Exposure to Celebrities as a Possible Explanatory Mechanism in the Perception of American Narcissism

An a priori power analysis was conducted to ensure an adequate sample size using R software’s “pwr” package (Champely, 2015). We estimated a medium effect size of Cohen’s d = 0.5, a conservative estimate based on the finding from Miller et al. (2015) that high visibility exemplars (e.g., actors/actresses) were rated as significantly more narcissistic (d = 1.03) than low visibility exemplars (e.g., teachers). Analyses indicate that for an independent sample t-test with power of 0.9 to find a medium effect at a significance level of .05, 86 participants were needed in each condition, suggesting that this study (N = 300) was sufficiently powered for a medium effect.

In the interest of transparency, we preregistered our hypotheses, experimental procedures (including measures), and our analytic strategy online on the Open Science Framework under the project title “Mechanisms Belying Perceptions of American Narcissism” (https://osf.io/t93cw/). Of note, we pre-registered that we intended to perform sequential analysis as recommended by Lakens (2014) on our data after our first session of running participants (anticipated N = 200) to test for a significant effect with an appropriately adjusted alpha value. We underestimated the flow of recruitment, and we gathered 300 participants before ceasing data collection. After performing our first analyses (N = 300), we decided to terminate data collection based on a sufficient degree of confidence in our results, as suggested by Lakens.

Participants were recruited from a large, Southeastern U.S. university research participant pool and received course credit in exchange for their participation. To disguise the purpose of the study, participants were told the investigation concerned the relationship between personality and memory interference. Upon entering the lab, participants were administered informed consent and randomly assigned to either the neutral or celebrity magazine condition and seated on pre-designated sides of the room. After completing a demographics form, closed binders containing two magazines were passed out to each participant. Participants in the celebrity magazine condition were given recent issues of Lifestyle and People Magazine, whereas participants in the neutral magazine were given recent issues of Consumer Reports and Scientific American. All participants were given 15 minutes to review the magazines in preparation for a quiz about their content. After the exposure, the magazines were collected, and participants completed the “memory interference task,” which consisted of the PNC questionnaires. Lastly, participants were administered a manipulation check disguised as a memory test to assess whether participants had paid attention to the magazines. IRB approval was obtained for all aspects of this study.

Five Factor Model Rating Form (FFMRF; Mullins-Sweat, Jamerson, Samuel, Olson, & Widiger, 2006). The FFMRF is a 30-item rating for assessing the five domains and 30 facets of the FFM. Each facet is assessed with a single item (1 = extremely low, 5 = extremely high). Domain scores are simply the average of the six facets that comprise each domain. Internal consistencies for the PNC ratings for Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness were α = .57, .62, .57, .72, and .74, respectively.

Narcissistic Personality Inventory–13 (NPI-13; Gentile et al., 2013). U.S. PNC ratings of narcissism were based on a 13-item, forced choice version of the NPI, which provides both a total score (α = .87) and three subscale scores: Leadership/Authority (4 items; α = .75), Grandiose Exhibitionism (5 items; α = .80), and Entitlement/Exploitativeness (4 items; α = .75).

Structured Clinical Interview for DSM–IV Personality Disorders–Personality Questionnaire (SCID-II P/Q; First, Gibbon, Spitzer, Williams, & Benjamin, 1997). The SCID-II P/Q is a self-report measure of personality psychopathology found in the DSM-IV and DSM-5. Responses are in a dichotomous, “yes” or “no” format. In the present investigation, only items NPD (SCID-II P/Q: 17 items) and AVPD (SCID-II P/Q: 7 items) scales were used. Alphas for PNC ratings of NPD and AVPD were .85 and .57, respectively.

Manipulation Check. A 10-item self-report manipulation check (disguised as a memory task) was used to simultaneously maintain the cover story of the study and to assess attention to the magazines. The items were a combination of multiple choice and True/False items, which were balanced across condition.

