Psychopathic Personality Inventory-Revised (PPI-R)

The Psychopathic Personality Inventory (PPI-Revised) is a personality test for traits associated with psychopathy in adults. The PPI was developed to assess these traits in non-criminal (e.g. university students) populations, though it is still used in clinical (e.g. incarcerated) populations as well. The PPI is a self-report assessment. It is intended to comprehensively index psychopathic personality traits without assuming particular links to anti-social or criminal behaviors. It also includes measures to detect impression management or careless responding.

Test review of Psychopathic Personality Inventory-Revised
DeMauro, G. E. (2007).
In K. F. Geisinger, R. A. Spies, J. F. Carlson, & B. S. Plake (Eds.), The seventeenth mental measurements yearbook. Retrieved from http://marketplace.unl.edu/buros/

DESCRIPTION
The self-report Psychopathic Personality Inventory-Revised (PPI-R) is a revision of the PPI designed to reduce reading load, eliminate cultural idioms, reword or eliminate items with poor psychometric properties, and develop new items borrowed (with permission) from the Multidimensional Personality Questionnaire (Tellegen, 1978/1982) for the Stress Immunity and Unlikely Virtues validity scales. The Total PPI-R score is a combination of eight constituent content scales: Machiavellian Egocentricity, Social Influence (called Social Potency on the PPI), Coldheartedness, Carefree Nonplanfulness, Fearlessness, Blame Externalization (called Alienation on the PPI), Rebellious Nonconformity (called Impulsive Nonconformity on the PPI), and Stress Immunity. Three factor scores are also available: Self-Centered Impulsivity, Fearless Dominance, and Coldheartedness (the same as the content score).

DEVELOPMENT
The PPI development was based on a series of considerations designed to sample the domain: (a) Dimensions of psychopathy were delineated avoiding criminal antisocial behaviors; (b) the test was structured to minimize socially desirable or undesirable responding; and (c) a 4-point Likert-type format was used to eliminate central tendency in option selection.

Each stage of PPI test development was guided by principal components factor analyses with orthogonal (Varimax) and oblique rotations. Items were decomposed to the eight scales of the PPI-R. The Recaptured Item Technique (Meehl, Lykken, Schofield, & Tellegen, 1971) provided an independent validation of the factor labels. Exploratory factor analyses of PPI-R data generated by the norming sample, using both Promax and Varimax rotations, decomposed the scores to three superordinate factors accounting for 47% of the variance: Self-Centered Impulsivity, Fearless Dominance, and Coldheartedness.

The three rounds of PPI test development used samples of 241 male students (mean age = 20.4 years), 253 male students (mean age = 20.8 years), and 610 students consisting of 249 men (mean age = 21.7 years) and 361 women (mean age = 22.1 years). The raw scores were standardized within gender in the final round to avoid spurious effects in factor analysis.

Each round of development retained items that loaded .3 or higher on any factor or on the Total score, using principal components analyses. Items with slightly lower correlations were rewritten. Test development proceeded through four stages: (a) Of the original 136 items, 28 were deleted and 7 were rewritten; (b) additional items were written bringing the total to 188, 35 of which were deleted and 6 rewritten; (c) in the third round, 210 items were administered, 45 were deleted, and 5 new items were added to improve coverage of the domain; and (d) factor analyses identified eight content scales and three validity scales for 187 items.

An embedded 10-item Deviant Response (DR) scale was used to identify aberrant response patterns and eliminated between 3.4% and 5.4% of the respondents. A Variable Response Inconsistency scale was developed in the third round of test development and used to eliminate respondents from validity studies. An embedded 14-item Unlikely Virtues scale (see Tellegen, 1978/1982) was used as a variable in later analyses.

After eliminating respondents for validity scale anomalies, the tryout sample was 683 undergraduates, 256 men and 427 women. Principal axis factor analysis with Promax rotation identified eight factors. Items were retained if they loaded .15 or higher on the Total score and .3 or higher on a content scale. Some items with slightly lower correlations were retained to improve domain coverage. A 40-item Inconsistent Responding scale was developed to augment the 15-item scale and increase internal consistency.

TECHNICAL

Reliability
Composite reliability coefficients, for the norming sample of 985 community and college examinees and for 154 inmate offenders, are reported for Total score (.93 and .86, respectively), the Self-Centered Impulsivity factor score (.92 and .91, respectively), and the Fearless Dominance factor score (.91 and .76, respectively). Test-retest and alpha coefficients are also reported for Total score and the eight content scales. Test-retest coefficients for 51 members of the norming sample over an interval ranging from 12 to 45 days ranged from .82 for Coldheartedness to .95 for Social Influence. The coefficients for the Self-Centered Impulsivity and Fearless Dominance factor scores and for the Total score were .92, .95, and .93, respectively.

Alpha coefficients for written (.95) and computerized (.92) forms of the PPI-R are reported (Sandler, 2004). Content scale coefficients for these two forms ranged from .75 to .90. Over a 3-week interval, the test-retest coefficients were .94 (written then computerized) and .90 (computerized then written), and the 16 content scale coefficients ranged from .66 to .91, with 13 values at or above .80.

Validity
Abundant validity evidence is reported to address discriminant and convergent properties of the PPI-R test scores. Studies of the PPI also support the construct evidence of validity, but these are not summarized here for the sake of brevity. The greatest obstacle to interpreting these findings is the multidimensionality of the psychopathy construct. Among the eight content scales, only Machiavellian Egocentricity and Rebellious Nonconformity yielded as many as four correlations of .3 or higher with the other seven scales. Blame Externalization, Carefree Nonplanfulness, and Fearlessness yielded two of seven correlations greater than .3. Social Influence, Stress Immunity, and Coldheartedness were not correlated .3 or higher with any of the scales. Stress Immunity had negative correlations with four of the other seven scales. This raises serious questions about the interpretation of a Total score.

