Narcissism and Anger : Self-Esteem and Contingencies of Self-Worth as Mediating Self-Structures

Theory and research suggest that an internalization of psychological “structure” related to self-esteemmaymediate relationships of Maladaptive Narcissism with higher and Adaptive Narcissism with lower Anger. In the present study (N = 623), Self-Esteem and Contingency of Self-Worth Scales served as presumed indices of the presence or absence of structure. Maladaptive Narcissism predicted greater Anger and a relative absence of structure whereas Adaptive Narcissism displayed an opposite pattern of results. Constructs assessing structure partially mediated the Maladaptive Narcissism relationship with greater Anger. Mediation analyses also revealed that structure not only fully mediated the inverse Adaptive Narcissism linkage with Anger, but also suppressed an association with greater Anger that would otherwise have been evident. These data supported the idea that psychological structure related to self-esteem is noteworthy in explaining the implications of narcissism for personality and interpersonal functioning.

Interpersona | An International Journal on Personal Relationships interpersona.psychopen.eu| 1981-6472 therefore, exist initially outside the self in the social environment, as supportive selfobjects offer the soothing love and care that makes it possible for a child to experience an esteeming of the self even in disturbing circumstances.
If objects gradually and optimally frustrate this need for soothing, the individual will internalize a capacity to do for himself psychologically what previously was done for him socially.In other words, "structure" will develop within the personality that enables the individual to esteem the self even in the absence of social support.On the other hand, internal structure will fail to develop if selfobjects fail to meet basic esteem needs in an age appropriate fashion or if they traumatically frustrate the child.Maladaptive narcissism will be the result.Removal of social support will then be experienced as a threat, and anger, rage, and aggression will follow as a self-protective response.
Efforts to explore the role of "structure" in narcissistic anger must address the complexity of narcissism (Watson, 2005).Kohut's (1977) framework suggests a continuum of self-esteem regulation that ranges from maladaptive narcissism to progressively more adaptive forms of narcissism as structure develops within the self.Research in fact demonstrates that indices of narcissism record both more adaptive and more maladaptive forms of personality functioning, and that the mental health implications of one become more obvious after statistical procedures control for the other (Watson, Little, Sawrie, & Biderman, 1992;Watson, Sawrie, Greene, & Arredondo, 2002).In other words, maladaptive narcissism becomes more unambiguously maladaptive when procedures control for its covariance with adaptive narcissism.Conversely, adaptive narcissism becomes even more obviously adaptive after controlling for maladaptive narcissism.The obvious presumption would be that maladaptive narcissism would predict less evidence of structure and higher anger whereas adaptive narcissism would predict more structure and less anger.In addition to this obvious possibility, the present project most importantly tested the hypothesis that "structure" helps explain the linkages of narcissism with anger.Evidence supporting that suggestion would appear if presumed indices of the presence or absence of structure mediated such relationships.

Measures of Structure
One clearly relevant marker of structure would be a Self-Esteem Scale (Goldberg, 1999).Additional indices seem available in the Contingencies of Self-Worth Scale (Crocker, Luhtanen, Cooper, & Bouvrette, 2003).This instrument assesses 7 dimensions of personality functioning in which personal self-worth depends upon the achievement of goals.Two dimensions reflect "internal contingencies of self-worth" and point toward the presence of internalized self-structure.God's Love operationalizes self-esteem based upon the love and acceptance of God, and Virtue records the self-approval that follows from abilities to meet internalized ethical standards.Crocker et al. (2003)  Approval of Others makes self-esteem rest upon how other people evaluate the individual.Competition makes self-esteem a product of successfully bettering others.Crocker et al. (2003) discovered that all three of these external contingencies correlated negatively with self-esteem and positively with neuroticism.In other words, these external dimensions of contingent self-worth served as indicators of an absence of structure.

