Self-Serving Bias or Simply Serving the Self? Evidence for a Dimensional Approach to Narcissism

Abstract

Previous research has suggested that narcissism can be conceptualized as a multidimensional construct consisting of the related, but unique, dimensions of grandiosity and entitlement. The current studies examined the divergent associations of grandiosity and entitlement with respect to different types of self-serving strategies. In Study 1, we found that narcissistic grandiosity, but not entitlement, was positively associated with a self-enhancing strategy of unrealistic optimism. This association was not mediated by self-esteem. In Study 2, narcissistic entitlement, but not grandiosity, was predictive of unethical decision-making, an interpersonal self-promotional strategy that advances the self at the expense of others. Together, both studies support a model of narcissism consisting of a relatively intrapersonal dimension of grandiosity and a relatively interpersonal dimension of entitlement.

1. Introduction

Narcissism has been described as a paradoxical construct (Morf & Rhodewalt, 2001). On one hand, narcissists tend to think very highly of themselves and have an inflated sense of self-worth. On the other hand, they often behave in ways that suggest an underlying vulnerability of the self (Jordan, Spencer, Zanna, Hoshino-Browne, & Correll, 2003; Zeigler-Hill, 2006). In part because of this paradox, many researchers have either distinguished between multiple forms of narcissism (e.g., Watson & Biderman, 1993; Wink, 1991) or identified various subfactors of narcissism (e.g., Emmons, 1984; Raskin & Terry, 1988). Others have also characterized narcissism as a combination of high agentic and low communal traits (Campbell & Foster, 2007) or high extraversion and low agreeableness (Paulhus, 2001). Despite these seemingly different ways of categorizing narcissism, many conceptualizations make a distinction between components of narcissism generally characterized by a grandiose and inflated sense of self-worth and components characterized by high levels of entitlement, exploitation, and similar traits (Emmons, 1984, 1987; Raskin & Novacek, 1989; Watson & Biderman, 1993).

Consistent with these interpretations of narcissism, the entitlement/exploitativeness subscale (E/E) of the Narcissistic Personality Inventory (NPI; Raskin & Hall, 1979) diverges from the remaining subscales of the NPI with respect to a host of variables associated with narcissism (see Brown & Tamborski, 2012, for a discussion). Given this trend, Brown, Budzek, and Tamborski (2009) suggested a conceptualization of narcissism as comprising two overarching dimensions: a predominantly intrapersonal dimension of grandiosity and a predominantly interpersonal dimension of entitlement.

Brown et al. (2009) argued that the grandiosity dimension orients the narcissist toward maintaining an internal sense of self-importance, whereas the entitlement dimension orients the narcissist toward maintaining the status of the self vis-à-vis others. Brown and colleagues demonstrated that grandiosity, but not entitlement, predicted mental health (an intrapersonal outcome). Furthermore, they found that entitlement, but not grandiosity, predicted overt cheating, when behavior unambiguously violated social norms to satisfy self-interest. When cheating was more subtle and rationalizable, grandiosity, but not entitlement, predicted cheating.

Previous research has clearly demonstrated a link between narcissism (as measured by the NPI) and a variety of intrapersonal strategies. For example, narcissists tend to report lower actual/ideal self discrepancies (Emmons, 1984; Raskin, Novacek, & Hogan, 1991; Rhodewalt & Morf, 1995), display greater confidence in their responses to general knowledge questions (but not greater accuracy; Campbell, Goodie, & Foster, 2004), overestimate the potential benefits from risky ventures (Foster, Shenesey, & Goff, 2009), and discount the importance of a task after receiving negative feedback (e.g., Campbell, Reeder, Sedikides, & Elliot, 2000; Farwell & Wohlwend-Lloyd, 1998). Narcissists are also more willing than non-narcissists are to violate pro-social norms in order to protect their ego or to promote their own self-interests (i.e., interpersonal strategies). In particular, narcissism is associated with responding aggressively to insults, even if the victim is not responsible for the insult (Martinez, Zeichner, Reidy, & Miller, 2008; Twenge & Campbell, 2003). Narcissism is also positively associated with vengeance seeking (Brown, 2004), and punitiveness (Bushman, Bonacci, van Dijk, & Baumeister, 2003).

