Development and validation of an imprisonment-appropriate version of the Personality Inventory for DSM-5 (PID-5)


General approach / background

The main prognostic questions psychologists working in prisons need to deal with are early prison release, paroles and day paroles and the placement in open or closed correctional facilities.
The legal systems explicitly states what has to be considered when answering these questions. Examining these statutory requirements it becomes evident, that the prisoner´s personality is one important aspect that has to be considered. For example, the German criminal code  demands to consider the personality, past life, the context of the offence, the consequences in case of reoffending, the demeanor in detention and the living conditions outside of detention, when deciding about early prison release (§ 57 Abs. 1  StGB).
This juridically required focus on personality is in line with results of past research. Personality traits have been proven to be  relevant predictors of deviant behavior (Andrews & Bonta, 2010), recidivism (Gendreau, Little, & Goggin, 1996; Olver, Stockdale, & Wormith, 2014; Steiner, Cauffman, & Duxbury, 1999; Tinklenberg, Steiner, Huckaby, & Tinklenberg, 1996) and institutional misconduct (Holsinger, Lowenkamp, & Latessa, 2006).
Hence personality plays an important role for legal and empirical reasons when answering the main questions in the penal system. However, those personality inventories that are established in prisons (see Archer et al., 2006) appear inadequate because of their unreasonable number of items (e. g. Personality Assessment Inventory, PAI, Morey, 1991), their deficient psychometric properties or their lack of theoretical foundation (e. g. Minnesota Personality Inventory–2, MMPI-2, Butcher et al., 2001). Considering the serious consequences of prognostic misdeterminations in forensic settings it is stringently required to implement a practicable, theoretically well-grounded personality inventory whose psychometric quality is ascertained for self- and informant reports.

Research questions

The following questions will be investigated for the original version of the PID-5:

  1. Is the factorial structure of the PID-5 replicable in a sample of condemned prisoners?
  2. Is the reliability of the PID-5 replicable in a sample of condemned prisoners?
  3. Is the convergent validity of PID-5 and Big Five replicable in a sample of condemned prisoners?
  4. Does the PID-5 show associations to the prisoners conduct in the penal institution and statistical risk scores in a sample of condemned prisoners?

The following questions will be investigated for the imprisonment-appropriate version of the PID-5:

  1. Does the factorial structure of the imprisonment-appropriate version of the PID-5 show a better model fit than the original version of the PID-5 in a sample of condemned prisoners?
  2. Are reliability criteria in the imprisonment-appropriate version of the PID-5 superior to those of the original version of the PID-5 in a sample of condemned prisoners?
  3. Is the convergent validity of the imprisonment-appropriate version of the PID-5 comparable to the original version of the PID-5 in a sample of condemned prisoners?
  4. Does the imprisonment-appropriate version of the PID-5 show stronger associations to the prisoners conduct in the penal institution and statistical risk score than the original version of the PID-5 in a sample of condemned prisoners?

Additionally, the following exploratory question will be addressed:

  1. Which further personality measures are determinants of behavior in prison context?

Research design, participants, measures

Design. To develop such an inventory we plan the following three steps. First, we investigate the psychometric quality of the original self- and informant report PID-5 in a pilot study. The self reports are filled in by condemned prisoners, placed in socio-therapeutic departments of the penal institution. For each participating prisoner, one penal service officer and one psychologist are asked to fill in one informant report. Additionally, five prisoners, penal service officers and psychologists are asked for an interview to comment on each item of the PID-5. We aim to detect which questions are to complex, not answerable in this specific forensic context or prone to distortion. At the end of the first phase we analyze item characteristics, factorial structure, reliability, validity and self- informant report consensus.

Based on this information and the qualitative results of the interviews we select, adapt and add items to meet the special requirements of the forensic setting (e.g. basic language level, imprisonment-appropriate wording, behavior-related questions, specification of an reference group)  in the second phase. Furthermore items are planned to be sensitive to change as we want to capture personality development. Considering these aspects we plan to overcome the main problems of personality assessments in forensic settings (for an overview see Hosser et al., 2008).

In a third step, we validate the original and the imprisonment-appropriate version of the PID-5 in a large sample. We examine item characteristics, factorial structure, reliability, validity and self- and informant report consensus considering sample characteristics like type of offence and migrant background. Furthermore we want to know whether this adapted personality inventory is helpful for answering the main prognostic questions in penal systems. We further examine the power of the PID-5 to predict institutional misconduct or removal of paroles.

Participants. Data will be collected from juvenile and adult, male and female prisoners. Prisoners are placed in special departments for social therapy in prisons of Lower Saxony and Bremen, Germany. Participants can choose between money (10 €), two pounds of coffee or two packets of tobacco as compensation. We plan to assess N = 50 prisoners in the pilot study and N = 200 prisoners in the validation study.

Measures. The present study aims to investigate the psychometric quality of the PID-5 (PID-5; Krueger, Derringer, Markon, Watson, & Skodol, 2012) and the imprisonment-appropriate version of the PID-5 in a sample of condemned prisoners. In addition we assess the Big Five (BFI-15; Gerlitz & Schupp, 2005) for investigating the construct validity. Additionally some information regarding the (mis-)conducts in prison are assessed for investigating the criterion validity. Finally, a number of contentual relevant personality measures (u.a. NARQ, Back et al., 2013; SD3, Jones & Paulhus, 2014) are assessed for exploratory analysis and a number of moderator and control variables (including demographics, migrational background, type of offence) are measured. The codebook to the study (also available via the projects OSF page: https://osf.io/eefvq/) contains an overview of all variables that will be assessed (including the content of each item and the response scale of each item).


Researchers & cooperation partners


OSF Link : https://osf.io/eefvq/