The Personality Assessment Inventory (PAI) is used in the assessment of a broad range of clinical variables and interpersonal functioning in clinical and forensic settings. It has been identified as the second most frequently utilized broadband instrument in the evaluation of adults by forensic psychologists and the third most frequently utilized self-report measure in the assessment of parents undergoing child custody evaluation (CCE). The child custody evaluation context tends to be susceptible to, or to “pull for” positive, self-favorable presentation on the part of parents during interviews and psychological testing. A review of the literature finds that, whereas on average almost all of the PAI clinical scales are at average levels or suppressed in the CCE context, several scales tapping more positive personal and interpersonal functioning tend to be moderately elevated. This paper focuses particularly on the significance of elevations on the MAN-G (Grandiosity) and ARD-O (Obsessive-Compulsive) subscales in contexts that pull for positive response distortion, such as the CCE context.

PAI: Personality Assessment Inventory; CCE: Child Custody Evaluation; MMPI-2: Minnesota Multiphasic Personality Inventory-2; MMPI-2-RF: Minnesota Multiphasic Personality Inventory-2-Restructured Form; MCMI-III: Millon Clinical Multiaxial Inventory-Third Edition; PIM: Positive Impression Management; DOM: Dominance; WRM: Warmth; NEO-PI: NEO Personality Inventory; MCSD-SF: Marlowe-Crowne Social Desirability-Short Form; MAN: Mania; MAN-G: Mania Grandiosity; ARD: Anxiety-Related Disorders

A dominant proportion of psychologists who conduct child custody evaluations (CCE’s) for the courts administer standardized, self-report personality assessment measures to parents [1-3]. Among such measures are the Minnesota Multiphasic Personality Inventory-2 (MMPI-2) [4], the Minnesota Multiphasic Personality Inventory-2-Restructured Form (MMPI-2-RF) [5], the Millon Clinical Multiaxial Inventory-Third Edition (MCMI-III) [6], which recently has been updated, and the Personality Assessment Inventory (PAI) [7,8]. Psychological testing, as one component of a CCE, may be useful in helping the evaluator with the assessment of a parent’s personality qualities or identification of acute psychopathology that may be relevant to issues in the realm of parenting, e.g., severe depression or severe anxiety disorders, hostility and anger with poor impulse control, antisocial personality characteristics, severe narcissistic qualities, severely detached or schizoid qualities, ability to form stable, warm relationships, misuse of alcohol or other substances, as well as in the assessment of how stress might be impacting the parent’s relationship with the child. In this sense, Graham [9] noted that an evaluator may form higher order inferences related to parenting based on the empirical associations between MMPI-2 scores and particular behavioral correlates.

Various researchers and investigators [10,11] have noted that whereas in the therapeutic context it is in the best interests of patients/clients to report their symptoms and problems as openly and candidly as possible, in the forensic context conscious, intentional distortion of information presented by the examinee, or positive response distortion, is much more likely to occur. In the forensic context the examinee may have much to gain or to lose from the outcome of his or her case.

In assessment contexts in which the finding of healthy psychological functioning, or at least the non-presence of psychological dysfunction, would serve the interests of the examinee, it is expected that examinees will try to put their best foot forward in giving a favorable impression of themselves. Certain forensic contexts, such as CCE’s, illustrate such an assessment context. Personnel screening assessment, although non-forensic, also illustrates such an assessment context.

As noted by Weiner & Greene [12], forensic psychologists who employ tests such as the MMPI-2 or the PAI should be familiar with the types of profiles one can expect to find that are particular to the type of forensic context. There have been published reports on MMPI-2 test scores (and, more recently, MMPI-2-RF test scores) in samples of parents undergoing CCE’s that have revealed certain consistent findings, on average, across those samples (e.g., tendencies toward unrealistic assertions of virtue, defensiveness and underreporting of problems, and suppression of scores on clinical or substantive scales). Such “fake good” or defensive profiles are common in the CCE context [13-15]. Positive impression management has a high base rate in the custody evaluation context [16].

