Narcissistic Illusions in Self-Evaluations of Intelligence and Attractiveness
Total Page:16
File Type:pdf, Size:1020Kb
Narcissistic Illusions In Self-Evaluations of Intelligence and Attractiveness Marsha T. Gabriel, Joseph W. Crltelll, and Juliana S. Ee University of North Texas ABSTRACT In this study we compared the ability of narcissism and self- esteem to predict positive illusions in self-evaluations of intelligence and physi- cal attractiveness in a sample of 146 college students. Narcissism predicted both types of illusion for males and females; self-esteem predicted intelligence self-illusion for males. Both males and females overestimated their own intel- ligence, with males, but not females, also overestimating their attractiveness. Positive illusions for intelligence and attractiveness were correlated. Males showed greater positive illusions than females, with this effect at least partly attributable to observed gender differences in narcissism. Clinical theories (Kernberg, 1970; Kohut, 1971) argue that the purpose of narcissistic mental activity is to maintain a positive view of self. The clinical literature, based on unquantified observation, claims that nar- cissists distort reality to protect their fragile view of self. Narcissists purportedly see themselves as special and unique, possessing extraor- dinary abilities (Russell, 1985; Tobacyk & Mitchell, 1987). They exag- gerate their accomplishments, deny their faults, and adopt an attitude of superiority toward others (Lax, 1975). Thus narcissists are thought to possess a grandiose view of self that ranges across a number of life areas (Akhtar & Thomson, 1982). While an inflated, grandiose view of self is central to diagnoses of narcissistic pathology, this characteristic has not been verified empiri- cally. Previous research (Raskin, Novacek, & Hogan, 1991), however, has suggested a link between narcissism and measures relevant to gran- Send requests for reprints to Joseph W. Critelli, Department of Psychology, PO Box 13587, University of North Texas, Denton, TX 76203. Journal of Personality 62:1, March 1994. Copyright © 1994 by Duke University Press. CCC 0022-3506/94/51.50 144 Gabriel et al. diosity. These authors employed three such measures: the Grandiosity scale (Watson & Klett, 1972), consisting of Minnesota Multiphasic Personality Inventory (MMPI) items such as "I am a special agent of God"; composite observer ratings of the traits "condescending," "ego- tistical," and "conceited"; and the fit between self and ideal self, with personal adjustment partialled out. Although this research goes a long way toward establishing an empirical linkage between narcissism and grandiose self-perception, it does not compare subjects' views of self to an objective criterion to determine whether the self-perceptions of narcissists are, in fact, inflated or otherwise distorted. The present study extends this line of research by examining whether narcissistic individuals show positive illusions in self-perception as compared to an objective criterion. We selected two major life areas for exploring these effects: cognitive ability and physical attractiveness. In each domain, self-evaluations of those high in narcissism were com- pared to objective measures, providing an empirical test of distorted self-perception. For purposes of discriminant validation, we also con- trasted narcissism with self-esteem. Both narcissism and self-esteem share the characteristic of positive self-evaluation, and self-esteem has been related to self-enhancing illusions (Raskin et al., 1991; Taylor & Brown, 1988). Nevertheless, both theoretical and empirical descrip- tions of narcissism indicate that unrealistic self-perception should be a more central feature of narcissism than of high self-esteem (Buss & Chiodo, 1991). Subjects Participants were 146 volunteers (84 females, 62 males) who received course credit in utidergraduate psychology classes. Ages ranged from 17 to 45 years, with a mean of 21.6 {SD = 0.4). Racial composition of the sample was 82% Caucasian, 9% African-American, 6% Hispanic, and 3% other. Measures Narcissism was measured with the Narcissistic Personality Inventory (NPI; Raskin & Terry, 1988). This scale shows good internal consistency (alpha = .86; Auerbach, 1984) and appropriate correlations with other measures of nar- cissism (Prifitera & Ryan, 1984). The NPI has showti positive correlations with dominance and exhibition (Emmons, 1984) and negative correlations with Narcissistic Illusions 145 social interest (Jourbert, 1986). The NPI has also been used successfully to diagnose narcissism in clinical samples (Prifitera & Ryan, 1984). Self-esteem was measured by the Tennessee Self-Concept Scale Total P score (TSCS; Roid & Fitts, 1988). The TSCS is a widely used measure of self- esteem (Stanwyck & Garrison, 1982) with