Collateral Report of Psychopathy: Convergent and Divergent Validity of the Psychopathic Personality Inventory-Short Form

Abstract

Psychopathy is a personality syndrome comprised of interpersonal, affective, and behavioral features that has emerged as a correlate of intimate partner violence perpetration. One commonly used self-report measure of psychopathy is the Psychopathic Personality Inventory-Short Form. The current study employed a multi-trait, multi-method approach to test convergent and discriminant validity of the measure in partner-violent couples by comparing males’ self-report of psychopathy to the informant report of their female partner (N = 114). It was hypothesized that the female partner-report of the male’s psychopathy would be highly correlated with the male report of his own psychopathy, thus providing evidence for the construct validity and interrater reliability of the PPI-SF. Analyses found that male and female reports were correlated significantly on the two major factors of the PPI-SF. Furthermore, the female-report explained a significant amount of variance over and above men’s self-report on PAI scales designed to indicate antisocial personality traits.

Introduction

Psychopathy is a personality disorder comprised of interpersonal, affective, and behavioral features. It is typically expressed through deception, manipulation, lack of empathy, lack of insight, inflated and arrogant self-appraisal, and other antisocial traits (Cleckley, 1941; Hare, 1996). Although relatively uncommon in the general population (1–2%), (Hare, 1996; Neumann & Hare, 2008), 15–25% of criminals meet criteria for psychopathy (Hare, 1996). Furthermore, psychopathy has emerged as a significant correlate of antisocial, impulsive, and violent behavior (Camp, Skeem, Barchard, Lilienfeld, & Poythress, 2013), including intimate partner violence perpetration (IPV; Costa & Babcock, 2008; Mager, Bresin, & Verona, 2014) as well as treatment failure and recidivism (Hemphill, Hare & Wong 1998; Costa & Babcock, 2008; Rock, Sellbom, Ben-Porath, & Salekin, 2012).

The most commonly used measure of psychopathy is the Psychopathy Checklist, Revised (PCL-R; Hare, 1991,2003), but it has been criticized for having a long and difficult administration (Lilienfeld & Andrews, 1996; Kastner, Sellbom, Lilienfeld & Fowler, 2006; Lilienfeld, 2012; Tonnaer, Cima, Sijtsma, Uzieblo, & Lilienfeld, 2012). Self-report measures of psychopathy have also been successfully devised and are widely used because of their comparative ease and speed in administration. Since one feature of psychopathy is deceit, the veracity of self-reports of psychopathy is suspect. Therefore, the current study aims to show validity for a self-report measure of psychopathy by testing whether intimate partners can provide collaborative reports of psychopathy and whether adding partner report improves construct validity.

Overview of the Psychopathic Personality Inventory (PPI)

Lilienfeld and Andrews (1996) developed the Psychopathic Personality Inventory (PPI), a 187-item self-report questionnaire measure of psychopathy, which was designed and validated using undergraduate samples. The PPI is comprised of eight subscales. Seven of the eight scales load onto two higher order factors: Fearless Dominance (PPI-I) and Impulsive Antisociality (PPI-II; Benning, Patrick, Hicks, Blonigen, & Krueger, 2003).¹ Stress Immunity, Social Potency, and Fearlessness comprise PPI-I, and Impulsive Non-conformity, Blame Externalization, Machiavellian Egocentricity, and Carefree Nonplanfulness comprise PPI-II. PPI-I assesses the affective-interpersonal traits of psychopathy, whereas PPI-II assesses the behavioral-lifestyle traits of psychopathy. These two factors are seen as orthogonal, with little to no overlap. The revised version (PPI-R; Lilienfeld & Widows, 2005) contains 154 items, was adjusted for cross-cultural use, and lowered the reading level. Support for the two-factor model was maintained in the revised version of the measure (Lilienfeld & Widows, 2005).

