Self-concept Clarity's Role in Self–Other Agreement and the Accuracy of Behavioral Prediction

Abstract

The current set of studies investigated whether undergraduates with greater self-concept clarity also demonstrated greater accuracy based on self–other agreement and behavioral prediction. Study 1 demonstrated a positive association between self-concept clarity and target–informant agreement for a set of observable traits. Study 2 explored the association of self-concept clarity with target–informant agreement for the Big Five personality traits, a set of behaviors related to personality, and a set of everyday activities. Study 3 related self-concept clarity to a behavioral measure of accuracy by comparing participants' predicted and subsequent observed behavior on a laboratory task. As hypothesized, those with greater self-concept clarity showed greater self–other agreement on personality traits, behavioral descriptions of personality, and accuracy of predicted behavior.

Keywords:

• Self-concept
• Personality
• Interpersonal understanding
• Behavior
• Accuracy

Acknowledgments

Preparation of this article was supported in part by a grant from Anthony Marchionne Foundation Small Grants Program to the first author. Study 3 was funded by a grant from Psi Chi to the second author.

In addition to several anonymous reviewers, we would to thank the following individuals for their help on this project: Rebecca Hampson, Leigh-Ann Javas, and Alanna Raines.

Notes

1. We also assessed self–other agreement by taking the absolute value of the difference between the informant's ratings and the target's ratings and computing the mean, and with squared absolute values of the difference. In each case, analyses with these scores paralleled the self–other correlations. Difference scores analysis demonstrate a direction of difference between self and other, while correlations assess similarity between judgments of self and other. A great deal of prior research has made use of correlations when determining self–other agreement (see Colvin & Funder, 1991; Vogt & Colvin, 2005, for reviews). Thus, for parsimony, we only report analyses that used the self–other correlation.

2. Unless noted otherwise, we reported all results using one-tailed tests due to the directional nature of the predictions: Greater self-concept clarity was always predicted to positively correlate with more accuracy. Note, however, that even had we used the more conservative two-tailed tests, the only results that would have failed to reach significance are those in Study 2 for extraversion and conscientiousness. In those cases, the two-tailed tests would have reached the p < .10 level and were in the predicted direction.

3. We tested gender as a potential moderator in each of the reported regressions. In each analysis gender was not a significant moderator of the reported effect.

4. Because informants in Studies 1 and 2 were recruited based on their status as “the person who knows you [the target] best at school,” the degree of acquaintance ratings have a restricted range: None of the informants were strangers to the target. In Study 1, on a 7-point Likert scale scores ranged from 3.5 to 7, while in Study 2 scores ranged from 3 to 7. As a result, analyses controlling for acquaintanceship only account for variations in informants who already knew the target. If there were a greater range of informants (e.g., strangers as well as best friends), it is possible that degree of acquaintance could have moderated the association between self-concept clarity and self–other agreement.

5. We also analyzed the data with a regression predicting the actual score from the predicted score, self-concept clarity, and the interaction between the two. Consistent with our hypothesis, the interaction (predicted effect) was significant, β = 1.77, t(68) = 2.78, p = .007, sr ² = .08.