Fewer studies have reported correlations between self-enhancement tendencies and the NPI subscales, but those that do support a grandiosity/entitlement distinction. For example, Rhodewalt and Morf (1995, Study 3) found that the non-E/E subscales of the NPI were positively correlated with self-evaluative ratings in various domains (e.g., intelligence) and the certainty of those ratings. However, E/E was uncorrelated with both. In the same study, the composite NPI was negatively correlated with actual/ideal self-discrepancies, yet E/E was positively correlated with this discrepancy. Additional studies have reported a similar divergence between E/E and the rest of the NPI in predicting self-esteem (Bogart, Benotsch, & Pavlovic, 2004; Rose, 2002) and positive illusions (Hickman, Watson, & Morris, 1996).

In contrast with intrapersonal self-enhancement, studies that have examined the associations between the NPI subscales and interpersonal strategies suggest that they are primarily driven by E/E. Using a version of the NPI with seven subscales (Raskin & Terry, 1988), Reidy, Zeichner, Foster, and Martinez (2008) found that only entitlement and exploitativeness uniquely predicted the intensity and duration of an electric shock that participants administered to a competitor. Similarly, Antes et al. (2007) demonstrated that E/E was the only NPI subscale that consistently predicted unethical decisions. Finally, Exline, Baumeister, Bushman, Campbell, and Finkel (2004) showed a negative relationship between entitlement and forgiveness, but no relationship between the remaining NPI subscales and forgiveness. Thus, the composite E/E or the singular entitlement subscale of the NPI appears to be associated with promoting the self at the expense of others. Likewise, the remaining subscales are specifically associated with intrapersonal self-enhancing biases, such as overly positive self-evaluations and smaller actual/ideal self-discrepancies.

In the following studies, we examine the distinction between grandiosity and entitlement by exploring their roles in predicting a relatively intrapersonal self-enhancing strategy (i.e., unrealistic optimism) and a more interpersonal strategy that focuses on self-promotion without regard for the well-being of others (i.e., unethical decision-making). Previous research has already suggested such a dissociation using the NPI subscales (Antes et al., 2007; Hickman et al., 1996). However, these subscales suffer from unacceptably low internal reliabilities, questionable item content, and ambiguity regarding the most appropriate number of subscales (Tamborski & Brown, 2012). Furthermore, none of the aforementioned studies investigated whether E/E might interact with the remaining subscales. Therefore, we assessed grandiosity and entitlement with reliable and highly face-valid scales specifically designed to measure these two dimensions of narcissism.

2. Study 1

Study 1 examined whether narcissistic grandiosity and entitlement exhibit divergent associations with unrealistic optimism — a form of intrapersonal self-enhancement (Regan, Snyder, & Kassin, 1995). Because previous research has found that the NPI (particularly the non-E/E subscales) is positively associated with optimism (Farwell & Wohlwend-Lloyd, 1998; Hickman et al., 1996), we hypothesized that grandiosity would likewise be positively associated with unrealistic optimism. Additionally, Brown et al. (2009) reported a positive association between grandiosity and mental health (which included a simple index of optimism) that was only partially accounted for by self-esteem. For this reason, we expected that grandiosity would similarly demonstrate a positive relationship with unrealistic optimism that was not dependent on self-esteem. We also hypothesized that this relationship would be largely independent of entitlement, which we did not expect to be associated with optimism. Furthermore, the measure of optimism that we used allowed us to explore whether grandiosity and entitlement were related to optimism in general, or whether any associations might occur more strongly for positive versus negative events.

3. Method

3.1. Participants

One hundred ten undergraduates from the University of Oklahoma participated in partial fulfillment of a research exposure requirement in an introductory psychology course. Three participants did not complete half the items on our optimism scale and were excluded from the analyses. The majority of the remaining 107 participants were female (65%) and Caucasian (79%).

3.2. Measures and Procedure

At least three weeks prior to the study, participants completed measures of grandiosity and entitlement, as well as the Rosenberg Self-Esteem Scale (α = .87; Rosenberg, 1965). To assess grandiosity, we used the Narcissistic Grandiosity Scale (NaGS; Rosenthal, Hooley, & Steshenko, in preparation). The NaGS (α = .94) consists of 16 adjectives (e.g., glorious, omnipotent) that participants rated for their self-descriptiveness on a Likert scale ranging from 1 (not at all) to 7 (extremely). In addition to being highly face valid, the NaGS is moderately correlated with non-entitlement subscales of the NPI (rs ranging from .35 to .49; Brown et al., 2009).

To assess entitlement, we used the Psychological Entitlement Scale (PES; Campbell, Bonacci, Shelton, Exline, & Bushman, 2004). The PES (α = .86) contains nine statements (e.g., “If I were on the Titanic, I would deserve to be in the first lifeboat!”) to which participants rated their agreement on a Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Campbell and colleagues (2004) found the PES to be strongly correlated with NPI-assessed entitlement (r = .54) and predictive of selfish behaviors, such as taking candy meant for children.