The Psychopathic Personality Inventory-Short Form (PPI-SF; Lilienfeld & Hess, 2001) was developed before the revised long form was finalized, and it is the target measure of the current study. It is comprised of 56 items derived from items on the original version of the measure (PPI; Lilienfeld & Andrews, 1996) that showed the highest loadings on PPI-I and PPI-II (Lilienfeld & Hess, 2001). While the long-form PPI-R (Lilienfeld & Widows, 2005) has been shown to be comprised of two higher-order underlying factors, the short-form has not necessarily evidenced the exact same factor structure (Smith, Edens, & Vaughn, 2011). However, the PPI-SF is highly correlated to the PPI (r = .90) and has shown high internal consistency (α = .85; Lilienfeld & Hess, 2001). Although there is some preliminary support for its construct validity (Kastner et al., 2012; Tonnaer et al., 2012), only a handful of published studies have utilized the PPI-SF in research on medication treatment (Dunlop, DeFife, Marx, Garlow, Nemeroff, & Lilienfeld, 2011), moral judgment (Gao & Tang, 2013), psychopathy subtypes (Lee & Salekin, 2010), empathy (Mullins-Nelson, Salekin, & Leistico, 2006), personality features (Vaughn, Litschge, DeLisi, Beaver, & McMillen, 2008), and cognitive dissonance (Murray, Wood, & Lilienfeld, 2012). It is surprising that few studies have examined psychometric properties of the PPI-SF given the support for the long form.

The long form (PPI) has shown adequate overall validities, reliabilities, and utility of its built-in validity scales, providing considerable empirical support for its use as self-report measure of psychopathy. The PPI has acceptable convergent validity with the PCL-R (r = .43; Poythress et al., 2009) and test-retest reliability (r = .95; Lilienfeld & Andrews, 1996), and is sensitive to detecting inconsistent responding (Nikolova, Hendry, Douglass, Edens, & Lilienfeld, 2012). The PPI and PPI-R have two built-in validity scales (Deviant Responding and Variable Response Inconsistency) to detect deviant responding and inconsistent responding (Lilienfeld & Andrews, 1996) that have shown acceptable sensitivity and specificity in detecting invalid reports (Anderson, Sellbom, Wygant, & Edens, 2012).

Despite the strengths demonstrated by the PPI, one major concern remains regarding the validity of the measure because it lacks corroboration from an alternative source. Namely, critics argue that core features of psychopathy, such as deception, manipulation, and lack of insight, make respondents intrinsically inadequate reporters of their own psychopathy. This argument suggests that psychopathic individuals are more likely to demonstrate positive impression management or malingering with their responses (for a review, see Lilienfeld, 1994; Lilienfeld & Fowler, 2006). Positive impression management, or social desirability, refers to the respondent’s likelihood to falsely present the self in a positive light. On the other hand, malingering refers to the respondent’s likelihood to falsely endorse psychopathology. Although a recent meta-analysis found that psychopathic individuals are willing and able to accurately endorse socially undesirable traits and are not necessarily inclined to engage in positive impression management (Ray, Hall, Rivera-Hudson, Poythress, Lilienfeld, & Morano, 2013), reporting biases on the PPI have not thoroughly been tested. Furthermore, these types of reporting biases have yet to be examined on the short-form.

One way to test the accuracy of psychopathy reports is to compare an individual’s responses to those of an informant using the same measure. This type of convergent validation (McCrae, 1982) can be achieved by correlating a respondent’s ratings with those made by an informant that knows the respondent well. It can be assumed that the informant’s report will not be influenced by positive impression management or malingering. Therefore, the self-report may be validated if it corresponds to the objective informant report. In contrast, differing reports may suggest deceitful target responding or biased informant responding.

Several studies have examined the comparison of self-report and informant report on general personality characteristics to validate non-pathological self-report measures (McCrae et al., 2004; Goma-i-Freixanet, Wismeijer, & Valero, 2005; De Vries, Lee, & Ashton, 2008), but there are few studies using criminal or pathological populations. Spouse reports were used to validate the Minnesota Multiphasic Personality Inventory-2 (MMPI-2; Hathaway, S. R. & McKinley, J.C., 1989), which suggests that collateral report is useful for validating personality measures of psychopathology. However, few studies extended the use of collateral report validation to measures of psychopathy.