Sources of validity evidence include correlations of the PPI-R Total, content, and factor scores with two global measures of psychopathy, Levenson’s Self-Report Psychopathy Scale (Levenson, Kiehl, & Fitzpatrick, 1995) and the Self-Report Psychopathy Scale-II (SRP-II; Hare, 1991) and with two personality measures, the NEO Five Factor Inventory (NEO-FFI; Costa & McCrae, 1992) and the Personality Assessment Inventory (PAI; Morey, 1991). Correlations also are reported with measures of other constructs.

Based on the community/college and inmate samples, all eight content scales were significantly related to the SRP-II Total score. Seven of the scales were related to Levenson’s Primary Psychopathy, SRP-II Factor 2, the Anti-Social Personality Disorder scale of the OMNI-IV Personality Disorder Inventory (Loranger, 2001), a measure of pathology. Six of the eight scales were related to Levenson’s Secondary Psychopathy score, and four of the scales were related to the SRP-II Factor 1.

The PPI-R Total score was related to PAI Mania, Antisocial Features, and Aggression. Each of the eight content scores, as well, was positively or negatively related (r > .35) to those component scores of these personality measures consistent with theory.

The Self-Centered Impulsivity factor score was negatively related (r < -.35) to NEO-FFI Agreeableness and Conscientiousness and to PAI Treatment Rejection. It was positively related (r > .35) to 11 of the 18 PAI scales. The Fearless Dominance factor score was positively related to NEO-FFI Extraversion and negatively related to PAI Somatic Complaints, Anxiety, Anxiety-Related Disorders, and Depression. It was also negatively related to NEO-FFI Neuroticism. Convergence of PPI-R scores with scores on other instruments and with measures of behavior provides evidence in support of the construct validity of the PPI-R. The interpretation of the correlational studies could be facilitated by providing reliability estimates for the scores correlated and also providing the intercorrelations of external measures to enable an evaluation of their dimensionality.

Evidence supporting the use of the validity scales includes scoring differences in relation to response instructions, and correlations with corresponding scales of the PAI. Randomly completed protocols exceeded recommended cutoff scores on the 15-item (72.5%) and 40-item (78.5%) Inconsistent Responding scales.

Standardization
The 683 undergraduate item tryout protocols were supplemented with 329 adults who were representative of the 2002 U.S. population in race/ethnicity and education. Of these, 27 were eliminated for missing demographic information or extreme validity scale values, bringing the final sample to 985. An offender sample of 154 men also was tested to provide clinicians with comparative data.

Final item selection employed the same factor analytic criteria described above. The final PPI-R version consisted of 154 items in eight content scales and four validity scales.

Separate norms were developed by gender for the community sample in response to higher scores obtained by males. Within-gender norms also were developed for ages 18-24, 25-29, 30-39, 40-49, 50-59, and 60 years or older. Because there were no demographic differences in the offender sample, only one set of norms was developed for this group.

COMMENTARY
The validity argument rests most heavily on the convergent and discriminant properties of an instrument designed to assess a multidimensional construct. The dimensions of the construct, in turn, represent characteristics held important in the literature. Greater precision might be realized through profile analyses relating ranges of scores across Content areas to a theoretical framework. Similarly, although the test authors discourage the use of a single cutoff score, it is clear that reference to an informal normative cutoff score is common. Another approach is to set standards to describe the ranges of performance based on theory. This technique is common in educational achievement testing and could increase the interpretability of scores. It offers the advantage, as well, of defining the degree of psychopathy in criterion, rather than normative terms.

SUMMARY
The PPI-R has been developed through careful review of the theoretical bases for a multidimensional aspect of personality. The psychometric consequences are large and difficult. Nevertheless, the instrument provides enough support for interpreting scores to make it a valuable tool for counseling and research. Some suggestions are offered to enable the instrument to provide more criterion-related information.

REVIEWER’S REFERENCES

• Costa, P. T., Jr., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO PI-R) and NEO Five-Factor Inventory (NEO FFI): Professional manual. Odessa, FL: Psychological Assessment Resources.
• Hare, R. D. (1991). The Self-Report Psychopathy Scale-II. Unpublished test, University Of British Columbia, Vancouver, Canada.
• Levenson, M. R., Kiehl, K. A., & Fitzpatrick, C. M. (1995). Assessing psychopathic attributes in a noninstitutionalized population. Journal of Personality and Social Psychology, 68, 151-158.
• Loranger, A. W. (2001). OMNI Personality Inventories professional manual. Lutz, FL: Psychological Assessment Resources, Inc.
• Meehl, P. E., Lykken, D. T., Schofield, W., & Tellegen, A. (1971). Recaptured-Item Technique (RIT): A method for reducing somewhat the subjective element in factor naming. Journal of Experimental Research in Personality, 5, 171-190.
• Morey, L. C. (1991). The Personality Assessment Inventory professional manual. Odessa, FL: Psychological Assessment Resources.
• Sandler, J. C. (2004). Computer equivalency of the psychopathic Personality Inventory-Revised in a nonincarcerated population. Unpublished master’s thesis, Castleton State College, Castleton, VT.
• Tellegen, A. (1978/1982). Brief manual for the Multidimensional Personality Questionnaire. Unpublished manuscript, University of Minnesota.