Hypotheses
In summary, this study i explored the possibility that variables indicative of Kohut's (1977)  ).In terms of race, the majority of the sample was Caucasian (73.5%), with 18.9% African-American, 2.4% Hispanic, 1.6% Asian/Oriental, 0.6% Middle-Eastern, and 2.9% "other" or simply failing to respond.

Measures
All psychological scales appeared in a single questionnaire booklet that contained measures for use in several projects.The first page asked participants to report basic background information.Then, after an initial section that obtained data for a different investigation, the booklet presented all instruments in the order in which they are described below.All but the Narcissistic Personality Inventory (NPI: Raskin & Hall, 1981) utilized a 0 to 4 Likert scale.
This inventory uses a forced-choice format in which participants choose between a narcissistic (1) and a nonnarcissistic (0) response.Illustrative of the 11-item Exploitative/Entitlement (E/E) factor was the narcissistic assertion that "I expect a great deal from other people," in contrast to the non-narcissistic claim that "I like to do things for other people."Nine items defined a Leadership/Authority (L/A) factor which was exemplified in the self-report that "I see myself as a good leader."The non-narcissistic option said, "I am not sure that I would make a good leader." The Superiority/Arrogance (S/A) factor contained 11 items with a representative item suggesting, "I can make anybody believe anything."The more modest non-narcissistic choice said that "people sometimes believe what I tell them."Nine items defined Self-Absorption/Self-Admiration (S/S).Illustrative of this factor was the narcissistic self-report, "I like to look at my body," as opposed to the non-narcissistic alternative that stated, "My body is nothing special." Preliminary analyses indicated that all NPI data could be meaningfully and more conveniently examined by dichotomizing the factors into measures of Adaptive and Maladaptive Narcissism.Based on previous research and on a preliminary examination of the present data to ensure the validity of this approach, Adaptive Narcissism (α = .76,M = 0.50, SD = 0.20) reflected the average score on the L/A, S/A, and S/S factors.E/E defined Maladaptive Narcissism (α = .53,M = 0.33, SD = 0.20).Prior to creating these variables, procedures addressed the potential problem of cross-loadings.With regard to the narcissistic claim that "I am assertive," Emmons (1984) reported that this item helped define both L/A and S/A.The stronger loading, nevertheless, appeared on L/A, resulting in its elimination from S/A in this study.Another narcissistic option stated, "People always seem to recognize my authority."Emmons found that this statement loaded on both E/E and L/A.The stronger loading appeared for L/A; so, computation of E/E did not include this item.

Contingencies of Self-Worth -
The Contingences of Self-Worth Scale uses 35 items to record the seven dimensions of contingent self-esteem (Crocker et al., 2003).Five items express each domain.Again, five of the seven domains reflect external sources of contingent self-worth.An illustrative "Physical Appearance" (α = .69,M = 2.55, SD = 0.75) item said, "My self-esteem is influenced by how attractive I think my face or facial features are."Selfworth based upon "Competition" (α = .81,M = 2.57, SD = 0.82) appeared in such claims as, "My self-worth is affected by how well I do when I am competing with others."Indicative of "Academic Competence" (α = .78,M = 2.88, SD = 0.77) was the assertion, "I feel bad about myself whenever my academic performance is lacking." Expressing "Family Support" (α = .74,M = 3.04, SD = 0.71) was, for example, the statement, "Knowing that my family members love me makes me feel good about myself."Indicative of "Approval of Others" (α = .76,M = 1.78,SD = 0.94) was the admission, "My self-esteem depends on the opinions others hold of me."Two other measures operationalized internal contingencies of self-worth.A representative expression of "Virtue" (α = .75,M = 2.79, SD = 0.77) said, "Doing something I know is wrong makes me lose my self-respect.""God's Love" (α = .93,M = 2.80, SD = 1.17) appeared in such statements as, "I feel worthwhile when I have God's love."