Results of the manipulation check indicate that on average, participants scored substantially above chance in both the celebrity magazine (M = 6.8/10, SD = 1.33) and the neutral magazine condition (M = 7.1/10, SD = 1.52; Cohen’s d of comparison between conditions = –.21 [95% CI: –.44 to .02], p = .08), which was consistent with pilot testing. This suggests that participants were comparably attentive to the stimuli between conditions. Ten participants were excluded because their scores were at or below chance levels, which suggests that they may not have paid sufficient attention to the magazines. This resulted in N = 146 in the celebrity magazine condition and N = 144 in the neutral magazine condition. Of note, the pattern of findings was virtually identical before and after exclusion.

A series of independent sample t-tests were run to compare the two conditions with regard to mean scores on the PNC ratings of the NPI-13 total score and 3 factor scores (Grandiose Exhibitionism, Leadership/Authority, Exploitativeness/Entitlement), SCID-II NPD and avoidant personality disorder, as well as the five domains and 30 facets of the FFM. Descriptive statistics for all scales as well as t-scores, p-values, and Cohen’s d for the mean comparisons are listed in Tables 1 and 2.

Contrary to hypotheses, the results indicate no differences in terms of mean ratings for NPI-13 total score or any of the 3 factor scores, NPD, or AVPD. In terms of FFM domains, only Openness evinced a significant difference, with participants in the celebrity magazine condition rating Americans in general as higher in trait Openness. Participants in the celebrity magazine also rated Americans in general as higher in Gregariousness, Fantasy, and Values, and lower in Deliberation, although the effect sizes were small in nature.

Interests in differences among cultures are long-standing, but recent studies have afforded new insights into how individuals are perceived by others both inside and outside of their home country. Though some studies have cast doubt on the accuracy of PNC rating as they relate to organic differences, differences in perception of foreign cultures have been empirically demonstrated, and they have been found to influence behavior in meaningful ways (Shteynberg, Gefand, & Kim, 2009). Thus, it is of both theoretical and practical interest to explore the underpinnings of these phenomena. Research has consistently identified that both Americans and non-Americans alike view the typical American to be very high in trait narcissism and Extraversion, and low in Agreeableness. In fact, both American and non-American participants in several previous samples rate the average American as meeting diagnostic criteria for an NPD diagnosis under DSM-5 (Miller et al., 2015), despite the fact that this tends to be a relatively infrequent diagnosis. Despite the pervasiveness of this distorted perception of Americans’ national character, the particular mechanisms by which this view is developed and maintained are unknown.

The current study used an experimental design to investigate a potential mechanism by exposing participants to either a set of celebrity magazines or neutral magazines and subsequently gathering PNC ratings of American narcissism and FFM personality. It was hypothesized that participants in the celebrity-magazine condition would give higher PNC ratings of American narcissism, lower ratings of Agreeableness, and higher ratings of Extraversion than participants in the neutral magazine condition. In line with the availability heuristic, we expected that proximal exposure to American actors/actresses, athletes, and politicians (groups who are considered particularly narcissistic and featured prominently in the experimental stimuli) would result in amplified PNC ratings of American narcissism. This pattern of results would support the view that PNC ratings of Americans are, in part, driven by the visibility of narcissistic exemplars. However, contrary to hypotheses, participants in the celebrity magazine condition did not provide higher PNC ratings of American narcissism than participants in the neutral magazine condition. No differences were found in Agreeableness and Extraversion by condition either, which are believed to be the core personality domains that underlie narcissism/NPD (Miller, Lynam, Hyatt, & Campbell, 2017; Paulhus, 2001).