The Personality Assessment Inventory is the second most frequently utilized multiscale psychological test instrument in forensic evaluation of adults [17] and the third most frequently used standardized self-report measure of personality and psychopathology in child custody evaluations [2]. Some studies have reported on positive response distortion using the PAI under simulated and natural assessment conditions, primarily in forensic contexts [7,18-20]. The initial study by Morey [7] of the Positive Impression Management (PIM) scale to detect underreporting on the PAI, utilized a simulated self-favorable test instruction condition with a sample of 45 college students. The mean profile of the study group reflected elevation on the PIM scale, suppression of the clinical scale scores, with the exception of a moderate elevation on the Mania (MAN) scale due to elevation on the MAN-G Grandiosity subscale, moderate elevation on the Treatment Rejection (RXR) scale, and moderate elevations on the two Interpersonal scales, i.e., Dominance (DOM) and Warmth (WRM). One study [21] examined test scores on several measures including the PAI in parents undergoing parenting capacity evaluations, that is, evaluation for possible termination of parental rights, which is different from child custody evaluation. Positive self-presentation was found across all of the measures and the different measures of self-presentation were all positively correlated with each other. However, only one published report to date provides data on PAI test scores in a sample of parents undergoing CCE’s [22]. The study by Hynan [22] yielded results that were fairly similar to the simulated study by Morey [7], i.e., moderate elevations on the PIM, RXR, and WRM scales. The MAN scale was not elevated, and Hynan did not report subscale scores. The DOM scale also was not elevated.

Although not involving a forensic context, a recent study by Kurtz et al. [23], with a large sample of college students, used both a natural/honest test condition and a role-play condition to simulate a job application. Under the role-play condition the respondents were instructed to respond to the test in such way that describes them “in the best possible manner”. The students took the test twice, under each condition. In comparison to the natural/honest condition, scores increased under the role-play condition on the Mania (MAN) clinical scale, the Grandiosity (MAN-G) subscale, the Obsessive-Compulsive (ARD-O) subscale of the Anxiety-Related Disorders (ARD) clinical scale, and on each the RXR, DOM, and WRM scales. These findings suggest that in contexts that “pull for” the presentation of very favorable psychological adjustment, one may expect, on average, to find moderate or higher elevations on the PIM scale, and moderate elevations on the MAN scale or MAN-G subscale, and on the RXR, DOM, and WRM scales. It is reasonable that in contexts in which persons are motivated to present well, their PAI test profiles would yield the impression that they are highly virtuous (PIM), free of psychological dysfunction (suppression of clinical scales with the possible exception of MAN), with a high level of self-esteem and confidence (MAN-G), with orderly and organized qualities (ARD-O), with leader-like abilities to be assertive, effective, and able to take charge (DOM), while also being warm, empathic, sympathetic and patient with others (WRM). Such persons would be thought to make good candidates for employment and to possess positive parenting qualities.

Kurtz et al. [23] did not entertain considerations to explain the direction of score changes for the MAN-G and ARD-O subscales. Highly elevated scores on these subscales may be associated with inflated self-esteem that borders on delusional (MAN-G), and with the failure of obsessional ideation defenses to control anxiety (ARD-O). Yet, at moderate levels these subscales may be associated with benign, if not positive, adaptive qualities, i.e., self-confidence and being orderly, detail-oriented and conforming.

An examination of the associations between the PAI scales and the NEO Personality Inventory (NEO-PI) [24], as reported in the PAI manual, can help to further elucidate the positive personality qualities that may be tapped by some of the items contained within the MAN-G and ARD-O subscales, particularly at moderate score elevations. The ARD-O subscale was seen to have a correlation of .42 with the conscientiousness scale of the NEO-PI in community adults. The MAN-G subscale had correlations of .54 and .44 with the Extraversion and Openness scales of the NEO-PI, respectively. Yet, both the overall clinical MAN and ARD scales had moderate correlations with NEO-PI Neuroticism facets. For example, correlations between ARD with Anxiety, Hostility, and Depression on the NEO-PI were .58, .37, and .57, respectively. For MAN, r = .44 and .35, with Hostility and Impulsiveness, respectively. In contrast, ARD-O had only a small correlation with Anxiety (.24) and negligible correlations with the remaining NEO-PI Neuroticism facets. MAN-G had correlations near zero with Hostility and Impulsiveness but correlations of -21 and -20 with Anxiety and Depression, respectively. Thus, in contrast to their parent clinical scales, ARD-O and MAN-G tend to be less strongly associated with, or negatively associated with neurotic personality qualities as measured on the NEO-PI.

An additional consideration is that, as per the PAI Manual, as reported for the Community Adults sample, both the ARD and MAN clinical scales had negative correlations of small magnitude with the Marlowe-Crowne Social Desirability-Short Form (MCSD-SF) [25], whereas the magnitudes of negative association with the MCSD-SF scale were of medium size for most of the other clinical scales. Clearly, MAN and ARD were not highly inversely associated with social desirability. Further, ARD-O had a negative correlation near zero with the MCSD-SF, while MAN-G had a negative, but negligible association with the MCSD-SF, indicating no significant association with social desirability. An alternative possible explanation is that both ARD-O and MAN-G contain some items that are susceptible to social desirability while overall these subscales are not associated with social desirability.