Procedure
All research procedures complied with institutional guidelines for ethical research.Student participation was fully voluntary, completely anonymous, and rewarded with extra course credit.Statistical procedures scored all measures in term of the average response per item.

Results
Given the much larger number of women than men in this sample, preliminary analyses examined whether this Partial correlations controlling for gender appear for all but the narcissism variables in Table 2. Anger correlated negatively with Self-Esteem, Family Support, God's Love, and Virtue.Anger also predicted higher levels of Physical Appearance and Competition.Self-Esteem, in contrast, correlated negatively with Physical Appearance, Competition, and Approval of Others and positively with Family Support, God's Love, and Virtue.Of the 21 rela-Interpersona 2015, Vol.9(1), 59-71 doi:10.5964/ijpr.v9i1.163tionships among the Contingency of Self-Worth constructs, all but three associations observed for God's Love and one for Virtue proved to be significant and positive.Table 3 describes partial correlational findings for Adaptive and Maladaptive Narcissism.After controlling for only gender, Adaptive Narcissism predicted greater Self-Esteem and God's Love and lower Approval of Others; however, it failed to exhibit any linkage with Anger and also correlated positively with a Competition variable that had displayed connections with greater Anger and lower Self-Esteem.After controlling for Maladaptive Narcissism along with gender, Adaptive Narcissism did correlate negatively with Anger and also with Physical Appearance.In addition, the conceptually problematic linkage with Competition became non-significant.As expected, therefore, Adaptive Narcissism became more unambiguously adaptive when procedures controlled for Maladaptive Narcissism as well as for gender.
As Table 3 also reveals, Maladaptive Narcissism became more clearly maladaptive in partial correlations controlling for both Adaptive Narcissism and gender.After controlling for only gender, Maladaptive Narcissism predicted higher levels of Anger, Physical Appearance, and Competition and also lower levels of Virtue.However, no linkage appeared with Self-Esteem.After controlling for Adaptive Narcissism as well as for gender, these significant associations remained essentially unchanged, and Maladaptive Narcissism correlated negatively with Self-Esteem and God's Love and positively with Approval of Others.
These results demonstrated that in addition to controlling for gender, mediational analyses needed to examine Adaptive Narcissism after controlling for Maladaptive Narcissism and vice versa as independent variables in models predicting the dependent variable of Anger.Identification of potential mediators followed the recommendations of Baron and Kenny (1986) in examining only those potential mediators that displayed a significant association with the independent variable.After controlling for gender and Maladaptive Narcissism, Adaptive Narcissism exhibited significant associations with Self-Esteem (β = .56),Physical Appearance (β = -.11),Approval of Others (β = -.28), and God's Love (β = .19,ps < .05).Analyses, therefore, included these four variables as possible mediators of the Adaptive Narcissism relationship with lower Anger.In these procedures, significance tests of indirect effects used the bootstrap generated confidence interval (CI) estimations recommended by Hayes (2012) and maintained the conventional focus on unstandardized regression coefficients.In regressions controlling for Maladaptive Narcissism and gender, Adaptive Narcissism exhibited a negative unstandardized regression coefficient in its association with Anger, B = -.53,p < .001.After including Self-Esteem, Physical Appearance, Approval of Others, and God's Love as possible mediators, the direct effect of Adaptive Narcissism on Anger shifted in the opposite direction and proved to be significant, B = .33,p <. 05.As Table 4 makes clear, a significant total indirect effect appeared with all four mediators combined.Specifically, Self-Esteem, Physical Appearance, and Approval of Others proved to be significant mediators, but the indirect effect of God's Love was non-significant.For the three significant mediators, an examination of the indirect effect paths revealed that Adaptive Narcissism (1.) increased Self-Esteem which in turn decreased Anger, (2.) decreased Physical Appearance which in turn increased Anger, and (3.) decreased Approval of Others which in turn decreased Anger.
Noteworthy was the unexpected finding that Approval of Others predicted lower rather than the greater Anger when examined within the context of these other mediators.
In regressions controlling for Adaptive Narcissism and gender, Maladaptive Narcissism exhibited a positive association with Anger, B = 1.30, p < .001.After including Self-Esteem, Physical Appearance, Approval of Others, Competition, Virtue, and God's Love as mediators, the direct effect of Maladaptive Narcissism on Anger remained significant, B = .73,p < .001.A significant total indirect effect appeared with all six mediators combined (see Table 4).Specifically, Self-Esteem, Physical Appearance, Approval of Others, and Virtue were significant mediators, but the indirect effects of God's Love and Competition were non-significant.Apparent in these indirect effect paths were findings that Maladaptive Narcissism (1,) decreased Self-Esteem and Virtue which in turn decreased Anger, (2.) increased Physical Appearance which in turn increased Anger, and (3.) increased Approval of Others which in turn decreased Anger, Once again, Approval of Others displayed an unexpected negative connection with Anger when examined within the context of other mediators.

