The Effects of Anticipated Negative Feedback on Psychological States Among Narcissists

Although narcissism has long been researched in relation to anger, previous research examined narcissistic anger toward negative feedback that had already occurred. In this study, we investigated the effects of anticipation of evaluation (present vs. absent) and negative feedback (present vs. absent), using a creativity task paradigm, on state anger scores among 231 U.S. undergraduates (76% White, 60% women). We also measured undergraduates’ narcissistic tendencies and impressions of the creativity task. Multiple regression analyses revealed a significant interaction between narcissism and negative feedback on total anger scores, with narcissists responding with more anger than non-narcissists in the condition of negative feedback. We also found a significant two-way interaction between narcissism and anticipation of evaluation on total enjoyment scores. Anticipation of feedback inhibited narcissist-prone individuals from enjoying the task in the anticipation condition, but this pattern was not present in the no-anticipation condition. Implications and recommendations to better understand the nature of narcissism are discussed.


Narcissism and Anger
In their original "threatened egotism model," Baumeister et al. (1996) explained that there is a dark side of high selfesteem. They also consider the case of high but unstable selfesteem, which is one aspect of narcissism (cf. Baumeister, 1993;Bushman & Huesmann, 2010;Kernis, Grannemann, & Barclay, 1989). According to their model, when those with high and unstable self-esteem receive negative evaluations, a perceived threat (i.e., "threatened egotism") emerges because their inflated but uncertain self-worth has already been derogated or may be derogated in the future. Consequently, they attempt to reestablish and maintain a favorable view of self by rejecting all negative feedback. This process involves a negative emotion (i.e., anger) toward the source of the threat (i.e., negative feedback) and leads people with high/unstable self-esteem to express their anger in the form of aggression.
Thus, anger can be provoked when narcissists' aggrandized, but fragile self-worth, is threatened by negative 650921S GOXXX10.1177/2158244016650921SAGE OpenMatsuo and DeSouza research-article2016 1 Nagoya University, Japan 2 Illinois State University, Normal, USA feedback from others. In addition, anger is not only a response to threatened self-esteem but also a means of regaining and retaining one's sense of superiority. That is, anger and aggressive behavior often function as a symbolic dominance over others (Baumeister et al., 1996).

