Brief report
Predictive validity of explicit and implicit self-esteem for subjective well-being

https://doi.org/10.1016/S0092-6566(02)00532-9 Get rights and content

Abstract

In recent years, researchers have developed a variety of techniques to measure implicit self-esteem. Bosson, Swann, and Pennebaker (2000) examined the reliability and validity of these measures. Only some implicit measures were reliable, and even these measures failed to show convergent and predictive validity. In contrast, explicit self-esteem predicted subjective well-being (SWB). However, the predictive validity of explicit self-esteem measures may have been inflated because SWB was assessed by means of self-reports. The present article addresses this concern. We correlated self-reports and informant reports of subjective well-being with an explicit (Rosenberg’s self-esteem scale) and an implicit measure of self-esteem (preferences for initials). Explicit self-esteem was a significant predictor of all SWB measures. Preferences for initials were not significantly correlated with any of the SWB measures.

Introduction

Most variables in personality psychology are assessed with self-report measures because (a) they are easier to administer than alternative measures (e.g., assessment of response latencies, physiological indicators), and they have proven to be reliable and valid in numerous studies. For example, self-report measures show convergent validity with informant reports (Costa & McCrae, 1988; Diener, Smith, & Fujita, 1995; Funder, 1995), and they predict important outcomes such as longevity and marital satisfaction (Friedman et al., 1995; Watson, Hubbard, & Wiese, 2000). Subsequently, we will refer to self-report measures as explicit measures.

Explicit measures can be contrasted with implicit measures of personality traits. Implicit measures of personality have proven useful since the beginning of personality psychology (Murray, 1938; Wilson, Lindsey, & Schooler, 2000). They do not rely on individuals’ conscious representations of their selves to assess personality. Hence, they promise to overcome the limitation of explicit measures that participants have to be both willing and able to report their personality. Given the promising qualities of implicit measures of personality, it is important to examine the utility of these measures for personality assessment by examining their reliability and validity.

Bosson, Swann, and Pennebaker (2000) examined the retest-reliability of seven implicit measures of self-esteem. The results varied greatly across the different measures. Four measures showed unsatisfactory retest-correlations (rs<.30), which compromises their utility as personality measures. However, three measures of implicit self-esteem showed tolerable reliabilities, namely the Implicit Association Test (IAT, .69), Preferences for Name Initials (.63), and Preferences for Birth Dates (.53). Similar results have been obtained in other studies (Greenwald & Farnham, 2000; Koole, Dijksterhuis, & van Knippenberg, 2001).

Bosson et al. (2000) also examined the convergent validity of different implicit self-esteem measures. As expected, unreliable measures failed to show convergent validity with reliable measures. However, even reliable implicit measures failed to show convergent validity with each other (rs=−.11 to .23). Bosson et al. (2000) also examined the predictive validity of explicit and implicit self-esteem measures (see also Greenwald & Farnham, 2000). It is well known that explicit self-esteem is correlated with subjective well-being (high life-satisfaction, high positive affect, low negative affect), especially in individualistic cultures (Diener & Diener, 1995). Bosson et al. (2000) replicated this finding. Explicit self-esteem (e.g., Rosenberg’s self-esteem scale) predicted individual differences in positive and negative affect. Only one of the implicit measures (i.e., preferences for initials) was a significant predictor of positive affect, and none of the implicit measures predicted negative affect.

A minor caveat of Bosson et al.’s (2000) study was the reliance on self-report measures for the assessment of subjective well-being. As a result, the criterion variables may have shared method variance with the explicit self-esteem measure. For example, social desirable responding could have inflated scores on the explicit self-esteem measure and positive affect ratings. This shared bias could inflate the correlation between explicit self-esteem and positive affect. More importantly, social desirability could also attenuate the relation between implicit self-esteem and subjective well-being. The aim of this study was to address this concern. For this purpose, we compared the predictive validity of explicit and implicit self-esteem measures for informant reports of subjective well-being. We used preferences for initials as a measure of implicit self-esteem, mainly because it was feasible to use this paper–pencil method in a large sample. In addition, preferences for initials were the only implicit measure that was significantly correlated with well-being measures in Bosson et al.’s (2000) study.

Section snippets

Participants

One hundred and forty-one students at the University of Illinois, Urbana-Champaign participated in the study as part of a course on personality and well-being.

Explicit self-esteem

Explicit well-being was assessed at the beginning of the semester with Rosenberg’s self-esteem scale (Rosenberg, 1965).

Implicit self-esteem

Preferences for initials were also assessed at the beginning of the semester, but in a different session than the assessment of explicit self-esteem to avoid shared influences due to mood effects or other state

Preferences for initials

We followed Bosson et al.’s (2000) procedure to derived Preferences for Initials from the letter ratings. That is, we subtracted the overall liking of a letter averaged across all participants from the rating of initials. That is, if “B” was rated on average a 4 and Bob rated “B” a five, Bob’s liking of his first name initial would be 1 (i.e., 5−4). This procedure controls for differences in the liking of letters in general. For example, we found that participants on average liked “A” much more

Discussion

The present study compared the validity of an explicit measure of self-esteem with the validity of an implicit measure of self-esteem. Bosson et al. (2000) found that explicit self-esteem was a better predictor of self-reported well-being than preferences for initials. The present study replicated this finding. Furthermore, we extended the finding to informant reports of well-being. Informant reports of well-being were related to explicit self-esteem but unrelated to preferences for initials.

Acknowledgements

We would like to thank Romin Tafarodi and Shigehiro Oishi for helpful comments on a previous draft of this article.

References (15)

  • J.K. Bosson et al.

    Stalking the perfect measure of implicit self-esteem: The blind men and the elephant revisited

    Journal of Personality and Social Psychology

    (2000)
  • P.T. Costa et al.

    Personality in adulthood: A six-year longitudinal study of self-reports and spouse ratings on the NEO personality inventory

    Journal of Personality and Social Psychology

    (1988)
  • E. Diener et al.

    Cross-cultural correlates of life satisfaction and self-esteem

    Journal of Personality and Social Psychology

    (1995)
  • E. Diener et al.

    The satisfaction with life scale

    Journal of Personality Assessment

    (1985)
  • E. Diener et al.

    The personality structure of affect

    Journal of Personality and Social Psychology

    (1995)
  • H.S. Friedman et al.

    Childhood conscientiousness and longevity: Health behaviors and cause of death

    Journal of Personality and Social Psychology

    (1995)
  • D.C. Funder

    On the accuracy of personality judgment: A realistic approach

    Psychological Review

    (1995)
There are more references available in the full text version of this article.

Cited by (0)

View full text