FORUM on Corrections Research

Using psychological testing to predict institutional misconduct

by Brad Kelln,¹ David Dozois and Ian McKenzie²

Psychological assessment of offenders is important for efficient decision-making, prediction of institutional misconduct, treatment planning and parole.³ Assessment improved with the introduction of personality and psychopathology inventories, such as the Millon Clinic Multiaxial Inventory (MCMI-III).⁴

The MCMI was originally developed for diagnostic decision-making and treatment planning for psychiatric populations.⁵ To date, no studies have examined the predictive use of the MCMI personality profiles in a general inmate population. One advantage of using the MCMI-III in predicting institutional misconduct is that the instrument is already in wide use within the forensic system,⁶ and does not add significantly to staff workloads. Moreover, clinicians and case managers have been found to make unreliable and inaccurate predictions of serious misconduct if they cannot base their Psychological predictions on objective assessment data assessment The general assessment literature⁸ also supports the need to use existing objective assessment devices.

The current study used the MCMI-III on an inmate population to predict institutional misconduct. The incremental validity of MCMI-III was then examined to test its predictive power beyond readily accessible data such as demographic information.

Methodology

The study assessed 142 male offenders incarcerated in a medium-security federal institution (Drumheller Institution). Their average age was 30.77 years (SD = 8.99; range 18–67). Their average sentence length was 60.68 months (SD = 60.78). The sample was 59% Caucasian, 27% Aboriginal, 5% Asian, 4% Hispanic, 4% Black and 1% other ethnic groups. These figures are comparable with demographic statistics from western Canadian institutions.⁹ Just over one third (35%) of the sample were convicted of crimes against the person (i.e., sexual assault, physical assault and homicide), while 65% were convicted of other types of crimes (e.g., property, drug- and alcohol-related offences). The participants’ age, ethnicity, type of offence and sentence length were included in the evaluation process.

The MCMI-III is a 175-item inventory, scored in a true-or-false format and designed to be consistent with the Diagnostic and Statistical Manual of Mental Disorders — Fourth Edition (DSM-IV).¹⁰ The MCMI-III consists of 10 clinical scales, 11 basic personality scales and 3 severe personality scales.

Information on institutional misconduct was obtained from inmate records. Institutional misconduct was defined by five behaviours:

1. number of official reprimands;
2. number of days spent in segregation;
3. number of early lock-ups;
4. number of monetary penalties; and
5. number of days of program suspension.

MCMI-III data was collected during assessment, which was conducted shortly before the subjects’ admission to Drumheller Institution, as most of the inmates were routed through other institutions (e.g., Edmonton Maximum Security). Demographic information and institutional misconduct data were gathered from administrative records. The average time from admission date to the collection of behavioural misconduct information was 9.28 months (SD = 4.24).

Results

Reprimands are given for less serious misconduct.¹¹ Reprimands are also a non-behavioural consequence of misconduct, whereas other penalties have behavioural ramifications (e.g., paying fines, going to segregation). Therefore, the reprimand records of each inmate Table 1 analysed separately from more severe misconduct penalties. The results of this study are presented in two parts: individuals with only reprimands were compared with the rest of the sample ( e. g., control group 1), and individuals with behavioural penalties were compared with the rest of the sample ( e. g., control group 2).

In the prediction of reprimands, subjects were classified into two categories: reprimand (n = 21) and control group 1 (n = 121). In a discriminant function analysis (DFA),¹² the demographic variables of age, sentence length and type of offence produced a correct classification rate of 50% in the prediction of institutional reprimand (i.e., of the entire group, 50% were correctly identified as either receiving or not receiving an institutional reprimand). When the MCMI-III scales were added to the demographic variables, DFA correctly predicted the likelihood of reprimands for 80% of the inmates. MCMI-III scales that were strong predictors for the reprimand group (that is, for which the reprimand group scored significantly higher than the control group) were the Somatoform Clinical Syndrome scale, the Avoidant Personality scale and the Self-Defeating Personality scale.

To be included in the institutional misconduct group defined by behavioural penalties, subjects had to have received one or more of the following consequences: segregation, early lock-up, monetary penalty or suspension. Subjects were classified into two categories: the misconduct group (n = 41) and control group 2 (n = 101). The demographic variables alone predicted behavioural penalties for 69% of the inmates. This increased to 76% when the MCMI-III scales were included in the analysis. The results are summarized in Table 1.