To our knowledge, there are only three studies that compare self versus informant reports of psychopathic traits, and no studies have compared self versus informant reports on the PPI-SF. The first study found a small to moderate significant correlation (r = .32) between self-report and informant-report on an infrequently used psychopathy measure, the Psychopathy Q-Sort (Fowler & Lilienfeld, 2007). The remaining two studies showed moderate correlations between self and informant report on the PPI-R (r = .34 – r = .83; Jones & Miller, 2011; Miller, Jones, & Lynam, 2011), but these two studies were obtained from the same sample. Furthermore, the studies did not control for the type of target-informant relationship, which may confound the self-informant correlations. De Vries et al. (2008) found the informant to be a more accurate reporter of an individual’s personality when the pair was more closely acquainted, with friends, family members, and romantic partners respectively being more closely associated. Therefore, one would expect romantic partners to know the respondent well, and to provide an accurate report of the target’s personality characteristics. Partner-violent couples have also demonstrated moderate levels of concordance when comparing self- to spouse-report of behaviors (Burrus & Cobb, 2011).

The Current Study

To date, few studies have even examined the overall validity of the PPI-SF. No studies have tested the convergent validity of the PPI-SF using intimate partners of the target, and only a few studies have even compared self-report to a general informant-report of psychopathy. The current study will employ a multitrait-multimethod approach to test convergent and discriminant validity of the measure (Campbell & Fiske, 1959). Multiple traits will be evaluated, some of which are expected to correlate and others to be unrelated to psychopathy. Self- and partner-reports will be evaluated as multiple methods of reporting these traits. Validity can be established when there is agreement between two methods reporting the same trait.

The main research aim of the current study is to test the convergent validity of the PPI-SF (Lilienfeld & Hess, 2001), by comparing males’ self-report of psychopathy to the informant report of their female partner. We hypothesize that the female partner-report of the male’s psychopathy will be highly correlated with the male report of his own psychopathy on the PPI-SF, thus providing evidence for the construct validity and interrater reliability of the PPI-SF. Given the established features of psychopathy such as manipulation, positive impression management, propensity for violence, and recidivism (Cleckley, 1941; Hare, 1996; Hemphill, Hare & Wong, 1998; Costa & Babcock, 2008; Rock et al., 2012; Camp et al., 2013), it is expected that both self- and partner-reports of psychopathy will correlate positively with men’s perpetration of intimate partner violence, psychological abuse, and treatment resistance, negatively with impression management, and be unrelated to the use of positive communication skills.

A secondary aim of the current study is to test the incremental validity of adding the female report to her partner’s report of his psychopathy. We expect male- and female-reports to correlate on measures of violence, because these behaviors are concrete, quantifiable, and easily identified by both partners. However, there are other characteristics of antisocial behavior and psychopathy that are less concrete, like deceitfulness, impulsivity, a failure to plan ahead, and a lack of remorse. We expect that a person who knows the respondent well, namely, a romantic partner, will also provide an accurate report of these less concrete characteristics. We hypothesize that adding the female partner-report of the male’s psychopathy will explain significantly more of the variance in men’s antisocial personality but not in frequency counts of men’s overt violence. Men are assumed to be adequate reporters of frequency counts and severity of violence once they have self-reported any violence (Arias & Beach, 1987).

A third and final goal of this study was to determine the correlation between male and female’s PPI-I and PPI-II scores with various variables of interest. Male violence was measured by both male self-report and spouse-report. Male impression management was measured by the Impression Management and Self-Deceptive Enhancement scales on the Balanced Inventory of Desirable Responding (BIDR; Paulhus, 1991). Treatment rejection was measured by the RXR scale on the PAI. Antisocial personality was measured using the ANT scale on the PAI. We expected that higher levels of PPI-I and PPI-II would be associated with higher levels of both self- and spouse-reported violence. We also expected that psychopathy would be associated with impression management, antisocial personality, and treatment noncompliance. We predicted, however, that female report of male negotiation would not be significantly correlated with self- or spouse-report of either psychopathy factor, thus demonstrating divergent validity.

