PAI May Be Helpful for Gender-Responsive Assessment of Female Inmate Misconduct

Megan Davidson & Jon R. Sorensen, East Carolina University
Thomas J. Reidy, Monterey, California

Proper inmate assessment is critical to correctional management and institutional security. While many instruments have been developed to assist with this process, most of these tools have not been validated using samples of female inmates although distinct gender differences have been identified in the inmate population in terms of adaptation and misconduct. The Personality Assessment Inventory (PAI) is a multiscale measure of psychopathology that is being increasingly utilized in the correctional setting to assist with the inmate classification process. The current study contributes to the dearth of literature surrounding gender-responsive inmate classification by utilizing a sample of 2,000 female inmates to examine the incremental and predictive validity of the PAI in association with general and assaultive disciplinary infractions. Findings from this study reveal that the PAI scales presenting the strongest relationship to general and assaultive disciplinary infractions among this female sample included Aggression (AGG), Antisocial Features (ANT), Paranoia (PAR), and the Violence Potential Index (VPI). Moreover, findings derived from this study suggest that certain PAI measures, specifically ARD-T, DRU, and more general substance abuse and mental health indicators may be useful in gender-responsive assessments during the female inmate classification process.

Personality Assessment Inventory, PAI, psychological testing, validity, gender responsiveness, institutional misconduct

Assessment of inmate risk is an important step for correctional staff and mental health professionals to appropriately house and treat incoming inmates. Many risk assessments tools have been developed to aid this process but these tools have been tested and validated with male inmate populations. With distinct differences between male and female inmate misconduct, it is becoming a pressing issue to validate risk assessment tools for female populations. “Using assessment instruments tested exclusively on male inmates to guide correctional decision-making regarding female inmates has been associated with overestimating dangerousness and overutilization of restrictive housing, which may predispose females to engage in misconduct and undermine treatment efforts” (p. 72-73). “An effective risk assessment process for female inmates would incorporate factors that are most notably associated with female misconduct (i.e., trauma, substance abuse, mental illness, family support, etc.) into pre-existing gender neutral assessment instruments to provide a more accurate picture of the inmate at intake” (p. 73).

The Personality Assessment Inventory (PAI) is a multiscale measure of psychopathologywhich has been used on male inmates to assist with inmate classification. “The PAI is a self-report inventory that includes 344-items, each answered on a 4-point scale (false, slightly true, mainly true, and very true). The PAI is comprised of 22 non-overlapping scales and subscales measuring specific constructs categorized as response bias, clinical, treatment planning, and interpersonal” (p. 74). Some subscales that have been associated with inmate misbehavior are the Aggression Scale (AGG), the Antisocial Features Scale (ANT), and the Violence Potential Index (VPI).

“The primary aim of this study was to identify the predictive and incremental validity of PAI scales with an expanded population of imprisoned females utilizing general disciplinary infractions and inmate/staff assaults as outcome variables. A second objective was to determine the contribution of gender-specific correlates to disciplinary infractions, namely, pre-existing trauma, substance abuse, and mental health needs”(p. 78). Database information was collected from the Oregon Department of Corrections (ODOC) from August 2000 to December 2008. A final total of 2,181 female inmates were included in the sample after excluding inmates with invalid profiles. The PAI’s subscales of Traumatic Stress (ARD-T) and Drug Problems (DRU) were included as potential gender-responsive predictors. Additional information about mental health and substance abuse was collected from ODOC-created measures. Outcome measures were classified as either general or assaultive prison misconduct.

For the results, “paranoia (PAR), ANT, AGG, and the VPI have the strongest relationships to general disciplinary rule violations with AUCs of .6 or higher. While findings relating ANT, AGG, and VPI to outcomes were to be expected, PAR has not been discussed much in relation to disciplinary infractions” (p. 76). “In relation to the second objective of the study, our findings support the importation theory and prior research indicating that females enter prison with elevated levels of stress, exposure to traumatic stressors, depression, and noteworthy substance abuse problems,” (79).