A further psychometric consideration that is relevant to understanding the significance particularly of the ARD-O subscale is its internal consistency, as measured by the alpha coefficient. As reported in the test manual, the alpha coefficient for ARD-O indicates poor internal consistency, even while alpha for the ARD scale is in the acceptable range. The mean inter item correlation for ARD-O (.14) was the second lowest of the clinical subscales. Thus, particularly on the ARD-O subscale, test items may be tapping different, less consistent aspects of the same intended construct. MAN-G had an acceptable level of internal consistency. It is noted, though, that with respect to test-retest reliability, ARD-O had acceptable reliability and MAN-G had good reliability.

Extending the above considerations one step further, it is my opinion that some of the items within both the ARD-O and MAN-G subscales tap into relatively more socially positive attitudes and behavior while others tap into more dysfunctional behavior and attitudes. From my currently unpublished data set (N = 51) of parents undergoing CCE’s who completed the PAI, I formed two subdivisions of the subscales for both ARD-O and for MAN-G. Each of the subdivisions contains 4 items, or half of the items on each of these subscales. The subdivision items were selected using rational consideration and judgment of relative degree of positive functioning vs psychopathology along a dimension or continuum of the constructs believed to be measured by ARD-O and MAN-G, i.e., obsessive and compulsive ideation, rumination and rigid behavior, inflated self-esteem. A comparison of mean scores for the ARD-O and MAN-G subdivisions, tentatively labeled as SD (Socially Desirable) and NSD (Non-Socially Desirable) subdivisions for the respective subscales, yielded significant differences (p < .0001) for both subscales. This suggests that, for this sample, items within both the ARD-O and MAN-G subscales could be sorted into relatively positive functioning groupings and more pathological groupings. However, as per the PAI manual, PIM had negative associations of medium magnitude with both the ARD and MAN scales (correlations with PIM were not reported for the clinical subscales). Additionally, the association between MAN-G and ARD-O in the community normative sample was of small magnitude which would contraindicate that these subscales share an association mediated possibly by social desirability.

The PAI is a very well-constructed multidimensional measure of psychological symptoms and problems in adults. There is strong evidence for reliable and valid results with this test which is used in various assessment contexts. In addition to validity scales and clinical scales, the PAI includes Treatment Consideration scales and Interpersonal Scales which may be relevant in different forensic contexts. For example, in the CCE context, a test profile suggesting that a parent responded similarly to others who are controlling and forceful while also distant, disinterested, and not warm in interpersonal relationships may have significance for the evaluation. The CCE context tends to “pull for” positive response distortion during clinical interviews and psychological testing. As with the MMPI-2, there now are some published data as to how parents, on average, respond to the PAI in the CCE context [22]. Although Hynan [22] did not find elevation on the MAN scale, and he did not report scores for the clinical subscales, other studies suggest it is not uncommon to find elevations on MAN, and particularly MAN-G, and in some cases on ARD-O in situations that pull for positive response distortion. This review suggested that some items contained within both the MAN-G and ARD-O subscales may be associated with positive functioning, and thus susceptible to positive response distortion in contexts where that is likely to occur, such as in CCE’s. Based on an unpublished data set, I found that these two subscales could be subdivided into two subsets, one possibly representing positive personality qualities and the other more pathological personality functioning. Further study along these lines would be useful. In my opinion, it also is strength of the PAI that the ARD-O and MAN-G subscales include a range of items that, at moderately aggregated levels suggest more positive personality functioning, but at highly elevated levels is associated with dysfunction. Indeed, for both the ARD-O and MAN-G subscales, the clinical range is not reached until the raw score reaches 17, which, essentially means that nearly six of the eight items are being endorsed in the keyed direction at the highest level. Evaluators may find it helpful to examine which particular items are endorsed when any scale or subscale is elevated. With respect to MAN-G and ARD-O, it may be helpful to distinguish between those items suggestive of neat, orderly, detail-oriented characteristics versus compulsive, phobic characteristics, and between items suggestive of feeling well accomplished, but not espousing ideas of grandeur or self-exaltation.