Method

Participants

Community participants were recruited as part of a larger study through local newspaper advertisements and flyers requesting “couples needed.” During a telephone screening interview to determine eligibility, female partners were administered a modified version of the Conflict Tactics Scale (CTS; Straus, 1979) by trained undergraduates. To be eligible to participate, couples had to be at least 18 years of age, married or living together as if married for at least 6 months, heterosexual, and able to speak and write English fluently. Female partners had to report at least two male-to-female violent acts in the past year (n = 95). A small sample of distressed but non-violent couples were also recruited (n = 19). Those with incomplete PPI-SF reports were excluded from the present study, leaving a total of 114 couples in the current analyses.

Overview of Procedures

Data was collected as part of a larger study in which male participants were assessed twice and female participants were assessed once with their partner. In the first assessment period, male participants were administered a series of pencil and paper questionnaires, including the Psychopathic Personality Inventory-Short Form (Lilienfeld & Hess, 2001). In the second assessment period, both male and female participants were administered questionnaires followed by two marital interaction tasks and individual interviews about their history of domestic violence. Women completed the PPI-SF, modified to report on her partner’s, not her own, psychopathic traits. Couples were then reunited for debriefing and payment. Couples were paid $90 to $100 for their participation in the two 3-hour assessment periods.

Measures

Although several measures were included in the original study, only the Psychopathic Personality Inventory-Short Form, Conflict Tactics Scale-2, Personality Assessment Inventory, and Balanced Inventory of Desirable Responding are discussed here as the others are not relevant to the study at hand.

Intimate Partner Violence

The Conflict Tactics Scale-2 (CTS-2; Straus, Hamby, Boney-McCoy, & Sugarman, 1995) was administered to both male and female participants to assess for physical assault, psychological abuse, and, for discriminant validity, prosocial negotiation skills. The CTS-2 is a 78-item self-report that assesses the frequency of physical, sexual, and psychological abuse both committed and received by intimate partners. Five scales measure negotiation, psychological aggression, physical assault, sexual coercion, and injury. Internal consistencies on the CTS-2 subscales ranged from .60 to .78. Only women’s report of men’s use of negotiation and perpetration of physical and psychological abuse were analyzed here.

Psychopathy

The Psychopathic Personality Inventory-Short Form (PPI-SF; Lilienfeld & Hess, 2001) is a 56-item self-report measure that is derived from the original full-length Psychopathic Personality Inventory (PPI; Lilienfeld & Andrews, 1996). The PPI-SF has an analogous theoretical structure to the PPI, with individual items loading onto eight subscales in a mutually exclusive fashion, although this structure has failed to garner empirical evidence. However, as discussed previously, evidence supports a two higher-order factor structure of the full length PPI (Benning et al., 2003) and a similar structure has been found in the short form version as well (Smith et al., 2011). The first factor, PPI-I, assesses the affective and interpersonal traits of psychopathy and has been found to correlate positively with negative mood states such as depression, and anxiety (Edens & McDermott, 2010). The second factor, PPI-II, assesses the overt behaviors related to psychopathy and has been found to correlate positively with physical and verbal aggression and substance use (Claes et al., 2009). The PPI-SF was administered to both male and female partners as a measure of the male partners’ psychopathic thoughts and behaviors. In the current study, the PPI-SF demonstrated adequate to good internal consistency (with alphas greater than .65).

Antisocial Personality

The ANT scale of the Personality Assessment Inventory (PAI; Morey, 1991) was used to measure antisocial personality characteristics. The ANT scale is comprised of 24-items that make up three subscales: ANT-A, which measures criminality, ANT-S, which measures stimulus seeking, and ANT-E, which measures interpersonal deficits that are often associated with antisocial personality, such as lack of emotion and empathy. Male ANT scale scores were predicted using two models: one adding men’s self-report of PPI-I and PPI-II factor scores, and one including both men’s self-report and women’s spouse-report of PPI-I and PPI-II factor scores. This was done to test whether women’s ratings of their partners’ psychopathy would add incremental prediction of ANT scores.

