NPI Factors and 16-Item NPI Rosenthal, Montoya, Ridings, Rieck, and Hooley (2011) Table 2 examines the relation of four sets of published factor-based NPI subscales and a shortened version of the NPI to the criterion variables investigated in the current study. It includes correlations of the NPI subscales developed by Emmons (1987), Raskin and Terry (1988), Corry, Merritt, Mrug, and Pamp (2008), and Ackerman et al. (2011), and the 16-item version of the scale (Ames, Rose, & Anderson, 2006) with self-esteem, grandiose narcissism, psychological health, and aggression/anger. Results indicated that all 16 NPI subscales were positively correlated with the grandiose narcissism benchmark, although one failed to meet Cohen's (1988) conventional definition of a "small" correlation (i.e., z.10). This demonstrates that the subscales all converge with the narcissism construct to some degree. For self-esteem, 11 NPI subscales had a positive correlation.10. However, two NPI subscales were negatively correlated with selfesteem, one subscale was uncorrelated with self-esteem, and two subscales had a positive, but trivial (i.e., <.10) correlation. In terms of psychological health, nine NPI subscales had a positive correlation.10. However, three subscales were negatively correlated with psychological health, two were uncorrelated, and two shared a positive, but trivial correlation with psychological health. Finally, for aggression/anger, nine NPI subscales shared a positive correlation. However, one subscale was negatively correlated, two were uncorrelated, and four shared trivial correlations with aggression/anger. Overall, we believe that a function-based (i.e., IRT-based) analysis of the NPI offers a more precise method for assessing the validity of subgroups of the scale's items than does factor analysis. However, inspection of the correlations between the factor-based subscales 1
and the criterion variables also provides an informative means for investigating the better and poorer aspects of the NPI's validity. The final column of Table 2 presents correlations between the Ames et al. (2008) 16- item version of the NPI and the criterion variables. Compared to the full NPI, the 16-item version had a similar correlation with grandiose narcissism. The 16-item version also had weaker (although still positive) relations with self-esteem and psychological health, and a stronger relation with aggression/anger than did the full NPI. This suggests that the 16-item NPI may offer minor advantages over the full NPI in terms of discriminant and construct validity. Five Factor Model of Personality Miller, Maples, and Campbell (this issue) present correlations of the Five Factor Model of personality (FFM) with the NPI and its NPI-X and NPI-N subscales. Although not entirely applicable to the NPI's relation to psychological health, the FFM-based results in Miller et al.'s two datasets largely support their contention that the differences between the NPI-X and NPI-N subscales are a matter of degree. We investigated the relation of the NPI and its subscales to the FFM using eleven datasets (N = 29,587). We produced a composite FFM variable using the NEO Personality Inventory (Costa & McCrae, 1985), the Revised NEO Personality Inventory and NEO Five-Factor Inventory (Costa & McCrae, 1992), the Big Five Inventory (see Benet-Martinez & John, 1998; John & Srivastava, 1999), and the Big Five Aspects Scale (DeYoung, Quilty, & Peterson, 2007). Table 3 reports correlations of the composite NPI, Rosenthal and Hooley's (2010) NPI subscales, and grandiose narcissism (see Samuel & Widiger, 2008a) with the FFM. The NPI-N subscales had a stronger negative relation with Agreeableness than did the NPI-X 2
subscales. Conversely, the NPI-X subscales had a stronger positive relation with Extraversion, and a stronger negative relation with neuroticism than did the NPI-N subscales. The NPI-X subscales were also more strongly related to Conscientiousness and Openness to Experience than were the NPI-N subscales, although there are generally no specific hypotheses regarding the relation of narcissism to these two FFM factors. In the context of the current discussion, the most notable FFM finding is the significant, but trivial negative correlation between the IRT-X subscale and Agreeableness (z =.04). This contrasts starkly with the robust negative correlations between the two variables (rs =.40,.45) reported by Miller et al. (this issue). Low Agreeableness is widely regarded as a hallmark of narcissism. This is true not only in a clinical context (see Lynam & Widiger, 