During a separate laboratory session, participants completed a measure of unrealistic optimism. Thirty-six items describing potential future events were selected from previous research (Weinstein, 1980; Chambers, Windschitl, & Suls, 2003). The events were balanced in terms of valence and likelihood, including items such as “Going blind,” and “Winning a sweepstakes.” Participants responded to these items by indicating the probability that each event would happen to them compared to other students at their university of the same age and gender, using a response scale anchored by “well below average” (1) and “well above average” (5). Negative events were reverse scored, with higher scores indicating greater optimism. Reliability coefficients for the composite optimism measure, the positive events subscale, and the negative events subscale were .71, .77, and .82, respectively.

After completing the unrealistic optimism scale, participants engaged in several other tasks not relevant to the current study. After completing all tasks, participants were fully debriefed and dismissed.

4. Results and Discussion

Grandiosity was positively correlated with entitlement and self-esteem, but self-esteem was not associated with entitlement (Table 1). No significant gender differences emerged across either dimension of narcissism or unrealistic optimism, including both positive and negative subscales. As a result, gender was not considered further in our analyses. When we entered grandiosity, entitlement, and their interaction (after first mean-centering both predictors, Aiken & West, 1991), grandiosity significantly predicted optimism, β = .25, t(103) = 2.43, p = .02, sr²= .05, but entitlement did not, β = −.10, t(103) = −0.96, ns. The interaction between grandiosity and entitlement was not significant, β = .06, t(103) = 0.72, ns, so we dropped it for the following analyses. Adding self-esteem to the model only reduced the association between grandiosity and optimism slightly, β = .22, t(103) = 2.18, p = .03, sr²= .04, and self-esteem itself was only marginally significant, β = .16, t(103) = 1.68, p = .10, sr²= .02. A test of the indirect effect using bootstrapping (Preacher & Hayes, 2008) confirmed that self-esteem did not mediate the association between grandiosity and optimism, as the 95% bias corrected CI of the indirect effect included zero [0, .11]. Consistent with Brown et al. (2009; Study 2), self-esteem was not sufficient to explain the association between grandiosity and optimism. Further exploratory analyses revealed that the association between grandiosity and optimism was entirely driven by optimism for positive events, β = .39, t(103) = 3.91, p < .001, sr²= .13, an association that was also unmediated by self-esteem, β = .01, t(103) = 0.14, ns. Again, the 95% bias corrected CI for the indirect effect included zero [−.04, .06]. Grandiosity was not related to optimism for negative events, β = −.04, t(103) = 0.42, ns, and entitlement was not related to either form of optimism, βs < |.10|, t(103) < |1|, ns.

Table 1.

Zero-order Correlations among Self-esteem, Grandiosity, Entitlement, and Unrealistic Optimism (Including Positive and Negative Events).

	1	2	3	4	5	6	M	SD
1. Self-esteem	—						3.31	0.54
2. Grandiosity	.20*	—					3.42	1.00
3. Entitlement	.05	.36**	—				2.51	0.87
4. Optimism	.20*	.22*	−.02	—			3.28	0.30
5. Positive	.09	.38**	.09	.57**	—		2.80	0.46
6. Negative	.17	−.04	−.11	.74**	−.13	—	3.66	0.45

Note. N = 107;

^*p < .05

^**p < .01

5. Study 2

In Study 1, we demonstrated that when unrealistic optimism, an intrapersonal self-enhancement strategy, was simultaneously regressed on grandiosity, entitlement, and the grandiosity × entitlement interaction, only grandiosity was significantly predictive of optimism. Furthermore, this association was not mediated by self-esteem. In Study 2, we turned our attention to a more interpersonal self-serving strategy, namely, ethical violations. Previous research has demonstrated that entitlement is related to overt violations of social norms (Campbell, Bonacci, et al., 2004; Brown et al., 2009). We sought to replicate and extend these previous findings in two ways. First, we used a sample of graduate students instead of introductory psychology students. Second, we used an ethical decision-making measure developed to evaluate the effects of research ethics training (Mumford et al., 2006). Thus, the ethical dilemmas participants were asked to solve were complex, realistic depictions of situations the participants might actually experience in their professional careers. We hypothesized that of the variables of interest (i.e., entitlement, grandiosity, and their interaction), only entitlement would be predictive (negatively) of ethical decision-making.

