Although personality psychology has largely been built on findings based on self-report measures, descriptions of a target individual's personality from an observer (other-ratings) have bolstered and extended major findings from personality research. The veracity of these other-ratings, however, has been a longstanding research question and point of contention. Although work using other-ratings has spanned across multiple research streams, key issues are (1) how accurate other-ratings are (as indexed by a particular criterion) and (2) the conditions under which other-ratings are more or less accurate (as a result, for example, of a particular trait being easier to rate or of having higher quality information about the target). Four meta-analyses were conducted to empirically examine hypotheses derived from Funder's (1995) Realistic Accuracy Model, with each meta-analyses examining a particular accuracy criterion. Specifically, these meta-analyses examined how the ease of rating particular traits within the five factor model moderates accuracy ("good traits"). In addition, these meta-analyses examine how the quality of information available to others moderates accuracy ("good information"), either through having a particular type of relationship with the target (being a family member, friend, cohabitator, work colleague, incidental acquaintance, or stranger) or through observing particular target-related stimuli (visual information only, audio information only, audio and visual, personal objects, or text/electronic communication).

Study 1 meta-analyzed internal consistency and test-retest reliability coefficients for other raters. Findings indicate that other-ratings are measured at least as reliably as self-ratings, regardless of the particular trait being measured or of the particular information source. Study 2 meta-analyzed the consistency between two different other-raters' descriptions of a common target (interrater reliability). Interrater reliability was higher when rating more visible traits like Extraversion and Conscientiousness but lower for less visible traits like Emotional Stability and Openness. Interrater reliabilities were also higher when either (a) ratings came from others well-acquainted with the target or (b) both other-raters viewed a limited but overlapping set of the target's behaviors. Study 3 meta-analyzed correlations of other-ratings with self-ratings. Self-other correlations were highest when (a) the trait was highly visible, (b) opportunity to observe the target was high and not limited to only observing thin slices of behavior, and (c) the rating came from an other who was particularly intimate with the target. Finally, Study 4 meta-analyzed validities of other-ratings for predicting behaviors and behavioral outcomes. When predicting how strangers will perceive the target, ratings from acquainted other-raters predicted similarly to self-ratings. When predicting academic and job performance, however, validities for other-ratings were markedly higher than general findings for self-ratings. This was especially true for ratings of Conscientiousness and Emotional Stability when predicting academic achievement and for ratings of Conscientiousness and Agreeableness when predicting job performance.

The pattern of findings described in Studies 1 through 4 points to several general conclusions. Self-ratings and other-ratings both measure a common construct of personality dispositions. For other-ratings to be accurate, however, other-raters must have adequate opportunity to observe the target. This accuracy is enhanced when other-raters have access to internal aspects of the target's personality (thoughts, emotions, values, etc.) as a result of interpersonal intimacy. To the extent that such opportunities are lacking to observe both external and internal aspects of the target, other-raters may use implicit personality theories to fill in gaps in observation. Finally, Study 4 is suggestive of a relative predictive advantage of other-ratings over self-ratings. Whether resulting from greater accuracy of other-ratings or having a context and scope more closely aligned with the behavioral criterion, these behavioral predictions offer promising new directions for studying personality and applying personality measures. Broadly, these findings suggest that other-ratings can serve not only as an accurate, alternative method for collecting personality ratings but also as a tool illuminating fundamental aspects of personality and personality measurement.