Treatment Rejection

The RXR scale of the Personality Assessment Inventory (PAI; Morey, 1991) is an 8-item measure assessing treatment rejection. Specifically, the scale taps into into the individual’s willingness and motivation toward treatment. Pearson correlations were used to determine the relationship between male scores on the RXR scale and both male and female reports of male psychopathy.

Reporting bias

The Balanced Inventory of Desirable Responding (BIDR; Paulhus, 1991) is a 40-item measure assessing two factors of social desirability: self-deceptive enhancement and impression management. The BIDR was administered alongside the PPI-SF to determine whether subjects were reporting on their psychopathic behavior accurately or engaging in positive impression management.

Analyses

Intraclass correlations (ICCs) were used in the present study to establish interrater reliability between male and female reporters (Shrout & Fleiss, 1979). As discussed in Gomá-i-Freixanet & Wismeijer (2005), traditional Pearson correlations are not appropriate in determining convergent validity between self- and partner-report as they primarily measure the strength of a linear relationship. More specifically, self- and partner-reports that are consistently different will have a high Pearson correlation even though the scores on each item themselves are different between reporters. On the other hand, ICCs take into account the magnitude of respondents’ answers in comparison to one another, thus determining how similar respondents’ answers are above and beyond a simple one-to-one linear relationship. This technique therefore prevents inaccurate inflation of these correlations.

ICCs between male and female reporters were calculated for each of the eight theoretical PPI-SF subscales as well as the two higher-order PPI-I and PPI-II factors. A correlation matrix of ICCs was constructed to determine convergent and discriminant validity. The correlations on the principal validity diagonal should be larger than those on off-diagonals. Also included in this correlation matrix are the relationships between scores on the two higher-order factors of the PPI-SF, PPI-I and PPI-II. Additionally, a Pearson correlation was computed comparing partner-report to male’s self-report of violence. Finally, two stepwise multiple regressions were conducted to assess for incremental validity of women’s report of men’s psychopathy in predicting men’s antisocial personality and physical assault frequency. Men’s PPI-I and PPI-II self-reported psychopathy scores were entered in step one, followed by women’s reports of men’s PPI-I and PPI-II psychopathy scores in step two, to predict first Antisocial Personality as measured by the PAI-ANT scale and then men’s frequency of physical violence perpetration as measured by the CTS-2.

Results

Means and standard deviations for each group’s response on each subscale and factor are provided in Table 1. Additionally, unpaired t-tests were used to compare male and female means for each subscale and factor score. The Type 1 error rate per comparison was held standard at .05, and a Bonferroni correction of .05/8 = .00625 was used to correct for multiple comparisons of subscales and .05/2 = .025 was used to correct for multiple comparisons of factor scores. Higher subscale and higher-order factor scores are associated with greater psychopathic tendencies. Given that higher ICCs directly correspond to lower t-values (i.e., lower differences between group means), it was expected that subscale score differences between males and females would generally be insignificant. Females reported significantly higher scores than males on three subscales, (Machiavellian Egocentricity, Coldheartedness, and Carefree Nonplanfulness) whereas males reported significantly higher scores than females on one subscale (Stress Immunity). Additionally, women reported that their partners were significantly higher on PPI-I than did men but there was no significant difference between the men and women’s reports on PPI-II scores. ICCs between men and women are presented in Table 2. PPI-I scores of men and women were moderately correlated as were PPI-II scores of men and women. Seven of the eight subscales were also significantly correlated between male and female scores with correlations ranging from r = .31 to r = .62, thus providing evidence for the consistency in measurement of psychopathy between partners. Male and female reports of male IPV were also moderately correlated (r = .43, p < .01).

Table 1.

