Date of Award


Document Type

Doctoral Research Project

Degree Name

Doctor of Psychology (PsyD)



First Advisor

Radhika Krishnamurthy

Second Advisor

Julie Costopoulos

Third Advisor

Ashok Pandit

Fourth Advisor

Robert A. Taylor


Clinical and forensic psychologists often evaluate sex offenders to determine the level of risk they pose to the community and identify their treatment needs to reduce the risk of future recidivism. Personality assessments are administered in sex offender evaluations to aid in answering these referral questions, while also providing information regarding the evaluees’ response style to the test and descriptions of their personality functioning. The Minnesota Multiphasic Personality Inventory (MMPI) and its revisions have been the most widely used personality assessments in sex offender evaluations, with previous research demonstrating that sex offenders often respond defensively to the test by minimizing or denying their psychological problems, thus limiting the interpretability of the test results. The current study aimed to develop empirically-derived optimal cutting scores for various Minnesota Multiphasic Personality Inventory-2-Restructured Form (MMPI-2-RF) substantive scales (i.e., Higher-Order (H-O), Restructured Clinical (RC), Specific Problems (SP), and Personality Psychopathology Five (PSY-5) scales) for sex offenders to adjust for defensiveness and denial of psychopathology. Archival MMPI-2-RF data from a sample of N = 142 adult male sex offenders, previously deemed a defensive subgroup through cluster analysis, was compared to MMPI-2-RF data collected from a community sample of N = 135 adult men to compare means and standard deviations on the substantive scales and derive optimal cutting scores for sex offenders using receiver operating characteristic (ROC) analyses. Multivariate analysis of variance (MANOVA) results followed by a series of univariate analyses of variance (ANOVAs) demonstrated statistically significant differences in scores between the sex offender sample and the community comparison sample on 29 of the 40 substantive scales. ROC analyses produced area under the curve (AUC) values of greater than .70 for three of the substantive scales (i.e., RC7, RC9, COG). With the exception of the RC6 and JCP scales that produced very low AUC values and exceptionally high optimal cutting scores, the optimal cutting scores for all other substantive scales fell between 40.5 (SHY) and 62.5 (AGGR-r). Specifically, three scales fell between optimal cutting scores of 40-44, 14 scales between 44-49, 13 scales between 50-54, seven scales between 55-59, and one scale between 60-64. Alternative cutting scores at equal intervals of 40, 45, 50, and 55 showed that a T score of 45 was optimal for 10 scales, T 50 for 25 scales, and T 55 for five scales. These empirically-derived optimal cutting scores and alternative cutting scores developed to enhance practical applications can potentially be used in sex offender evaluations to adjust for defensive responding, providing a more accurate interpretation of sex offenders’ personality characteristics and psychopathology. Implications of these findings were discussed.


Copyright held by author