Unmasking Narcissus: A competitive test of existing hypotheses on (agentic, antagonistic, neurotic, and communal) narcissism and (explicit and implicit) self-esteem across 18 samples

ABSTRACT

Research on narcissists’ explicit self-esteem (ESE) and implicit self-esteem (ISE) is characterized by a wide array of competing hypotheses and inconsistent empirical findings. Using data from 18 samples (total N = 5,547), we moved beyond classical null-hypothesis testing and employed an information-theoretic approach combined with Response Surface Analysis to provide a first competitive test of all plausible ESE-ISE combinations for different aspects of narcissism and across a wide set of ISE measures. Agentic and communal narcissism were positively, neurotic and antagonistic narcissism were negatively related to ESE. Contradicting the mask model and other more complex models, no consistent evidence was found for relations between narcissism and ISE, and ESE-ISE discrepancies, respectively. Implications for understanding narcissistic self-regulatory dynamics are discussed.

KEYWORDS:

• Narcissism
• explicit self-esteem
• implicit self-esteem
• response surface analysis
• information-theoretic approach

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here.

Notes

1. We do not follow the more classic distinction between grandiose and vulnerable narcissism because it entangles either neurotic and antagonistic aspects (vulnerable narcissism) or antagonistic and agentic aspects (grandiose narcissism).

2. We point out that the datasets used in our study have already been analyzed elsewhere, partly with similar research questions in mind. Although the exact analyses we conducted have not been carried out by any of the co-authors before, we cannot rule out that the co-authors had prior knowledge of the data concerning associations of narcissism to ESE and ISE measures. However, this does not apply to the first author, who previously had no access to the data.

3. Reliabilities of Ackerman’s subscales are consistent with those reported in the literature (Ackerman et al., 2011): higher Cronbach’s alpha for the two agentic facets LA and GE and low Cronbach’s alpha for the antagonistic facet EE, which consists only of four items.