1. Ackerman MJ, Ackerman MC (1997) Custody evaluation practices: A survey of experienced professionals (revisited). Professional Psychology: Research and Practice 28(2): 137-145.
2. Ackerman MJ, Pritzl TB (2011) Child custody evaluation practices: A 20-year follow-up. Family Court Review 49(3): 618-628.
3. Bow JN, Quinnell FA (2002) A critical review of child custody evaluation reports. Family Court Review 40(2): 164-176.
4. Butcher JN, Graham JR, Ben-Porath, YS, Tellegen A, Dahlstrom WG, et al. (2001) MMPI-2 (Minnesota Multiphasic Personality Inventory 02). In: Manual for administration, scoring, and interpretation, University of Minnesota Press, Minneapolis, Minnesota, USA.
5. Ben-Porath YS, Tellegen A (2008/2011) MMPI-2-RF (Minnesota Multiphasic Personality Inventory-2 Restructured Form) In: Manual for administration, scoring and interpretation. University of Minnesota Press, Minneapolis, Minnesota, USA.
6. Millon T, Davis R, Millon C (1997) MCMI-III manual. (2nd edn), MN: National Computer Systems, Minneapolis, USA.
7. Morey LC (1991) The Personality Assessment Inventory professional manual. FL: Psychological Assessment Resources, Odessa, Ukraine.
8. Morey LC (2007) Personality Assessment Inventory (PAI): Professional manual. (2nd edn), FL: Psychological Assessment Resources, Lutz, USA.
9. Graham JR (2012) MMPI-2: Assessing personality and psychopathology. (5th edn), Oxford University Press, New York, USA, pp. 688.
10. Galatzer-Levy RM, Gould J, Martindale D (2009) From empirical findings to custody evaluations. In: Galatzer Levy RM, et al. (Eds.), The scientific basis of child custody decisions. (2nd edn), John Wiley & Sons Inc., Hoboken, New Jersey, USA, p. 1-47.
11. Melton GB, Petrila J, Poythress NG, Slobogin C (2007) Psychological evaluations for the courts: A handbook for mental health professionals and lawyers (3rd edn), Guilford Press, New York, USA.
12. Weiner IB, Greene RL (2008) Handbook of personality assessment. New Jersey: John Wiley & Sons, Inc., Hoboken, USA.
13. Bagby RM, Nicholson RA, Buis T, Radovanovic H, Fidler BJ (1999) Defensive responding on the MMPI-2 in family custody and access evaluations. Psychological Assessment 11(1): 24-28.
14. Bathurst K, Gottfried AW, Gottfried AE (1997) Normative data for the MMPI-2 in child custody litigation. Psychological Assessment 9(3): 205-211.
15. Butcher J (1997) Frequency of MMPI-2 scores in forensic evaluations. MMPI-2 News and Profiles 8(1): 2-4.
16. Morey LC, Hopwood CJ (2007) Casebook for the Personality Assessment Inventory (PAI): A structural summary approach. FL: Psychological Assessment Resources, Lutz, USA.
17. Archer RP, Buffington Vollum JK, Stredny RV, Handel RW (2006) A survey of psychological test use patterns among forensic psychologists. Journal of Personality Assessment 87(1): 84-94.
18. Cashel ML, Rogers R, Sewell K, Martin Cannici C (1995) The Personality Assessment Inventory (PAI) and the detection of defensiveness. Assessment 2(4): 333-342.
19. Morey LC, Lanier VW (1998) Operating characteristics for six response distortion indicators for the Personality Assessment Inventory. Assessment 5(3): 203-214.
20. Peebles J, Moore RJ (1998) Detecting socially desirable responding with the Personality Assessment Inventory: The positive impression management scale and the defensiveness index. Journal of Clinical Psychology 54(5): 621-628.
21. Carr GD, Moretti MM, Cue BJH (2005) Evaluating parenting capacity: Validity problems with the MMPI-2, PAI, CAPI, and ratings of child adjustment. Professional Psychology: Research and Practice 36(2): 188-196.
22. Hynan DJ (2013) Use of the Personality Assessment Inventory in child custody evaluation. Open Access Journal of Forensic Psychology 5: 120-133.
23. Kurtz JE, Henk CM, Bupp LL, Dresler CM (2015) The validity of a regression-based procedure for detecting concealed psychopathology in structured personality assessment. Psychological Assessment 27(2): 392-402.
24. Costa PT, McCrae RR (1985) The NEO Personality Inventory manual. FL: Psychological Assessment Resources, Odessa, Ukraine.
25. Reynolds WM (1982) Development of reliable and valid short forms of the Marlowe-Crowne Social Desirability Scale. Journal of Clinical Psychology 38(1): 119-125.