Although these results do not support our hypotheses, several comments are warranted. First, participants in both conditions provided high PNC ratings of American narcissism/NPD and an FFM profile that were consistent with prior literature, thus replicating previous work as to the evaluation of the “typical” American. Indeed, participants in both conditions rated the typical American as pathologically narcissistic on the NPD measure, exceeding the diagnostic cut-off score by approximately 100%. This is particularly noteworthy considering the low base rate of this disorder (point prevalence ~1%; Torgersen, 2013). In fact, the lack of differences between the groups may be due, in part, to somewhat of a ceiling effect, in that both groups provided PNC ratings of Americans’ narcissism that were relatively extreme (i.e., endorsing half or more of the narcissism items as being representative of Americans). It may be difficult to experimentally “budge” these ratings given that participants come in to these sessions with strong, a priori evaluations of Americans’ narcissism. However, it should be noted that while participants endorsed that Americans were very narcissistic, this ceiling effect is not a product of the measures themselves. In other words, participants did not rate Americans as uniformly narcissistic, as they did not endorse every item on measures of narcissism on behalf of Americans in general. Thus, the extent to which a ceiling effect exists remains unclear.

Additionally, while no hypothesized group differences were found in terms of FFM traits, the pattern of results suggests that Americans were perceived as relatively low on Agreeableness and high on Extraversion, which is consistent with prior literature and the FFM profile of NPD (Lynam & Widiger, 2001). Finally, in line with previous findings, participants across conditions did not rate the PNC of Americans as broadly disordered, as evidenced by comparable, sub-threshold PNC ratings of AVPD. This lends support to the notion that the narcissistic profile may be uniquely descriptive of PNC ratings of Americans.

Several limitations are of note. First, research on PNC ratings can be difficult to interpret, as PNC ratings bear little relation to self and informant reported personality data (Terraciano et al., 2005). Regardless, cultural stereotypes have been demonstrated to influence behavior toward members of another culture above and beyond individualizing information, suggesting that while PNC ratings may not be existentially precise, they are powerful in shaping inter-cultural interactions (Chaiken & Trope, 1999). Second, our analyses can only speak to the influence of proximal exposure to American celebrities on American citizens. Our study remains mute to the influence this type of exposure may have on citizens of other countries, although American culture, including the role of celebrities, is believed to have a relatively large global influence. Additionally, while the preponderance of celebrities featured in these magazines were of American nationality, there were some individuals (e.g. David Beckham) who were not, thus we are unable to make specific claims about the influence of American-born celebrities in particular. Similarly, while our sample size was adequately powered, the collegiate composition of our sample is restricted in terms of terms of age, education, and racial diversity.

Although the current study provides insight into how brief exposures influence (or fail to influence) PNC of Americans, we are hopeful that future research will continue to investigate the nuances of how narcissistic exemplars may reify PNC of Americans. For example, the duration of magazine exposure in this study was relatively brief compared to, say, a typical episode of reality television or a political debate. Thus, while a 15 minute exposure to a celebrity magazine may fail to inflate the PNC of American narcissism (that, notably, is quite high without an exposure to a narcissistic exemplar), it is possible that more consistent exposure (e.g. frequent readership of celebrity magazines, actual interactions with celebrities) may play a more potent role in strengthening these attitudes.

On the other hand, it may be such that cultural indicators of American narcissism are so pervasive and deep-rooted that PNC ratings are impervious to new, confirming (or disconfirming) information. To address the opposite side of this question, it would be interesting to test if exposures to acts of American selflessness (e.g. Americans performing charity work) could lower PNC ratings of American narcissism. Furthermore, it would be interesting to see if brief exposures could manifest in longitudinal shifts in PNC attitudes, and perhaps more importantly, if they could lead to behavior change in cross-cultural interactions. To fully investigate this possibility, it is imperative that these and similar hypotheses are tested in individuals of all ages from a variety of cultural backgrounds. Future work will prove critical in elucidating the development and maintenance of PNCs, and how these perceptions are related to important interpersonal outcomes.

The authors have no competing interests to declare.

The author(s) of this paper chose the Open Review option, and the peer review comments are available at:https://doi.org/10.1525/collabra.52.pr