales, and also found significant mean differences on 33 of the other 51 MMPI-A scales and subscales, 13 of which were large enough to be clinically meaningful. Most of the MMPI-A patterns and configurations were consistent with the literature on male juvenile delinquents.

II: Initial Development of the System

Authors’ Note: This research was supported in part by USPHS Grant No. 18468 (NIMH: Center for Studies of Crime and Delinquency) to E. I. Megargee and in part by the Federal Bureau of Prisons. All statements in this research study are opinions of the authors and do not represent the official opinions, policies, or attitudes of NIMH or the Federal Bureau of Prisons. This study is based on a master’s thesis by James Meyer, Jr. under the supervision of Edwin Megargee. Requests for reprints should be sent to Dr. Edwin I. Megargee, Department of Psychology, Florida State University, Tallahassee, FL 32306.

Seven MMPI alcohol and drug abuse scales: An empirical investigation of their interrelationships, convergent and discriminant validity, and degree of racial bias.

This study investigated the interrelationships, convergent and discriminant validity, and magnitude of racial differences on five Minnesota Multiphasic Personality Inventory (MMPI) alcohol scales and two drug scales with youthful samples (mean age = 22.3 years) of heavy and moderate alcohol users, heavy and moderate drug users, and non-psychoactive-substance-using inmates at a federal correctional institution (n = 1,048). Five scales showed some degree of convergent or discriminant validity. Racial differences were observed on five scales with blacks scoring lower on all these scales except one.

Personality Factors Associated with Over- and Undercontrolled Offenders

Summary Construct validity studies correlating Megargee et al.s O-H (Overcontrolled Hostility) scale with other MMPI-based measures of hostility and control might reflect common methods variance rather than trait validity. Could reliable differences between high and low O-H Ss be found on an inventory that did not share a common theoretical basis, common test construction methods or a common item pool with the MMPI? The 16 PF, Form C was administered to 75 youthful offenders with high and low O-H scale scores. As predicted, the high O-H group scored significantly higher on Factors C, G, and Q3 and significantly lower on Factors E, L, M and N, thus adding to the evidence for the construct validity of the O-H scale.

Assessment Research in Correctional Settings; Methodological Issues and Practical Problems.

With prison populations at record levels, classification and programming based on individual assessments are needed to allocate resources efficiently. However, there is a dearth of assessment research in correctional settings, partly because few psychologists are prepared for the special challenges posed by correctional research. This article discusses the methodological and practical problems that are especially common in or unique to assessment research in correctional settings, especially in prisons. It examines the distinctive culture of correctional institutions and the special concerns of their staffs ; reviews considerations peculiar to planning prison-based studies ; describes difficulties that may be encountered in obtaining approval by correctional institutions, funding agencies, and human subject review boards ; and presents problems that may be encountered in collecting the data and in analyzing and interpreting the results.

Racial Identity, Length of Incarceration, and Parole Decision Making:

Previous research has disclosed significant differences in the severity of sentences imposed on black and white offenders. This study investigated racial differences in black-white sentence patterns at the final stage of the criminal justice process, focusing on the actual amount of time served and the frequen cy of parole in a sample of 958 youthful offenders from a cohort of 1,345 consecutive admissions to a federal correctional institution. No racial differences were found in the actual amount of time served, but a significantly higher percentage of whites than blacks were granted parole. Blacks who were granted parole served a smaller portion of their sentences than did whites who were granted parole, but nonparoled blacks served a greater portion than did nonparoled whites. Discriminant function analysis did not show race to be significantly associated with decision to grant parole; instead, the primary factor was the entry sentence, with offenders sentenced to relatively short terms (less than eighteen months) being paroled less fre quently than those with longer terms. Since blacks were more likely to have the shorter sentences, the racial difference in percentage granted parole was partly attributable to the severity of the sentence imposed originally. Examination of thirteen offender variables reflecting personal and social history, legal factors, and institutional adjustment measures revealed signifi cant differences between paroled and nonparoled inmates but not between racial groups. Further discriminant analysis of factors associated with parole within black and white subsamples indicated that, of the thirteen variables examined, entry sentence, violence of offense, and adult maladjustment and deviance were significantly associated with likelihood of parole for both racial groups. Institutional adjustment also appeared to play an important role in decisions regarding blacks, while variables relating to personal and social ad justment before incarceration seemed to be more important among whites. It is speculated that the observable overrepresentation of blacks in penal and other correctional institutions is attributable to decisions made earlier in the criminal justice process, rather than to biased parole-granting procedures, at least in this federal institution. Further research on possible interaction of social deficits and criminal justice decision making instead of race and sen tence severity is suggested.