Paradox in Narcissists and Unexplored Areas
Although previous researchers have investigated the underlying cognitive processes in narcissism, they discuss just one side of narcissists' competitiveness and the need for admiration, which is their active involvement in competition and interactions with others to flaunt their superiority over them.
Regarding this notion, Elliot and Thrash (2001) pointed out the "paradox of narcissism." Thus, poor performance may result in negative feedback from others or lowering selfesteem, but narcissists must actively get involved in competition because that is the only way to maintain their self-esteem. Elliot and Thrash refer to the paradox of narcissism as the "fear of failure," in which narcissists compete to validate their ideal self, but simultaneously fear negative feedback. Therefore, to protect their self-esteem from possible threats, narcissists become sensitive to feedback-related situations, which results in fragile high/unstable self-esteem. When narcissists perceive a situation as an ego-threatening event, they feel distressed and may become angry even before the actual feedback is provided (cf. Barry, Chaplin, & Grafeman, 2006;Baumeister et al., 1996). That is, if narcissists anticipate that they might not perform well even when failure has not actually happened, they are frustrated and may become angry. This anger stems from narcissists' notion that they are afraid of failure and want to reject all actual and possible events that may cause a reduction in self-esteem.
However, scant research has been done about the paradoxical characteristic (i.e., "fear of failure") in narcissists in relation to anger. The existing literature has focused only on the performance-based threat that has already happened. In addition to emphasizing private events (i.e., one's performance) alone, most of the literature has only manipulated the positive or negative feedback condition and focused on the outcome of the feedback brought by others (e.g., Barry et al., 2006;Bond, Ruaro, & Wingrove, 2006;Bushman & Baumeister, 1998;Rhodewalt & Morf, 1998;Stucke & Sporer, 2002). This methodology lacks the perspective of the "fear of failure" that accompanies anticipation of negative results. It also lacks variation in experimental conditions, because it is based on the anger effect elicited by feedback that has already been given. However, narcissistic anger, which is an expression of "fear of failure," should also be observed before actual feedback is given.
The anticipation of negative feedback can possibly elicit narcissists' anger by the frustration that stems from perceived internal threat when they judge they are unlikely to surpass others in a given task.
It is possible that they would become angry just by the anticipation of negative evaluation, which would evoke as strong reaction as actual negative evaluation about their performance. If narcissists' anger might come just from anticipation, it might potentially be directed at anyone around them, not just the individuals who would give them negative evaluation. The anticipation-based anger would then possibly hurt narcissists' interpersonal relationships because others would be unwilling to interact with narcissists whose anger would appear to happen for no good reason. Consequently, others would be unwilling to interact with narcissists.
Gender is also an understudied area in the narcissismanger link, although some previous research suggested the importance of taking gender differences into consideration. According to the American Psychiatric Association (2013), up to 75% of those diagnosed with narcissistic personality disorder are men. The literature on narcissism indicates that men tend to show more narcissistic characteristics than women as ways to control the situation, dominate others, and exhibit their excellence (Carroll, 1987;Philipson, 1985;Richman & Flaherty, 1990), which are congruent with traditional male gender roles. Also, narcissists' affective states about a task depended on participants' gender in relative evaluation situations (Morf, Weir, & Davidov, 2000). Morf et al. (2000) found that male narcissists experienced more enjoyment and positive emotions in the conditions where performance evaluation was implied, but this pattern was not found among female narcissists, using a college student sample scoring in the top and bottom third score range on the Narcissistic Personality Inventory (NPI; Raskin & Hall, 1979, 1981. Few studies examined gender differences regarding narcissism and anger, which is of interest to us. For example, Hibino, Yukawa, Kodama, and Yoshida (2005) examined narcissism as one of the possible variables involving anger expression in a Japanese junior high school student sample. In a class period, participants voluntarily answered a retrospective questionnaire on an anger-triggering event. They found that a narcissistic characteristic (excessive need for admiration) took different paths to aggressive behavior by gender. For boys, the link was straightforward; excessive need for admiration led to anger, which motivated them to engage in aggressive behavior (e.g., revenge on the person who caused the event, complaint about the event to others, and displacement of anger toward objects). However, girls' excessive need for admiration led to a cognitive process that emphasized their victimhood, which motivated them to act aggressively. That is, unlike boys, girls tended to overestimate the seriousness of an anger-triggering event; they perceived being treated poorly, resulting in aggressive behavior.
In addition, Barry et al. (2006) found a strong positive relationship between narcissism and expressed aggression, with men showing this trend more strongly than women in a college student sample. Similarly, Thomaes, Stegge, Olthof, Bushman, and Nezlek (2011) found that male, but not female, narcissists expressed anger when humiliated in Study 2. Thus, men and women express anger differently.
Concerning violence (i.e., a more serious form of expressing aggression) in heterosexual dating relationships, Ryan, Weikel, and Sprechini (2008) found that significant relationships between college students' scores on physical assault/ sexual coercion and three subcategories of narcissism (i.e., overt narcissism, covert narcissism, and sexual narcissism) depended on gender; only men showed significant correlations between covert narcissism and physical assault and between sexual narcissism and sexual coercion.
Considering that narcissists desire to participate in competition and expected to win, the investigation of narcissists' psychological states about such competition is also of interest to us. Some previous work found that narcissism is related to positive psychological states (e.g., enjoyment) and negative ones (e.g., fear) simultaneously (Rose, 2002; see also Wink, 1991). Narcissists think positively about their own performance (albeit unrealistically). However, they are also nervous about their performance and damaging their selfimage, which causes them to be vulnerable to threat. These conflicting psychological states should be observed when asked about the task they have engaged in. It is possible that they feel negative about the task itself (that brought unfavorable feedback) because the task is the source of their frustration. Their negative internal states would be a contributor to support narcissists' "fear of failure."

The Current Study
We experimentally tested the narcissistic paradox of "fear of failure." Along with anger and gender, we also examined enjoyment, interest, and boredom. We argue that anticipation of negative feedback from others activates narcissists' ambivalent attitudes (i.e., activating both their participation in a task and their reluctance to engage in it due to fear of failure). Narcissists are not intrinsically motivated to engage in the task (cf. Morf et al., 2000); therefore, they would not enjoy it.
Likewise, because they cannot accept anything that casts doubt on their superiority, they would not find the task interesting when evaluation is announced beforehand. Then, in the case of anticipated outcomes, to protect their self-esteem, we expect narcissists to view the task as unenjoyable.
Based on the extant literature and the rationale above, we propose the following hypotheses: Hypothesis 1: College students who score high on narcissistic tendencies will respond to anticipated negative feedback with more anger than college students who score low on narcissistic tendencies, Hypothesis 2: Those who score high on narcissistic tendencies will also respond to actual negative feedback with more anger than those who score low on narcissistic tendencies, Hypothesis 3: Male narcissists will show the above patterns more strongly than female narcissists, and Hypothesis 4: Narcissists will enjoy the assigned task less (i.e., show less interest and more boredom) than nonnarcissists when they anticipate being evaluated.