Means and Standard Deviations for Men’s Self-Report and Women’s Collateral Report on PPI-SF Subscale and Factor Scores and t Test Comparisons

	Men		Women
Subscale	M	SD	M	SD	t
Machiavellian Egocentricity	14.21	3.97	16.00	4.86	−3.06*
Social Potency	19.84	3.94	20.65	4.47	−1.45
Coldheartedness	16.42	3.09	17.74	3.69	−2.94*
Carefree Nonplanfulness	12.58	3.23	15.80	4.55	−6.17*
Fearlessness	17.17	5.24	17.36	5.03	−.37
Blame Externalization	16.95	5.43	18.50	5.34	−2.18
Impulsive Nonconformity	15.80	4.97	16.05	3.77	−.42
Stress Immunity	21.32	3.77	18.21	4.62	5.57*
PPI-I	59.54	11.23	66.37	12.46	−4.34*
PPI-II	58.26	8.79	56.22	9.39	1.70

Note.

^*p < .05.

Table 2.

Intraclass Correlation Matrix for Self- and Partner-Reports on PPI-SF Subscales

	Spouse-Reports
Subscale	Mac- Ego	Soc- Pot	Col	Car- Non	Fea	Bla- Ext	Imp- Non	Str- Imm	PPI- I	PPI-II
Self-Reports
Machiavellian Egocentricity	.06	.08	.22	−.04	−.05	.24	−.05	.18	.07	.08
Social Potency	.32*	.53***	−.02	.03	−.08	−.16	.08	.20	.07	.26
Coldheartedness	.31*	−.15	.41**	.14	−.17	.25	−.33	.05	.12	−.09
Carefree Nonplanfulness	−.07	−.71	.09	.35*	−.03	.02	−.07	−.27	.06	−.29
Fearlessness	−.39	.06	.06	−.02	.48***	−.02	.26	.24	−.04	.35*
Blame Externalization	.22	−.47	−.08	.08	−.26	.62***	.02	−.41	.32*	−.52
Impulsive Nonconformity	−.12	−.28	.03	−.15	.24	.15	.54***	.14	.13	.08
Stress Immunity	.03	.36*	.18	−.09	−.05	−.09	.008	.31*	−.03	−.13
PPI-I	.05	−.34	.05	.06	−.01	.39**	.17	−.08	.35*	−.30
PPI-II	−.09	.36**	.08	−.03	.24	−.11	.15	.30*	−.01	.48***

Note.

^*p < .05,

^**p < .01,

^***p < .001

Table 3 provides correlations between male and female PPI-I and PPI-II factors with various variables expected to reveal convergent and discriminant validity. Both female-reported PPI-I and PPI-II scores were significantly correlated with women’s reports of men’s physical and psychological violence. Neither PPI-I nor PPI-II was significantly correlated with women’s reports of men’s violence when using men’s self-report of psychopathy. Men’s self-deceptive enhancement was significantly negatively correlated with men’s self-report of PPI-I scores and significantly positively correlated with male-report of PPI-II scores. Self-deceptive enhancement was also significantly positively correlated with female-report of PPI-II scores. Similarly, men’s impression management was significantly negatively correlated with his self-report of PPI-I scores but significantly positively correlated with his PPI-II scores. However, men’s impression management was unrelated with women’s collateral report of psychopathy. This pattern was seen again when examining the relationship between factor scores and treatment rejection (the RXR scale of the PAI): Men’s self-report of PPI-I was significantly negatively correlated with RXR but positively correlated with PPI-II. In terms of women’s report, only their report of men’s PPI-I scores were significantly negatively correlated with RXR. Self-report of PPI-I and PPI-II were unrelated to couples use of negotiation. Additionally, partner-report of PPI-I and PPI-II were also, as expected, unrelated to couples use of negotiation, demonstrating discriminant validity.

Table 3.