Discussion
Maladaptive Narcissism predicted higher Anger and lower Self Esteem, whereas Adaptive Narcissism displayed linkages with lower Anger and higher Self-Esteem.Such outcomes conformed to claims that healthy self-esteem develops out a maturation of narcissism (Watson, 2005).Kohut (1977) more specifically argued that this maturation proceeds through an internalization of the loving care received from selfobjects, a process which gradually creates structures within the self that enable an individual to maintain self-esteem without social support.Defined by an absence of internalized structure, immature narcissism can lead to anger and sometimes aggression in response to frustrations associated with the loss of social support.In contrast, more mature narcissism emerges from an internalization of structure that makes it possible for the individual to weather the loss of social support without anger.In this investigation, Self-Esteem and Contingencies of Self-Worth scales served as presumed indices of structure, and mediational results confirmed that these measures exerted important and largely expected influences on the associations of Adaptive and Maladaptive Narcissism with Anger.

Narcissism and Anger
Most straightforward were findings for Maladaptive Narcissism.Self-Esteem, Physical Appearance, Approval of Others, and Virtue partially mediated the connection of Maladaptive Narcissism with greater Anger.As expected, indirect effect paths revealed that Self-Esteem and Virtue did appear to operate as indices of structure in that they predicted both lower Maladaptive Narcissism and lower Anger, whereas Physical Appearance did seem to record an absence of structure because it exhibited an opposite pattern of effects.The expectation was that Approval of Others would display results parallel to those of Physical Appearance, but complexities appeared.The indirect effect path for this measure exhibited the expected positive connection with Maladaptive Narcissism, but a negative rather that positive linkage appeared with Anger.This unexpected outcome obviously requires clarification, but these data most importantly confirmed that a lack of structure at least partially mediated the connection of Maladaptive Narcissism with greater Anger.
Especially noteworthy were mediation results for Adaptive Narcissism.A negative association of Adaptive Narcissism with Anger became positive when Self-Esteem, Physical Appearance, and Approval of Others served as mediators.
Indirect effect paths reconfirmed Self-Esteem as an index of structure because it predicted greater Adaptive Narcissism and lower Anger, and Physical Appearance again seemed to operationalize the absence of structure because it exhibited an opposite pattern.Unanticipated results emerged once again for Approval of Others.As hypothesized, Approval of Others was like Physical Appearance in displaying a negative indirect effect path with Adaptive Narcissism; but with Anger, the path once more proved to be negative rather than positive.Most importantly, however, these results revealed that Self-Esteem, Physical Appearance, and Approval of Others not only fully mediated the Adaptive Narcissism relationship with lower Anger, but also suppressed a positive connection with Anger that would have been evident without the influence of structure.Such a suppression effect seemed consistent with previous observations that Self-Esteem may help explain the positive mental health implications of Adaptive Narcissism (Watson et al., 1999) and also with Kohut's (1977) understanding that self-esteem matures out of narcissism with the internalization of structure.
All significant partial correlations for the Contingencies of Self-Worth measures conformed to expectations.Presumed indices of the absence of internalized structure, including Approval of Others, Physical Appearance, and Competition, predicted lower Self-Esteem; although, the hypothesized linkage with greater Anger was non-significant for Approval of Others.All three measures displayed positive ties with Maladaptive Narcissism, and Approval of Others and Physical Appearance correlated negatively with Adaptive Narcissism.God's Love, Virtue, and Family Support served as presumed measures of internalized structure, and connections with Self-Esteem and Anger that were positive and negative, respectively, confirmed that presumption.God's Love and Virtue also predicted lower Maladaptive Narcissism, and a positive connection appeared between God's Love and Adaptive Narcissism.Only Academic Competence seemed irrelevant to the prediction of self-esteem, anger, and narcissism; and this lack of associations seemed generally in line with ambiguous outcomes observed previously for this measure (Crocker et al., 2003).Except for a few findings for God's Love and Virtue, all Contingency of Self-Worth measures displayed positive partial correlations with each other, and these results pointed toward a psychological commonality that these measures seek to record.In short, partial correlations offered broad support for the validity of the Contingencies of Self-Worth Scale.
