Validation of the German Five-Factor Narcissism Inventory and Construction of a Brief Form Using Ant Colony Optimization

Abstract

Narcissism is a multifaceted construct commonly conceptualized as comprising grandiose and vulnerable aspects in a two-factor model. While the manifold correlates of these aspects imposed a challenge for research on the structure of narcissism, recent models converge in a three-factor structure of agentic-extraverted, antagonistic, and neurotic aspects, capturing variance in different conceptualizations and correlates of narcissism. We construct and validate a German adaptation of the Five-Factor Narcissism Inventory (FFNI), a measure assessing these aspects based on the Five-Factor Model. In eight samples ( N = 2,921), we found the German FFNI to align with both, two- and three-factor models. The factors display good criterion validity with other narcissism measures, (non-)clinical personality dimensions, interpersonal styles, and (mal-)adaptive adjustment. Neurotic and antagonistic narcissism discriminated between individuals with/without mental disorder diagnoses, and displayed a characteristic profile in incarcerated offenders. Since the FFNI is comprehensive but long, we constructed a 30-item brief form (FFNI-BF) optimizing the internal structure and external validity using ant colony optimization. The FFNI-BF displayed good psychometric characteristics and similar, in certain aspects even advantageous criterion validity. We conclude that the German FFNI validly measures key aspects of narcissism, and the FFNI-BF captures these in a concise manner.

The nonlinear association between grandiose and vulnerable narcissism: An individual data meta‐analysis

“…To have comparable scores across studies, we used regression‐based estimation of FFNI‐LF scores for studies using the FFNI‐SF (see Supporting Information), which is why we cannot report internal consistencies here. Studies used either the original English version, a German version (manuscript in prep., for material, see Jauk et al., 2021), or a Polish version (Rogoza et al., 2021).…”

Too Beautiful to be Fake: Attractive Faces are Less Likely to be Judged as Artificially Generated

“…In the first part of the study, participants answered a series of personality questionnaires presented in the order below. These include the Mini-IPIP6 (24 items, Sibley et al, 2011) measuring 6 personality traits, the SIAS-6 and the SPS-6 (6 items each, Peters et al, 2012) assessing social anxiety levels, 5 items we devised pertaining to expectations about AI-generated image technology ("I think current Artificial Intelligence algorithms can generate very realistic images"), of which we mixed with 5 items from the general attitudes towards AI scale to lower the former's saliency and the ATTRACTIVENESS AND REALITY 9 possibility of it priming the subjects about the task, (GAAIS, Schepman & Rodway, 2020) the FFNI-BF (30 items, Jauk et al, 2022) measuring 9 facets of narcissism; the R-GPTS (18 items, Freeman et al, 2021) measuring 2 dimensions related to paranoid thinking; and the IUS-12 (12 items, Carleton et al, 2007) measuring intolerance to uncertainty.…”

A matter of measure? Assessing the three dimensions of narcissism.

“…Since exploratory factor analyses of the FFNI subscales (Miller et al, 2016) and items (Rogoza et al, 2021) yielded a three-factor structure which appeared consistent with the TriMN, a growing body of research has been applying the FFNI(-SF) to capture agentic extraversion, narcissistic antagonism, and neuroticism. The factor analytic results were sometimes ambiguous (Rogoza et al, 2021), however, and attempts at modeling a simple structure for the FFNI using confirmatory factor analytic approaches provided unsatisfactory results (see Jauk et al, 2023).…”

“…The aim of our first study was to assess and compare the respective factor structure of the FFNI-BF and a combination of NARQ and HSNS items as two potentially alternative operationalizations of the TriMN. With respect to the FFNI-BF, prior constrained measurement model analyses (i.e., confirmatory factor analyses [CFA]) indicated relatively poor fit for the simple structure model with three correlated factors (comparative fit index [CFI] = .65; root-mean-square error of approximation [RMSEA] = .09; Jauk et al, 2023). Using exploratory structural equation modeling, Jauk et al (2023) identified a properly fitting measurement model for the FFNI-BF (CFI = .97; RMSEA = .04), which, however, did not follow a simple structure as it included numerous cross-loadings and yielded distorted parameter estimates due to probable overfitting.…”

“…With respect to the FFNI-BF, prior constrained measurement model analyses (i.e., confirmatory factor analyses [CFA]) indicated relatively poor fit for the simple structure model with three correlated factors (comparative fit index [CFI] = .65; root-mean-square error of approximation [RMSEA] = .09; Jauk et al, 2023). Using exploratory structural equation modeling, Jauk et al (2023) identified a properly fitting measurement model for the FFNI-BF (CFI = .97; RMSEA = .04), which, however, did not follow a simple structure as it included numerous cross-loadings and yielded distorted parameter estimates due to probable overfitting. A simple structure, however, is usually desired in psychometric testing, as it warrants clear and unambiguous factor interpretations when each item is assigned to only one, not multiple factors (Furr, 2021).…”