Participants
The sample consisted of 237 (n = 143 or 60% women) undergraduates enrolled in general education classes at a large U.S. Midwestern public university with a mean age of 19.36 years (SD = 1.76). The majority of participants self-reported to be White (n = 181, 76.4%). Participation was voluntary. There was no financial compensation given, but students were offered the opportunity to earn extra credit points for their participation.

Measures and Task
NPI. Originally constructed as a 54-item scale aiming at measuring one's narcissistic tendency, the NPI is composed of sentences that are answered in a "yes" (coded 1) and "no" (coded 0) format. Responses are summed to create a narcissistic tendency score. Higher scores indicate higher levels of narcissism. A sample item includes, "I have a natural talent for influencing people" (Raskin & Hall, 1979, 1981. Unlike instruments that are designed to measure pathological narcissism, the NPI identifies individual differences in narcissistic tendencies in a nonclinical population (cf. Emmons, 1984;Raskin & Terry, 1988). Raskin and Hall (1981) reported a strong correlation between the 40-item NPI with its 54-item version (r = .98). del Rosario and White (2005) reported a Cronbach's alpha coefficient of .80 with a sample of college students for the 40-item NPI. We used the 40-item short version of the NPI in the current study, which had a Cronbach's alpha of .76.

State-Trait Anger Expression Inventory (STAXI).
We used the 10-item State Anger subscale (Spielberger, 1988). Each item is rated on a Likert-type scale, ranging from 1 (not at all/ almost never) to 4 (very much/almost always). Responses are summed, with higher scores indicating higher levels of anger. A sample item includes, "I feel like hitting something." Using a college sample, Fuqua et al. (1991) reported an alpha of .91 for the State Anger subscale. In our study, the Cronbach's alpha coefficient for the STAXI was .88.
Creativity test. The "Lange-Elliot Creativity Test" asks participants to come up with as many uses as possible for a brick (Part 1) and a candle (Part 2), and each of the two task lasts about 5 min (Sedikides, Campbell, Reeder, & Elliot, 1998). In the current study, we only used Part 1 (i.e., uses of a brick) for the sake of time. In the feedback-anticipating conditions, there was a clear statement regarding feedback. Although participants were led to believe that this creativity test was an established measure of one's creativity level, this bogus test served as an unfamiliar task without practice, which was new to everyone and seemed rather difficult (i.e., threat for narcissists).
Impressions of the task. Based on Sedikides et al.'s (1998) study, we asked participants three questions about their impressions of the task: (a) How much they enjoyed the creativity task, (b) how interesting it was, and (c) how boring it was. They rated each question on a 4-point, Likert-type scale that ranged from 1 (not at all/almost never) to 4 (very much/ almost always). We summed their scores (the boring item was reverse-scored), creating a total enjoyment index score that ranged from 3 to 12 (α = .79).

Experimental Manipulations
There were four experimental conditions: feedback-anticipation/ negative feedback given, feedback anticipation/no negative feedback given, no anticipation/negative feedback given, and no anticipation/no negative feedback given. In all conditions and before the creativity task began, participants were informed that the experimenter had already obtained reliable norms concerning the task at 130 students from the same university and had found a strong relationship between high scores on the task and high IQ level (i.e., intelligent students did well on the task). This cover story was intended to give participants a good opportunity to show evidence of their intellectual ability and also to stimulate their competitive motivation. The experimenter added that they were now collecting more data.
In the feedback-anticipating condition, at the time the directions of the task was given (i.e., before the task), participants were told that they would receive the evaluation of their performance after the task in the form of percentile rank. The instructions for the no-anticipation condition did not include any possibility of feedback.
Concerning actual feedback, after the creativity task, participants in the no-negative feedback condition received nothing, whereas participants in the negative feedback condition received a bogus performance evaluation sheet, which indicated that they performed poorly on the task and clearly showing that their score was on the 31st percentile (significantly below the norm). The experimenter also verbally told them that they did poorly on the task.

