Citation

Kennedy TD. Crim. Justice Behav. 1986; 13(2): 165-184.

Copyright

(Copyright © 1986, SAGE Publishing)

DOI

10.1177/0093854886013002003

PMID

unavailable

Abstract

Trends in the classification of prison inmates--readily-scored objective models and computerized psychometric profiles--are discussed in the context of a status evaluation of two psychometric approaches: the Megargee MMPI Typology for federal inmates and the Eber Diagnostic and Risk Assessment Profile for state prison inmates. Testing of youthful offenders has demonstrated that the Megargee system differentiates 10 types of inmates that vary on a broad array of variables. Cross-validation attempts with other than federal youthful offenders have yielded mixed results. Recent studies suggest that the instrument is not effective as a predictor of inmate violence or aggression during incarceration. Test-retest reliability investigations reveal a marked instability and higher than anticipated changes in group membership. The most frequent application of the typology has been in making intrafacility housing assignments. The original validation sample of 3,323 "uncontaminated" cases of Georgia state prison inmates on which the Eber system is based has not been cross-validated with other states' inmates although test results have been used in making classification decisions, institutional and program assignments, and release recommendations. Incremental validity and test-retest reliability have not been determined. Unpublished concurrent- and predictive-validity studies have yielded mixed results demonstrating the problems in predicting low base-rate behaviors. At least one state has experienced a marked reduction in the percentage of inmates initially classified to maximum security since the system's inception. Poor training and supervision of test users and organizational factors appear to detract from the system's potential usefulness.

Language: en