“With regard to the predictive validity of the PAI, findings from this study reveal that the PAI scales presenting the strongest relationship to general and assaultive disciplinary infractions included AGG, ANT, PAR, and VPI. The AUCs for these scales fell in the small to medium range with higher AUCs for assaultive infractions, and their relationship held even after background characteristics were accounted for in a baseline model.” These findings differ from previous literature that “reported that only ANT emerged as a robust predictor of general and aggressive misconduct for female inmates, but did not add explanatory power once demographic and criminal history variables were controlled. The authors concluded that ‘consideration of theoretically relevant PAI indicators other than ANT in conducting risk assessment is unlikely to yield additional useful information, at least in female offenders’ Findings from the current study challenge this assertion. Specifically, AGG, PAR, and VPI significantly contributed to incremental validity in our study. The discrepant results are most likely attributable to the considerable difference in sample sizes between the two studies (113 vs. 2,311 inmates).” (p. 78).

“The unique finding that PAR emerged as an unexpected and strong predictor of general and assaultive rule violations among this female sample suggests that the connection between misconduct and mental health symptoms may be more pronounced in female inmates because of pre-existing conditions and/or gender differences in adaptation to the prison experience…The identification of PAR as a new gender-responsive risk factor provides for a more expansive understanding of potential mental health reactions of women to traumatic life events and imprisonment leading to disciplinary infractions. Manifestations of PAR as a gender-responsive risk factor can include hypervigilance in monitoring the environment for harm, feelings of persecution, wariness toward others, hostility, resentment, and even psychosis. Further, when viewed within the context of elevated PAI scales STR, ARD-T, and DEP, female inmates with a history of adverse and traumatic life experiences may be more prone to mental health deterioration manifested by interpersonal conflicts, failure to follow rules, and assaultive behavior” (p. 79).

“An additional unique contribution of this study involves examining the relationship between defensive profiles and disciplinary infractions among a large sample of female inmates. Response distortion measures are important to consider in a prison context where there are strong inducements for inmates to deny or underreport abnormal personality traits, psychopathology, and antisocial tendencies…Findings from the current study suggest that female inmates with strong antisocial traits attempting to present an overly positive impression and deny minor faults to which most people admit are equally as likely to commit rule infraction as inmates not possessing these traits” (p. 79).

“In relation to the second objective of the study, our findings support the importation theory and prior research indicating that females enter prison with elevated levels of stress, exposure to traumatic stressors, depression, and noteworthy substance abuse problems. Significant pre-existing psychological and traumatic distress were indicated by PAI scores greater than 1 SD above the mean for community samples on the scales of DEP, ARD-TS, and STR. Similar elevations on scales measuring aggressive, antisocial and borderline features revealed evidence of behavior and personality disturbance. Historical information for this group further revealed a pattern of violent arrests and elevated substance abuse. These findings are consistent with other studies investigating the background and special needs of female inmates” (p. 79).

“Although elevated rates of stress and depression are present in this sample, neither scale strongly predicted disciplinary violations. However, after controlling for background factors, the experience of trauma (ARD-T) by female inmates added predictive value for general rule infractions. Additionally, the ODOC mental health needs indicator added predictive power to the base models for general misconduct but exerted less influence than PAI scales as predictors of misconduct. It is appears that the violations in this study may relate more to personality disorder than more serious mental illness” (p. 79).

“While mixed findings were presented from the current study regarding the gender-specific correlates of misconduct previously identified in the literature, the current study offers ample support for the incorporation of gender-specific factors into the female inmate assessment process. Findings derived from this study also suggest that certain PAI measures, specifically ARD-T, DRU, and the mental health scales may be useful gender-responsive scales to be incorporated into the female assessment and classification process. Measures such as the PAI that add incremental validity to demographic and criminal history variables can be useful tools in the classification process to plan for treatment needed during incarceration, anticipate management problems, and provide early identification of community services required upon parole” (p. 80).

“This study attempted to disaggregate severity of injury associated with violent infractions, but this analytical strategy was fruitless because of overall rare occurrence of injurious assaults among the female sample (0.7%). The comparatively low rate of serious violence among female offenders is a universal finding that clearly indicates women react differently to imprisonment either because of personality and life experiences imported from their community, a difference in their collective response to the experience of incarceration, situational factors, or some combination of thereof” (p. 78).

As always, please join the discussion below if you have thoughts or comments to add!