2001; Miller, Reynolds, & Pilkonis, 2004; Samuel & Widiger, 2006), but also in the context of "normal" narcissism as operationalized by the NPI (see Miller & Campbell, 2008; Paulhus, 2001; Samuel & Widiger, 2008b). In fact, the level of narcissistic (dis)agreeableness assessed by the composite NPI has been found to rival that assessed by a clinical measure of DSM IV NPD (Miller & Campbell, 2008). In comparison, the near-zero relation in this large dataset between the IRT-X subscale and (dis)agreeableness suggests that, in contrast to Paulhus's (2001) terse description of narcissists as "disagreeable extraverts" (p. 228), individuals who endorse the IRT-X items might be more aptly described simply as "extraverts." It is important to note, however, that the operational definitions of Agreeableness used by NEO-based FFM measures (Costa & McRae, 1985, 1992) are more closely tied to the NPI than are the operationalizations of Agreeableness used by other FFM measures (see Miller, Gaughan, Maples, & Price, in press). This observation was borne out for the IRT-X subscale 3
in our meta-analytic results. The correlation between IRT-X and NEO-based Agreeableness was significant (z =.25, k = 4), whereas the correlation between IRT-X and the other measures of Agreeableness was not (z =.02, k = 7). Further, the NEO-based correlation was stronger than was the non-neo-based correlation, χ 2 (1) = 10.90, p <.05. This indicates that the robust negative correlations between IRT-X and NEO-based Agreeableness reported by Miller et al. (this issue) do not fully conflict with our meta-analytic results. However, it is not entirely clear which operational definition of Agreeableness (i.e., NEO-based versus non- NEO-based) most closely corresponds to the underlying latent construct. It is clear that the NEO-based definition of Agreeableness is not identical to the definition on which most other measures of Agreeableness are based (Miller et al.). Accordingly, the overall trivial relationship between the IRT-X subscale and (dis)agreeableness found in the meta-analysis may provide further evidence that the IRT-X items are not optimal as measures of narcissism, particularly in their relation to important criterion variables. 4
Table 2. Correlations between criterion variables, factor-analyzed NPI subscales, and the 16-item NPI, Appendix A. 7-Factor (Raskin & Terry, 1988) k N Self-Sufficiency Authority Superiority Vanity Exhibitionism Exploitativeness Entitlement Self-esteem 43 36,092.33* (.35,.31).32* (.34,.30).30* (.32,.28).22* (.24,.20).09* (.11,.07).08* (.10,.06).01 (.01,.03) Grandiose narcissism 3 1,068.16* (.20,.12).25* (.29,.21).06* (.10,.02).22* (.26,.18).24* (.28,.20).20* (.24,.16).33* (.37,.29) Psychological health 17 9,704.20* (.23,.17).18* (.21,.15).15* (.18,.12).13* (.16,.10).03* (.06,.00).01 (.04,.02).09* (.06,.12) Aggression/Anger 8 20,332.07* (.04,.10).07* (.10,.04).01 (.04,.02).04* (.07,.01).19* (.22,.16).17* (.20,.14).30* (.33,.27) Table 2, Continued. 4-Factor (Emmons, 1987) 2-Factor (Corry et al., 2008) 3-Factor (Ackerman et al., 2011) 16-Item (Ames et al., 2006) S/S L/A S/A E/E L/A Ex/En L/A GE E/E Self-esteem.38* (.40,.36).30* (.32,.28).18* (.20,.16).06* (.04,.08).30* (.32,.26).13* (.15,.11).34* (.36,.32).22* (.24,.20).11* (.09,.13).25* (.27,.23) Grandiose narcissism Psychological health.28* (.32,.24).22* (.26,.18).30* (.34,.26).24* (.28,.20).28* (.32,.24).25* (.29,.21).25* (.29,.21).23* (.27,.19).28* (.32,.24).38* (.42,.34).21* (.24,.18).17* (.20,.14).08* (.11,.05).14* (.11,.17).15* (.18,.12).03 (.06,.00).17* (.20,.14).12* (.15,.09).14* (.11,.17).10* (.13,.07) Aggression/Anger.00 (.04,.04).09* (.13,.05).14* (.18,.10).37* (.41,.33).10* (.14,.06).23* (.27,.19).09* (.13,.05).11* (.15,.07).35* (.39,.31).21* (.24,.18) Note. *p <.05. 95% confidence intervals are reported in parentheses. For Emmons (1987): S/S = Self-absorption/Self-admiration; L/A = Leadership/Authority; S/A = Superiority/Arrogance; E/E = Exploitativeness/Entitlement. For Corry et al. (2008): L/A = Leadership/Authority; Ex/En = Exhibitionism/Entitlement. For Ackerman et al. (2011), L/A = Leadership/Authority; GE = Grandiose Exhibitionism; E/E = Entitlement/Exploitativeness. 5