For the Approval of Others measure, therefore, unexpected effects in the mediation analyses were specific to procedures examining this construct within the context of other mediators.Self-Esteem and Physical Appearance were the only other significant mediators in both analyses; and so, these two measures presumably were critical in producing the counterintuitive negative indirect effect paths between Approval of Others and Anger.Perhaps these unexpected outcomes revealed that this correlate of immature narcissism had noteworthy positive potentials.Such a possibility may receive support within Kohut's (1977) theory.Early in life, the seeking of approval from others would be an adaptive process when those others served as loving and supportive selfobjects who optimally frustrated that need and thus promoted its internalization as self-approval.In therapy for narcissistic disorders, the seeking of approval could have beneficial effects as well.The Kohutian therapist essentially operates as an empathic and supportive selfobject who, among other things, attempts to meet but also to optimally frustrate the client's need for approval.Once again, an internalization of self-approval would be the hoped for result.In these developmental and therapeutic scenarios, therefore, the need for approval would have beneficial potentials, and the unexpected Approval of Others mediation findings perhaps pointed in that direction.
As in earlier projects, Adaptive Narcissism emerged as a more unambiguous measure after controlling for Maladaptive Narcissism and vice versa, and this pattern supported attempts to conceptualize narcissistic functioning along a continuum of self-esteem regulation (Watson, 2005; also see Trumpeter, Watson, O'Leary, & Weathington, 2008).Various interpretative frameworks exist for conceptualizing narcissism, and one previous study differentiated between grandiose and vulnerable narcissism and examined their relationships with Contingencies of Self-Worth (Zeigler-Hill, Clark, & Pickard, 2008).Correlations identified vulnerable narcissism as clearly maladjusted, but grandiose narcissism as operationalized by the NPI yielded mixed implications.Grandiose and vulnerable narcissism also correlated positively.Findings of the present investigation first suggest that more unambiguous outcomes might have appeared if grandiose narcissism had been reexamined after partialing out vulnerable narcissism (and vice versa).More basically, the present study also reveals that at least some features of grandiose narcissism as recorded by the NPI may betray a lack of structure and display their own forms of vulnerability, a possibility in fact implied as a footnote consideration in this previous examination of narcissism and Contingencies of Self-Worth p. 765).The broader suggestion, therefore, is that narcissism measures might be better conceptualized along a continuum of self-esteem regulation rather than as constructs that can be dichotomized.
Gender differences appeared.Most important was the observation that men scored higher than women on both Adaptive and Maladaptive Narcissism.Such outcomes seemed consistent with previous demonstrations that gender development and sex role orientations make sense within Kohut's (1977) understanding of self-maturation (Sawrie, Watson, & Biderman, 1991;Watson, Biderman, & Boyd, 1989;Watson, Biderman, & Sawrie, 1994).
As with any project, limitations necessitate interpretative caution.Especially noteworthy is the realization that "structure" was a metaphor based on theory and that Self-Esteem and Contingency of Self-Worth scales served as hypothetical, indirect indices of that metaphor.Better measures of structure may be available, and even more basically, other more fruitful theoretical approaches may exist for explaining narcissism and self-esteem regulation.
In additional, all data were essentially correlational.Strong assumptions about causality, consequently, remain unwarranted.Even if analyses of mediation involved examination of causal models, procedures did not manipulate the independent variables or the mediators; so, observed outcomes could have reflected, at least in part, the covariance of other unexamined influences.This study examined college undergraduates.Samples examining participants of other ages or different backgrounds might yield important contrasts with the present results.Finally, E/E served as the index of Maladaptive Narcissism and displayed a relatively low internal reliability.A more psychometrically adequate operationalization of Maladaptive Narcissism could uncover even more robust outcomes.
In conclusion, this study found that "structure" presumably related to self-esteem regulation did as hypothesized mediate the linkages of Adaptive and Maladaptive Narcissism with Anger.Complexities associated with Approval of Others suggested that especially this variable seems to deserve additional research attention.For Maladaptive Narcissism, only a partial mediation effect appeared, and future studies might usefully explore what other factors might be necessary to produce a full mediation effect.Also of interest would be efforts to examine whether operationalizations of structure might mediate Adaptive and Maladaptive Narcissism relationships with other aspects of personality functioning, such as empathy, shame, and perfectionism (Watson et al., 1994(Watson et al., , 1996(Watson et al., , 1999)).At the broadest and most important level, however, the present data suggested that the conceptual framework of Kohut (1977) may supply useful guidance in future research attempts to clarify the dynamics of narcissism and self-esteem.

