Narcissism and Reactions to a Self-Esteem Insult: An Experiment Using Predictions from Self-Report and the Rorschach Task

Abstract

We used self-reported narcissistic grandiosity and vulnerability and a component derived from 11 potential grandiosity and narcissism variables (GNVs) coded from Rorschach behavior to predict fluctuations in self-esteem and their links to anger and defensive reactions. We assessed state mood, state self-esteem, and performance attributions in 105 college students who underwent a self-esteem manipulation involving success followed by failure on cognitive testing. Self-reported grandiosity predicted the disavowal of effortful ability as a factor in failure, but we did not replicate other previously reported findings for this variable. Self-reported vulnerability predicted oscillations in self-reported mood and self-esteem. The GNV scale predicted spontaneously expressed hostility and externalization following self-esteem insult, and attributions mediated its relationship with anger expressed after failure. We discuss implications of these results and recommend additional replication research.

Acknowledgments

We are grateful to the participants who took part in the study and to Andrea Kiss, Beth Runski, and Sean Walsh who helped with the data collection. We also thank Wilson Hsiao and Raymond Voss for their help with in the initial stages of the MediaLab programming.

Data availability statement

We chose not to place raw data in a suitable repository because of the nature of the assessment material used in the study. Specifically, responses to the Rorschach and to open-ended questions can be potentially revealing of participants’ personal information. However, researchers wishing access to the raw data or the syntax used for the analyses may contact the first or second author.

Declaration of interest statement

Gregory J. Meyer is a member of a company that sells the manual for the Rorschach Performance Assessment System (R-PAS) and associated products and two of the 11 Rorschach variables used in this study are included in R-PAS.

Notes

1 Rhodewalt and Morf (1998) used an extreme groups design and thus their results using self-reported grandiose narcissism do not provide a good basis for extrapolation. Similarly, the average effect size (r = .41) from previous research with the GNVs in relation to clinician ratings of narcissism in their therapy clients does not provide a good basis for detecting the subtler effects expected from this experimental manipulation with nonpatients.

2 We did not anticipate that some respondents would experience the “success” condition as a “failure” and in turn provide external justifications for their less than perfect performance. However, they did (e.g., “I started to get bored there was too much time in between questions”) so we coded it in line with our overarching hypotheses and while blind to any other results.

3 In general, participants who took credit for their success and thus were coded as aggrandizing did not simultaneously externalize responsibility for their less than perfect performance in the Success condition. This produced a substantial negative correlation (-.52) between these two variables.

4 The higher average measures reliability coefficients would have been appropriate for the free-text ratings because all responses were jointly coded. However, we report the single measure coefficients for consistency.