Table 3. Correlations between composite NPI, NPI subscales, and Five Factor Model personality traits, Appendix A. Benchmark NPI Subscale Composite NPI Self-esteem Grandiose narcissism IRT-X IRT-N Expert-X Expert-N EFA-X EFA-N Agreeableness.16* (.13,.19).28* (.31,.25).37*.04* (.01,.07).31* (.28,.34).12* (.09,.15).21* (.18,.24).09* (.06,.12).23* (.19,.26) Extraversion.49* (.51,.46).38* (.41,.35).09*.59* (.62,.56).32* (.35,.29).62* (.65,.59).37* (.40,.34).64* (.67,.61).35* (.38,.32) Neuroticism.17* (.14,.20).57* (.54,.60).11*.16* (.13,.19).09* (.06,.12).21* (.18,.24).15* (.12,.18).20* (.17,.23).15* (.12,.18) Openness to experience.17* (.20,.14).20* (.23,.17).07*.25* (.28,.22).10* (.13,.07).18* (.21,.15).17* (.20,.14).21* (.24,.18).14* (.17,.11) Conscientiousness.08* (.11,.05).40* (.43,.37).10*.20* (.23,.17).03* (.00,.06).10* (.13,.07).06* (.09,.03).17* (.20,.14).02 (.05,.01) Note. *p <.05. k = 11, N = 29,587. Grandiose narcissism correlations with the FFM factors are from the meta-analysis conducted by Samuel and Widiger (2008a). Bolded NPI subscale correlations are of a significantly greater magnitude than are the correlations for their respective paired subscales, p <.05. 6
References, Appendix A Ackerman, R. A., Witt, E. A., Donnellan, M. B., Trzesniewski, K. H., Robins, R.W., & Kashy, D. A. (2011). What does the narcissistic personality inventory really measure?. Assessment, 18, 67 87. Ames, D. R., Rose, P., & Anderson, C. P. (2006). The NPI-16 as a short measure of narcissism. Journal of Research of Personality, 40, 440 450. Benet-Martınez, V., & John, O. P. (1998). Los Cinco Grandes' across cultures and ethnic groups: Multitrait multimethod analyses of the Big Five in Spanish and English. Journal of Personality and Social Psychology, 75, 729 750. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd edition). Hillsdale, NJ: Erlbaum. Corry, N., Merritt, R. D., Mrug, S., & Pamp, B. (2008). The factor structure of the Narcissistic Personality Inventory. Journal of Personality Assessment, 90, 593 600. Costa, P. T., Jr., & McCrae, R. R. (1985). The NEO Personality Inventory manual. Odessa, FL: Psychological Assessment Resources. Costa, P. T., Jr., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO-PI R) and NEO Five-Factor Inventory (NEO-FFI) professional manual. Psychological Assessment Resources. DeYoung, C. G., Quilty, L. C., & Peterson, J. B. (2007). Between facets and domains: 10 aspects of the Big Five. Journal of Personality and Social Psychology, 93, 880 896. Emmons, R. A. (1987). Narcissism: Theory and measurement. Journal of Personality and Social Psychology, 52, 11 17. John, O. P., & Srivastava, S. (1999). The Big Five trait taxonomy: History, measurement, and theoretical perspectives. In L. A. Pervin, & O. P. John (Eds.), Handbook of personality: Theory and research (pp. 102 138). New York: Guilford Press. Lynam, D. R. & Widiger, T. A. (2001). Using the five-factor model to represent the DSM IV personality disorders: An expert consensus approach. Journal of Abnormal Psychology, 110, 401 412. Miller, J. D., & Campbell, W. K. (2008). Comparing clinical and social personality conceptualizations of narcissism. Journal of Personality, 76, 449 476. 7
Miller, J. D., Gaughan, E. T., Maples, J., & Price, J. (in press). A comparison of Agreeableness scores from the Big Five Inventory and the NEO PI R: Consequences for the study of narcissism and psychopathy. Assessment. Miller, J. D., Maples, J., & Campbell, W. K. (this issue). Comparing the construct validity of scales derived from the Narcissistic Personality Inventory: A reply to Rosenthal and Hooley (2010). Journal of Research in Personality, 45. Miller, J. D., Reynolds, S. K., & Pilkonis, P. A. (2004). The validity of the five-factor model prototypes for personality disorders in two clinical samples. Psychological Assessment, 16, 310 322. Paulhus, D. L. (2001). Normal narcissism: Two minimalist accounts. Psychological Inquiry, 12, 228 230. Raskin, R., & Terry, H. (1988). A principal-components analysis of the Narcissistic Personality Inventory and further evidence of its construct validity. Journal of Personality and Social Psychology, 54, 890 902. Rosenthal, S. A., & Hooley, J. M. (2010). Narcissism assessment in social personality research: Does the association between narcissism and psychological health result from a confound with self-esteem? Journal of Research in Personality, 44, 453 465. Samuel, D. B., & Widiger, T. A. (2006). Clinicians judgments of clinical utility: A comparison of the DSM IV and five-factor models. Journal of Abnormal Psychology, 115, 298 308. Samuel, D. B., & Widiger, T. A. (2008a). A meta-analytic review of the relationships between the five-factor model and DSM IV-TR personality disorders: A facet-level analysis. Clinical Psychology Review, 28, 1326 1342. Samuel, D. B., & Widiger, T. A. (2008b). Convergence of narcissism measures from the perspective of general personality functioning. Assessment, 15, 364 374. 8