6. Method

6.1. Participants

The study sample included 134 graduate students (46 males and 80 females; 8 did not specify gender) attending the University of Oklahoma, with a mean age of 29.3 (SD = 6.3) years. All participants were enrolled in a research-oriented program in the social (28%), biological (44%), or health sciences (28%). Eighty participants (59.7%) identified themselves as Caucasian, 27 (20.1%) as Asian, 6 (4.5%) as African American, 4 (3.0%) as Hispanic, 2 (1.5%) as Middle Eastern, and 6 (4.4%) as other (or did not report). Because 13 individuals did not provide data on one or more of the measures, the data of the remaining 121 participants were used for all analyses.

6.2. Measures

6.2.1. Predictors

Grandiosity and entitlement were assessed using the NaGS (α = .93) and the PES (α = .90), respectively. Participants also completed the Balanced Inventory of Desirable Responding (BIDR; Paulhus, 1991), a 40 item measure of response bias consisting of two, 20-item subscales: self-deceptive enhancement (SDE; α = .71) and impression management (IM; α = .74).

6.2.2. Ethical decision-making

Ethical decision-making was assessed by asking participants to first read 12 “parent” scenarios relevant to their discipline (i.e., social, biological, or health sciences), which had been developed using case studies of actual ethical violations. Each scenario contained general background information followed by six events (only three of which were ethically charged) that might plausibly occur given the information in the parent scenario. Each event was followed by six to eight response options. Participants were asked to imagine themselves as the actor in the scenario and choose what they considered to be the two best courses of action for each event. For the ethically charged events, the response options were evenly split between high ethicality, moderate ethicality, and low ethicality responses (as rated by an independent panel consisting of three psychologists and a subject matter expert in the relevant area; for more information of the validation and construction of this measure, see Mumford, et al., 2006). Ethicality scores were created by assigning numerical values to responses of high ethicality (3), moderate ethicality (2), and low ethicality (1). Participants’ overall score was computed by averaging the ethicality scores from each scenario.

6.3. Procedure

The current study was part of a larger project investigating research integrity. Participants were recruited via flyers, e-mail, and telephone. The study was described as an investigation of prior educational experiences on integrity and problem solving, and participants were offered $100 as compensation (participants spent 4 to 5 hours completing measures as part of the larger study). Participants were run in groups ranging from 1 to 20 but completed all measures individually.

After arriving at the laboratory and providing informed consent, participants completed a battery of individual difference measures including the NaGS, the PES, and the BIDR. Afterwards, all participants completed the measure of ethical decision-making, which was described as a problem-solving exercise for realistic workplace scenarios in order to help reduce socially desirable responding.

7. Results and Discussion

Overall, participants were moderately ethical in their responses (M = 2.07; SD = 0.19). Grandiosity scores (M = 3.73; SD = 1.12) were somewhat higher than in Study 1, t(226) = 2.19, p < .03, d = −0.29. Entitlement scores were also higher than those in Study 1, but still below the midpoint of the scale (M = 3.43; SD = 1.24), t(226) = 6.41, p < .001, d = −0.86 (see Table 1 for Study 1 means).

In order to examine our hypothesis regarding the associations between entitlement, grandiosity, and ethical decision-making, we regressed participants’ ethical decision-making scores on entitlement, grandiosity, the grandiosity × entitlement interaction (after mean-centering grandiosity and entitlement), SDE, IM, age, and gender. Neither grandiosity, β = .04, t(114) = −0.36, ns, nor the grandiosity × entitlement interaction, β = −.06, t(114) = −0.67, ns, was associated with ethical decision-making. Entitlement, however, was negatively predictive of ethical decision-making, β = −.21, t(114) = −2.00, p < .05, sr²= .03.¹ The regression weights for IM, SDE, age, and gender were not significant, all βs < |.14|, t(114) < |1.34|, ns. As predicted, respondents scoring higher in entitlement chose less ethical options on a realistic index of ethical decision-making.

8. General Discussion

The present studies provide additional support for a model of narcissism characterized by an intrapersonal dimension of grandiosity and an interpersonal dimension of entitlement (Brown et al., 2009). In Study 1, regression analyses demonstrated that only grandiosity explained significant variance in unrealistic optimism (above and beyond self-esteem, entitlement, and the grandiosity × entitlement interaction), particularly for positive events. In Study 2, entitlement was the only variable that significantly predicted ethical decision-making in a regression model including grandiosity, the grandiosity × entitlement interaction, age, gender, IM and SDE.