good test-retest reliability (r = .92, two-week interval) atid intemal consistency (r = .91; Roid & Fitts, 1988). It has been used to discriminate psychiatric groups from normals (Crites, 1965) and to predict treatment outcome in substance abuse patients (O'Leary, Chaney, & Hudgins, 1978). A global, rather than domain-specific, measure of self-esteem was selected to make this variable comparable in generality to the NPI. Intelligence was measured by the Shipley Institute of Living Scale (Zachary, 1986). This scale contains two subtests. Vocabulary and Abstraction, with a combined split-half reliability of .92, and a test-retest reliability over 12 weeks of .78. The test has been correlated .74 with the WAIS-R (Zachary & Gorsuch, 1985) and is considered a valid estimation of general intelligence (Zachary, Crumpton, & Spiegel, 1985). Scores on the Shipley Scale were transformed into WAIS-R equivalents according to the Shipley manual (Zachary, 1986) and then transformed into percentile scores for college students according to norms for college students from the WAIS-R standardization sample (Rey- nolds, Chastain, Kaufman, & McLean, 1987). Self-rated intelligence was ob- tained by having subjects rate their own intelligence on a percentile scale as compared to the average college student. Intelligence self-illusion was defined as self-rated intelligence independent of actual intelligence, i.e., the residual after actual intelligence was statistically removed from self-rated intelligence. Physical attractiveness was measured by comparing photographs of sub- jects against a set of selected yearbook photos. Three hundred photos of each sex were randomly selected from the university yearbook. Five male and five female undergraduates independently sorted photos into categories approxi- mating a normal distribution of attractiveness for each gender. Raters were given seven categories ranging from "extremely low attractiveness" to "ex- tremely high attractiveness," along with the number of pictures to be sorted into each category to approximate a normal distribution (Geiselman, Haight, & Kimata, 1984). Fifty pictures of each gender were selected based on main- taining a normal distribution and on consistency of ratings across raters. Separate sets of male and female pictures were randomly ordered on charts to be used as comparisons for rating pictures of subjects during the experi- ment. Physical attractiveness scores were obtained by comparing the subject's picture with each standard picture of the same sex (Mueser, Graw, Sussman, & Rosen, 1984). This type of direct picture-to-picture comparison provides a relatively objective basis for quantifying physical attractiveness. A subject's percentile attractiveness score was twice the average of the number of pic- tures for which the subject was rated "more attractive" by the two independent 146 Gabriel et al. raters. The two female experimenters, who were blind as to subjects' levels of narcissism and self-esteem, served as raters after having achieved acceptable reliability with pilot subjects (r = .80). Self-rated attractiveness was obtained by having subjects rate their own attractiveness on a percentile scale as com- pared to the average college student. Attractiveness self-illusion was measured by the residual remaining when actual attractiveness was statistically removed from self-rated attractiveness. Procedure Subjects completed demographic items and rated their own intelligence and attractiveness. This material was returned to the experimenters in a folder to prevent them from inadvertently seeing subjects' self-ratings. Shoulder-level pictures were taken of each subject against a standardized background with an automatically developing camera. Subjects were then given the intelligence measure. Following this, the NPI and TSCS were given in random order. While these measures were being completed, each of the two female experiment- ers independently rated pictures of the subject in a separate room. Details of this procedure can be found in Gabriel (1990). All procedures complied with ethical standards of the American Psychological Association. RESULTS Coefficient alphas for narcissism and self-esteem were .86 and .90, re- spectively. Attractiveness ratings reached acceptable levels of interrater reliabiUty (intraclass r = .58, p < .001; Bartko, 1976). Means and standard deviations for the major variables appear in Table 1. For males, self-rated intelligence was greater than actual intelligence, t{6l) for paired samples = 5.03, p < .01, and self-rated attractiveness was greater than actual attractiveness, r(61) for paired samples == 5.57, p < .01. For females, self-rated intelligence was greater than actual intelligence, r(83) for paired samples = 6.97, p < .01, but self-rated and actual attractiveness did not differ, f(83) for