Procedure
All participants in each session (one to two same-sex students) were randomly assigned to one of the four conditions. The participants signed the consent form and completed the demographic questionnaire and the NPI in the same room. After that, we took each participant to a separate room where he or she worked individually on the creativity task with the manipulations described earlier. Next, participants responded to the 10-item State Anger subscale that was embedded randomly with other items that asked about enjoyment, interest, and boredom about the creativity task (cf. Sedikides et al., 1998).
On completion, all participants were thoroughly debriefed, emphasizing that the feedback sheet was created to make the performance appear negative, and performance was not actually evaluated. They were told that the experimenters would examine participants' reactions to the task after some were led to believe that they had performed poorly. Also, participants were told that the task they performed was not a legitimate test of creativity and that confidentiality would be kept at all times.

Results
Six participants were excluded from the data analysis because three were suspicious about the true purpose of this study, two did not follow the directions on the creativity task, and one had a session that was seriously interrupted by another participant who arrived late, reducing the data set from 237 to 231. Of these, 57 participants were assigned to the anticipation/no-negative feedback condition, 59 to the anticipation/negative feedback condition, 58 to the no anticipation/ no-negative feedback condition, and 57 were to the no anticipation/negative feedback condition. Table 1 shows the means and standard deviations for the total rating scores by gender.
Three multiple regression analyses were performed after centering, with each condition of anticipation/no anticipation, negative feedback/no negative feedback, and man/ woman dummy coded. We conducted separate analyses focusing on negative feedback and anticipation because such investigation would show the clearest effects. Concerning the first analysis, the top half of Table 2 shows significant main effects of the total NPI score and negative feedback. The interaction of negative feedback and the total NPI score was also significant; the interaction of negative feedback with gender was not significant; last, three-way interaction was not significant. A simple slope test revealed that nonnarcissists experienced significantly greater anger when negative feedback was given than when the feedback was not presented, but no difference was observed for narcissist about the level of anger between no-negative feedback condition and actual-negative feedback condition. For the anticipation condition (see bottom half of Table 2), we did not find any significant main effects, two-way interactions, or threeway interaction on the total anger score.
These results supported Hypothesis 2, with narcissists having higher anger scores than non-narcissists in the actualnegative feedback condition. However, Hypothesis 1 (anticipation of negative feedback) and Hypothesis 3 (gender differences) were not supported.
Finally, we ran a separate multiple regression analysis (after centering) with the total NPI score anticipation, and their interaction as predictors and the total enjoyment score as criterion variable. Table 3 shows that the interaction of the total NPI score and anticipation was statistically significant, supporting Hypothesis 4. That is, high NPI scorers reported feeling less enjoyment in the task when they were informed that they would be given feedback (anticipation condition) than low NPI scorers. However, this pattern was not present in the case of the no-anticipation condition.
A series of post hoc power analyses were conducted using the software package, G*Power (Faul, Erdfelder, Lang, & Buchner, 2007). Along with our sample size of 231, we found the observed power for the effects of the first analysis about negative feedback, the second about anticipation, and the third about enjoyment were .97, above .99, and above .93, respectively.