Pearson Correlations of Male-Self and Female Collateral Report of PPI-SF Factors with Men’s Behavior

	M PPI-I	F PPI-I	M PPI-II	F PPI-II
Variable
Female report of male physical abuse	.16	.19*	−.08	−.21*
Female report of male psychological abuse	.10	.30**	−.04	−.23*
Male self-report of physical abuse	.40**	.12	.04	.01
Male self-report of psychological abuse	.22*	.16	.05	.10
Self-deceptive enhancement (BIDR)	−.32**	−.03	.53***	.31**
Impression Management (BIDR)	−.44***	−.14	.24*	.14
Treatment compliance (PAI-RXR)	−.32**	−.22*	.38***	.15
Composite score of men’s negotiation	−.01	.07	−.04	.07

Note.

^*p < .05,

^**p < .01,

^***p < .001

Finally, multiple regression was used to assess if women’s report of men’s psychopathy adds incremental validity in the prediction of men’s 1) antisocial personality and 2) intimate partner violence. The first step included only male report of PPI-I and PPI-II, whereas the second step included both male and female report of both psychopathy factors. Table 4 provides the results of the first multiple regression predicting male’s PAI-ANT scores from men’s and women’s reports of men’s psychopathic features. Although the model predicting antisocial personality from only male report of psychopathy was significant, adding female report of male’s psychopathy to this model explained significantly more variance in ANT scores, raising the total R² to .509 (p < .0001). The second multiple regression, predicting men’s perpetration of IPV as reported by the female partner, was not significant. This suggests that adding women’s collateral report adds incremental validity in predicting men’s antisocial personality but not intimate partner violence.

Table 4.

Summary of Multiple Regression Analyses for PAI Antisocial Personality Score

Variable		β	p	R2	R2 Change	F Change
Model 1	MPPI-I	.682	<.0001	.47	.47	49.10***
	MPPI-II	.208	.004
Model 2	MPPI-I	.684	<.0001	.51	.04	3.84*
	MPPI-II	.142	.052
	FPPI-I	.099	.193
	FPPI-II	.214	.007

Note.

^*p < .05,

^**p < .01,

^***p < .001

Discussion

The current study attempted to utilize the Psychopathic Personality Inventory-Short Form for two main purposes. First, we sought to validate a brief self-report measure, which could provide for more efficient yet valid assessment of psychopathy. In lieu of criminal records as corroborating evidence of psychopathic behavior, as is required in the administration of the PCL-R, spouse-reports were tested to convergently validate the PPI-SF. A second aim was to test the incremental validity of adding a female partner-report to the male respondent’s self-report. The PPI-SF self- and spouse-report were moderately correlated with each other and correlated as expected with other measures, further validating the measure. Furthermore, the female partner-report explained a significant amount of variance over and above men’s self-report on PAI scores designed to indicate antisocial personality traits.

Analyses found that male and female reports were correlated significantly on reports of male IPV as well as on the two major factors of the PPI-SF. The PPI-I may be understood as the emotional–interpersonal facet of psychopathy that reflects psychopathic traits related to dominance, low levels of anxiety, and recklessness. The significant male-female correlation on this facet implies that spouses were able to gauge their partner’s disposition to a significant extent. The correlation between male and female reports was also significant for the PPI-II factor. PPI-II assesses psychopathic traits that are expressed externally and overtly, such as antisocial behaviors aimed toward others. As expected, the target’s spouse was able to accurately detect behaviors that are more overt and manifested externally, in comparison to thoughts, feelings and internal traits representative of psychopathy.

With regard to the utility of partner report, our analyses found that women’s reports on the PPI were significant predictors of their reports of men’s physical and psychological abuse towards them. This may partly be due to shared method variance or criterion contamination since females were all victims of abuse and may not be representative of female partners in general. Further, only female-report of violence was used to avoid impression management by male IPV perpetrators, which could also contribute to shared method variance. Typically, studies of IPV correlate perpetrators’ self-report of personality traits with victims’ reports of violence, but this method introduces method variance that may obscure results. Furthermore, women’s report on the PPI- I were also significant predictors of a PAI variable that is designed to predict treatment compliance. Higher scores of their partner’s dominance and recklessness related with lower treatment compliance, as expected.