Table 1

In this analysis, 11 variables emerged as significant predictors of institutional misconduct.Those receiving behavioural penalties were younger and scored higher on the Narcissism, Aggressive-Sadistic, Schizoid, Antisocial, Aggressive, Passive-Aggressive and Borderline personality scales. Higher scores on the Compulsive personality scale are related to increased self-control and thus it is not surprising that those receiving behavioural consequences for misconduct were found to score lower on the Compulsive personality scale.

For the clinical scales, the misconduct group scored higher on the Alcohol Dependence and Thought Disorder scales than the control group. The misconduct group also obtained significantly higher scores on the Disclosure Modifying Index, indicating that they were more open and honest in responding to the MCMI-III. Such an unexpected finding may be a statistical artifact of elevated pathology in other clinical and personality scales (i.e., the Disclosure scale on the MCMI-III is composed entirely of other scales and thus having significantly higher scores on a large number of the individual scales would increase the overall score of the Disclosure index) or may be attributed to an impulsive response style.

Conclusion

The results of this study indicate that the MCMI-III strengthens the data forensic practitioners can use to predict misconduct in correctional facilities. This information, in turn, may be used to identify the individuals who should be considered higher priorities for receiving appropriate psychological interventions (e.g., impulse control training, problem-solving skills training) to reduce the amount of behavioural misconduct in the prison system. Further research is necessary to establish cutoff scores for use within specific institutions based on the needs of the institution and the demographic representation of its population.

1. The authors would like to thank the staff at Drumheller Institution for their support in the completion of this study. This research is supported by the Medical Research Council of Canada, the Killam Foundation and the Alberta Heritage Foundation for Medical Research. This support is gratefully acknowledged. The first two authors can be reached at Program in Clinical Psychology, EdB 292, University of Calgary, 2500 University Drive, N.W., Calgary, Alberta T2N 1N4.

2. Drumheller Institution, P.O. Box 3000, Drumheller, Alberta T0J 0Y0.

3. R. Serin, “Psychological intake assessment: Contributing to contemporary offender classification,” Forum on Corrections Research, 9,1 (1997): 51–54.

4. T. Millon, C. Millon and R. Davis, Millon Clinical Multiaxial Inventory – III Manual (Minneapolis, MN: National Computer Services, 1994).

5. T. Millon, Manual for the Millon Clinical Multiaxial Inventory
(Minneapolis, MN: National Computer Services, 1983); see also S. Wetzler, “The Millon Clinical Multiaxial Inventory (MCMI): A review,” Journal of Personality Assessment,
55 (1990): 445–464.

6. J. R. Weekes, S. J. Morison, W. A. Millson and D. H. Fettig, “A comparison of native Metis and Caucasian offender profiles on the MCMI,” Canadian Journal of Behavioral Science, 27 (1995): 187–198.

7. R. P. Cooper and P. D. Werner, “Predicting violence in newly admitted inmates: A lens model analysis of staff decision making,” Criminal Justice and Behavior, 17 (1990): 431–447.

8. G. Groth-Marnat, Handbook of Psychological Assessment (New York: Wiley, 1990); see also P. E. Meehl, “What can the clinician do well?” Problems in Human Assessment, D. N. Jackson and S. Messick, eds. (New York, NY: McGraw-Hill, 1967): 594–599.

9. Correctional Service of Canada, Basic Facts About Corrections in Canada (Ottawa, ON: 1993).

10. American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders, 4th ed. (Washington, DC: 1994).

11. D. Lovell and R. Jemelka, “When inmates misbehave: The costs of discipline,” The Prison Journal, 76 (1996): 165–179.

12. Discriminant function analysis operates by finding the line of best fit for the data (i.e., the linear composites of the dependent variables are calculated so that they are successively maximized). Stepwise discriminant function analysis removes redundancy (or multicollinearity) among variables, and this yields information about the unique variance accounted for by the predictors. See E. J. Pedhazur,
Multiple Regression in Behavioral Research, 2nd ed. (Fort Worth, TX: Harcourt Brace, 1982).