Discussion
We found that narcissists' anger was provoked by actual negative feedback about their performance on the creativity task, which is in keeping with previous studies (e.g., Barry et al., 2006;Baumeister et al., 1996;Stucke & Sporer, 2002). We did not find support for the anticipation of feedback hypothesis. We speculate that having told all participants that the creativity task was valid and was related to intelligence may have prompted some anticipation of evaluation or judgment in all four conditions; in addition, presumably the experimenter would know the "results" of the creativity task for each participant regardless of feedback given. Future studies should ask participants how much they anticipated feedback and how negative the feedback was to them to ensure that such manipulations are indeed relevant to our participants.
It is also feasible that our fear of failure model may need to be modified, as third variable(s) may be acting as mediators or moderators, affecting the narcissism-anger relationship. For example, Stucke and Sporer (2002) indicated that self-concept clarity is an important factor in considering possible interactions between narcissism and performance feedback on anger expression.
Our study still contributes to the literature because it revealed that anticipation of being evaluated affected participants' reactions to the task, suggesting that narcissist-prone individuals were sensitive to the situation and showed their psychological states differently from non-narcissist-prone individuals. It is noteworthy that anticipation of evaluation prevented narcissist-prone individuals from enjoying the creativity task. We speculate that this pattern reflects narcissists' inner ambivalence between active participation in a competitive situation to validate their greatness and reluctance to engage in an unfamiliar task that might cause failure. When narcissists feel bored about a task and do not enjoy it, they may be actually reluctant to engage in it because they do not want to experience unwanted outcomes (i.e., fear of failure). However, they need to get involved in the competitive situation anyway because of their desire to obtain admiration from others. As discussed earlier, narcissism features competitive motivation and need for admiration. These two characteristics could result from their tendency to participate actively, which is one aspect of narcissism. Furthermore, narcissists want to be in the public eye and involve themselves in competitive  Note. NPI total score was centered at its mean. Total enjoyment score is the sum of enjoyment, interest, and boring (reverse-scored) items. NPI = Narcissistic Personality Inventory.
situations with others because these situations offer opportunities to boast of their greatness and superiority over others. We cast light on the "paradox of narcissism" theory developed by Elliot and Thrash (2001) and empirically tested it for the first time. No current theory predicts our findings on anticipation of negative feedback, but our attempt was at least a good starting point to examine unexplored areas of narcissism. That is, our findings captured narcissists' impression of a task before, not after, receiving feedback about the task. As the previous research (e.g., Morf et al., 2000;Rhodewalt & Morf, 1998) focused on the psychological states caused by performance feedback (i.e., after feedback), the current study did cast light on a possible cognitive process when narcissists are waiting for feedback.
Narcissists are often described as having such confidence about their competency that they eagerly participate in competitive situations (cf. Morf et al., 2000). Yet, the current findings suggest that they do not enjoy these situations because of upcoming feedback. Together with the psychological unstableness of narcissists' self-esteem, their inner ambivalence between the undesirable self (possibly performing poorly) and their desire to do better than others emerge when they anticipate evaluation. Narcissists' derogation of the task could be a manifestation of their ambivalence between taking advantage of competitions (approach) and avoiding such opportunities for the purpose of protecting their self-esteem.
We recommend that future studies examine the "paradox of narcissism" by experimentally manipulating approach (e.g., desire to show off one's abilities) and avoidance (e.g., fear of anticipating/receiving negative feedback) conditions separately, and possibly showing their simultaneous activation. It might also be helpful to test whether the effect of feedback anticipation on anger by manipulating how many people would know the results of the creativity task (e.g., just the experimenter, as was the case in the current study, vs. many individuals, such as other research participants or confederates) and to include self-concept clarity (Stucke & Sporer, 2002) as a possible mediator/moderator.
Although the current study did not find evidence of significant gender differences concerning the narcissism-anger link, future research should continue to examine the dispositional aspect of narcissism by gender. According to Morf and Rhodewalt (2001), it is possible that narcissists try to achieve their goals in different and unique ways suitable to their gender roles. In the United States, hegemonic masculinity reinforces success, power, and competition, which have been linked to aggression and violence (Crowther, Goodson, McGuire, & Dickson, 2013), whereas "emphasized femininity" reinforces feminine submissiveness (DeSouza, 2013). Following such gender roles, narcissistic characteristic(s) and expressiveness of anger may be different for women and men (Ryan et al., 2008), especially in Latin countries where machismo and marianismo (emphasized femininity) seem to be even stronger than in North America (Baldwin & DeSouza, 2001). Thus, gender differences may be pronounced in more traditional cultures than in more egalitarian cultures such as in Scandinavia. We recommend a cross-cultural investigation to test for gender differences in the narcissism-anger link.

Conclusion
Our study provided theoretical insights into personality psychological science by identifying variables in narcissism that were affected by anticipation of negative feedback. Narcissists are situation-sensitive; their unexpected anger and reactions (witnessed by other people) are likely to have negative effects on their engagement in tasks, actual achievement, and interpersonal relationships. Narcissists' sensitivity to their social worlds (i.e., reactions about their performance from others) originally comes from the paradoxical concept they bearthey want to be superior to others, but they are worried about possible negative feedback. Their self-evaluation derives not from themselves, but from their environment, which results in high/unstable self-esteem. As Morf and Rhodewalt (2001) discussed, narcissism should be viewed as a self-regulatory processing system with paradoxical features.
Future research on narcissism should be interwoven with other self-related concepts and attitudes toward the external world in cognitive processes working for establishing narcissists' self-understanding and their interpersonal relationships to better understand the complicated mechanisms of narcissism and its dynamism.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) received no financial support for the research and/or authorship of this article.