Notes
i) Data from this project appeared in a thesis submitted by the first author in partial fulfillment of the M.S. degree in Psychology at the University of Tennessee at Chattanooga.Grisel Garcia is currently pursuing a Ph.D. degree at the University of Rhode Island.
describe as "external contingencies of self-worth."Family Support makes reference to self-worth based upon compliance with what the family wants a person to do, and Academic Competence records self-satisfaction based upon abilities to meet personal goals for scholastic achievement.Though designated as "external," these two contingencies appear to be somewhat ambiguous relative to the concept of "structure."Crocker et al. suggested that Family Support is relevant to loving family relationships that promote healthy self-esteem and hence might reflect experience with Kohut's soothing and supportive selfobjects.Indeed, Crocker et al. found that Family Support predicted greater self-esteem, extraversion, agreeableness, conscientiousness, and openness to experience.Academic Competence also suggests an internalization of standards that might encourage academic achievement.While Crocker et al. observed positive correlations with agreeableness and conscientiousness, they nevertheless found that Academic Competence displayed no linkage with self-esteem and predicted greater neuroticism.This project used an empirical criterion for determining Narcissism and Anger whether these variables reflected structure.If a measure correlated positively with self-esteem, then it assessed "structure."Three other external contingencies of self-worth were clearly relevant to maladaptive narcissism.With Physical Appearance, positive self-assessment depends upon how attractive the individual feels in the eyes of others.
Interpersona 2015, Vol.9(1), 59-71 doi:10.5964/ijpr.v9i1.163 conceptualization of "structure" would help explain relationships of narcissism with anger.Support for this model would appear with the confirmation of five most important sets of hypotheses.First, maladaptive narcissism should predict greater and adaptive narcissism should predict lower levels of anger.

Table 2
Partial Correlations Among Anger, Self-Esteem, and Contingencies of Self-Worth Controlling for Gender

Table 4
Indirect Effects in the Prediction of Anger by Adaptive and Maladaptive Narcissism