Brown et al. (2009) suggested that instead of measuring narcissism as a unitary construct, researchers should assess what appear to be the two primary dimensions of narcissism, grandiosity and entitlement, separately. In two studies, we showed that grandiosity and entitlement are predictive of different self-serving strategies. On one hand, unrealistic optimism, a relatively intrapersonal self-enhancement strategy, was predicted only by grandiosity. On the other hand, unethical decision-making, a relatively interpersonal self-promotional strategy, was predicted only by entitlement. Entitlement (Study 1), grandiosity (Study 2), and the grandiosity × entitlement interaction (both studies) failed to explain any additional variance, again demonstrating the uniqueness of these dimensions in predicting different outcomes.

Interestingly, the relationship between grandiosity and unrealistic optimism in Study 1 appeared to be driven by the association between grandiosity and optimism for positive events. This finding could suggest that grandiosity is primarily associated with intrapersonal strategies that accentuate positivity, rather than discounting negativity. Consistent with this possibility, Rhodewalt and Morf (1995) reported that the NPI was positively correlated with self-serving attributions for positive (but not negative) events. Similarly, Foster et al. (2009) found that the association between narcissism and risk-taking was partially mediated by the perceived benefits of risky behaviors, but narcissism was unrelated to perceived costs.

A strength of the present work is that, in contrast to most studies involving nonclinical narcissism, Study 2 used a sample of graduate students from a variety of disciplines. However, this strength is also a source of potential weakness. Not only are graduate students different from the general population in a variety of ways (e.g., intelligence, education), but all participants in this study were also voluntarily enrolled in an ethics-training course. As such, participants in this study were likely more concerned than both the general population and the average undergraduate with behaving ethically. Nonetheless, the average ethicality score for participants was only 2.07 (SD = 0.19), making ceiling effects unlikely, and any restrictions in range on our ethicality measure would only serve to reduce associations with our predictors.

Another limitation of Study 2 is that participants did not engage in unethical behavior, but responded to hypothetical scenarios involving fictional actors. However, it is important to note that participants were asked not what course of action they think the actor would take, but what course of action the actor should take. Thus, even though we cannot conclude that entitlement was associated with unethical behavior per se, it was still predictive of unethical preferences.

A more general limitation is that we only examined one type of self-enhancing and self-promotional strategy in each study. Although narcissism is associated with a variety of intrapersonal self-serving strategies (Morf & Rhodewalt, 2001) and anti-social behaviors (e.g., Twenge & Campbell, 2003), previous studies have overwhelmingly used either the composite NPI or its subscales to measure narcissism. Future research should corroborate these results by examining additional intrapersonal and interpersonal variables with measures designed to independently measure grandiosity and entitlement (e.g., the NaGS and the PES).

The current studies demonstrate the value of measuring narcissism as a multidimensional construct composed of grandiosity and entitlement. Previous research on narcissism has painted a picture of narcissists as being in constant need of maintaining their self-esteem and sense of superiority and as generally feeling unconcerned with the needs and welfare of others. However, assessing narcissism as a unitary construct potentially obscures a more nuanced relationship with narcissism and its associated outcomes. By measuring grandiosity and entitlement separately with measures specifically designed to tap their respective constructs, we were able to show that each dimension independently and uniquely predicted a different “self-serving” orientation.

Several researchers (e.g., Emmons, 1984) have argued that the E/E or entitlement subscale of the NPI represent a more maladaptive form of narcissism, whereas the other subscales represent a relatively adaptive form. Even though narcissism, and grandiosity in particular, is associated with psychological well-being (Brown et al., 2009; Hickman et al., 1996; Sedikides, Rudich, Gregg, Kumashiro, & Rusbult, 2004), it is also associated with cheating in ambiguous circumstances (Brown et al., 2009; Study 3). Thus, although grandiosity might be considered “adaptive” in that it fosters optimism and potentially other positive illusions (Taylor & Brown, 1988), it might have important negative consequences as well, limiting the appropriateness of calling it “adaptive.” Regardless, further research is necessary to determine the extent to which grandiosity and entitlement diverge or converge in predicting additional attitudes and outcomes theoretically associated with narcissism.

Highlights.

• Grandiosity and entitlement were associated with different self-serving strategies.

• Narcissistic grandiosity was positively associated with unrealistic optimism.

• Narcissistic entitlement was negatively associated with ethical decision-making.

• Further support for a distinction between grandiosity and entitlement was obtained.