Analyzing male reports of their own psychopathic tendencies provided a similar profile to that of female-report for these variables. According to male report, higher levels of dominance, and recklessness, and low trait anxiety were associated with higher levels of impression management as observed through its relationship with PPI-I. This may suggest that the low anxiety component of PPI-I could be facilitating faking responses in a biased manner so as to make a favorable impression to the researcher. However, male reports of their psychopathy on PPI-I and PPI- II differentially predicted their own treatment compliance. PPI-I, which is associated with high dominance, low anxiety, and high levels of recklessness, was negatively related with treatment compliance. This could suggest that individuals with high PPI-I scores exhibit impulsivity and a low tolerance for treatment that may challenge one’s dominant nature. PPI-II, which is associated with general antisocial tendencies, positively predicted treatment compliance. It is unclear why this factor is positively associated with treatment compliance, but one possibility is that these individuals have more experience dealing with the legal and social consequences of their antisocial behaviors and thus they are more willing to admit to these problematic personality features and seek out treatment.

Furthermore, analyses provided evidence for the incremental validity of the female partner’s report over the male’s self-report on measures of antisocial traits. The PAI-ANT scale, which was used as a measure of antisocial traits in the current study, has been shown previously to correlate significantly with PCL-R scores (Edens et al., 2000; Kucharski et al., 2008); this demonstrated relationship provides further evidence for the validity of the PPI-SF. Considering that the analogous test of adding the female-report to the male-report did not significantly improve the prediction of violence, perhaps the additional variance accounted for by the wife’s report explains antisocial traits that do not include overt violence. Specifically, the female partner may haves additional insight to facets of psychopathy that are not necessarily readily observable, such as deceitfulness, a failure to plan ahead, and a lack of remorse, that are not captured by the men’s self-report.

These results may clarify some disparities between the opinions of critics and proponents for psychopathy self-report measures. Our results suggest that men may be accurate reporters of their own psychopathy, but their partners may be better reporters of other antisocial characteristics. Results from the current study provide evidence for the convergent validity of the PPI-SF as a self-report measure of psychopathy, and that adding partner-reports of their partners’ psychopathy may be important for more accurately capturing personality characteristics that self-reports miss.

Convergent validity was established using self- and informant-report on the PPI-SF. Together, this suggests that the PPI-SF may be a useful and reliable tool for accurate assessment of psychopathic behavior as manifested in IPV couples. Although interviewing for psychopathic traits remains the gold standard approach, it is also time consuming and sometimes not feasible. In such circumstances, the PPI-SF may be a useful option.

Limitations and Future Directions

There were several limitations in the current study. The majority of this sample consisted of partner-violent couples, and only partner-report of male-perpetrated IPV was used in analyses to reduce impression management. Therefore, these results may not be representative of a more general couple population. In criminal justice settings, where perpetrators may benefit by denying their psychopathic tendencies, there may be a greater disparity between self and partner report. Because this sample did not consist of perpetrators facing legal action, there may have been more accurate responding from both male and female reports as there were no consequences for detection of psychopathic personality. It is also important to note that this study only examined how well partner- and self-reports of male psychopathy correlated. As such, these results cannot be generalized to partner- and self-reports of female psychopathy.

Future studies may further test the utility of collateral reports and the PPI-SF. First, the accuracy of incremental validity of the partner report above and beyond that of the perpetrator could be studied in a more representative severe criminal sample in which the validity of self-report may be more questionable. Future studies may also compare self- and partner-report of psychopathy against other corroborating measures of the perpetrator’s psychopathy, such as criminal records, to determine the most accurate way of assessing for psychopathy. Furthermore, future studies could extend the current study’s convergent validation of the PPI-SF to include other psychopathy measures, such as the PPI-R and the PCL-R. This study utilized a predominantly violent community sample to test the validity of the PPI-SF. In contrast, future studies could test the validity of this measure on a predominantly nonviolent sample.