Self–Other Agreement in Personality Reports: A Meta-Analytic Comparison of Self- and Informant-Report Means

Abstract

Self-report questionnaires are the most commonly used personality assessment despite longstanding concerns that self-report responses may be distorted by self-protecting motives and response biases. In a large-scale meta-analysis (N = 33,033; k = 152 samples), we compared the means of self- and informant reports of the same target’s Big Five personality traits to examine the discrepancies in two rating sources and whether people see themselves more positively than they are seen by others. Inconsistent with a general self-enhancement effect, results showed that self-report means generally did not differ from informant-report means (average δ = −.038). Moderate mean differences were found only when we compared self-reports with stranger reports, suggesting that people are critical of unacquainted targets. We discuss implications of these findings for personality assessment and other fields in which self-enhancement motives are relevant.

Keywords

accuracy Big Five meta-analysis self-enhancement self-report informant report open data open materials

Person perception is a fundamental aspect of our daily lives. As social beings, we generally try to understand our own personality traits and the traits of other people. Not only do we strive to make sense and predict the behavior of people with whom we interact in daily life, but we also speculate about the traits of total strangers. Apart from being a compelling part of our everyday experience, self- and other perceptions of personality are crucial to understanding how people function, and these perceptions help us to better navigate our social worlds.

However, the accuracy of personality self-perceptions (and, relatedly, self-reports that the field of personality has heavily relied on) has often been questioned in the literature. These criticisms have highlighted that motives toward social desirability may result in (a) inaccurate self-assessments of personality (Alicke, Klotz, Breitenbecher, Yurak, & Vredenburg, 1995; Brown, 1986; Kunda, 1990) and (b) inaccurate self-reporting by respondents (Edwards, 1957; Paulhus, 1991). Specifically, self-reports may be distorted by ego-protective biases, such as the better-than-average or self-enhancement effect (Alicke, 1985; Sedikides & Gregg, 2007; Taylor & Brown, 1988). This better-than-average effect particularly influences people’s self-reports of personality characteristics and skills that are desirable, evaluative, positively valenced, controllable, and personally important (Dunning, Griffin, Milojkovic, & Ross, 1990; John & Robins, 1993).

In light of these criticisms, subsequent research has evaluated whether self-reports of personality are useful predictors (Cronbach, 1955; Funder, 1987; John & Robins, 1993; Kenny & Albright, 1987; Vazire, 2010). Large-scale, meta-analytic evidence reveals that not only are personality self-reports strong correlates of major life outcomes (e.g., Roberts, Kuncel, Shiner, Caspi, & Goldberg, 2007), but they also converge well with personality reports from knowledgeable informants (e.g., Connelly & Ones, 2010).

However, finding that personality self-reports have strong predictive power does not necessarily imply that self-reports are not inflated or enhanced. Indeed, from as early as Cronbach’s (1955) studies, researchers have distinguished elevation accuracy (the tendency for individuals’ ratings to not be higher or lower than an accuracy criterion) from rank-order accuracy (consistency between the rank ordering of self-reports and an accuracy criterion). Therefore, it is possible that personality self-reports could show both elevation inaccuracy (e.g., inflated self-perceptions) and strong correlations with criteria (as are present in research using personality self-reports to predict peer reports or life outcomes). Indeed, research has found that personality self-reports that have been manipulated to be inflated (through either simulations or instructions to present oneself favorably) can still produce strong correlations with criteria (Komar, Brown, Komar, & Robie, 2008; Mueller-Hanson, Heggestad, & Thornton, 2003). There have long been large-scale examinations finding low elevation accuracy of self-perceptions (higher self-report means) in other domains (e.g., abilities, Mabe & West, 1982; job performance, Harris & Schaubroeck, 1988). However, there is no available large-scale analysis of whether personality self-reports show elevation accuracy. Thus, an important unanswered question is, “Just how inflated are personality self-reports?”

The Present Research

The purpose of the present research was to use meta-analytic integration to evaluate the elevation accuracy of personality self-reports. Elevation accuracy must be evaluated against a specific accuracy criterion (“Are self-reports of personality more elevated than ____?”). In the present study, we drew data from existing studies to compare means of self-reports of personality with means of informant reports among studies using the same measure to assess the same sample of target individuals. Informant reports of personality have long been used as a valuable criterion for evaluating the accuracy of self-reports (Funder & West, 1993). This representation of elevation accuracy as the difference between self- and informant reports generally corresponds to Kwan, John, Kenny, Bond, and Robins’s (2004) self-insight index of self-enhancement. Finding differences in these means suggests a general trend of systematic errors that inflate or deflate at least one rating source. Given the tendencies for people to self-enhance and given the previous findings on conditions of self–other agreement, we proposed six hypotheses.

Hypothesis 1a was that, on average, mean personality reports were likely to be more positive for the self than the informant. Previous research has highlighted that accuracy varies on the basis of characteristics of the traits (evaluativeness and visibility) and the raters (specifically, the raters’ acquaintance with the target; Connelly & Ones, 2010; Funder, 1995; John & Robins, 1993). First, accuracy is suggested to be highest for traits that are less evaluative and highly visible. Making judgments of the self on traits that are more evaluative is more ego involving because they are directly related to one’s self-worth, whereas judging other people is less ego involving. Second, raters are more likely to provide accurate assessments on visible traits because raters will have a high quantity and quality of trait-relevant observations (Funder, 1995).

Hypothesis 1b was that mean differences were likely to be larger for more evaluative than less evaluative traits. Hypothesis 1c was that mean differences were likely to be larger for less visible than more visible traits. We expected that our indices of elevation accuracy would differ depending on the relationship that the informant has with the target. Specifically, raters who have the most intimate relationships with targets are likely to have had discussions with targets about their inner thoughts and are more likely to validate targets’ self-perceptions (Swann, 1983). In addition, friendship biases (i.e., having favorable views of friends) could potentially mimic self-enhancement effects in the aggregate (Wood, Harms, & Vazire, 2010). Informants who interact with targets frequently but do not necessarily have intimate relationships are more likely to observe situation-specific trait information, although they are less likely to show friendship biases (e.g., work colleagues; Kenny, 2004). Thus, we proposed Hypothesis 1d, which was that elevation accuracy would be highest among studies using family or friends as informants, followed by work colleagues, followed by strangers. Additionally, it is possible that self- and informant reports would show differences in variability. Such variability differences reflect (a) the extent to which self-ratings may be more extreme than informant ratings and (b) whether a particular type of rater makes a greater differentiation among targets. For instance, studies on personality faking that have directly instructed respondents to portray a positive impression in their responses have tended to find less variability in personality scores (likely because respondents are encountering ceiling effects in endorsing positive items; Hooper, 2007). In contrast, finding greater variability in self-reports could suggest that informants avoid extreme positive and negative item endorsements. Accordingly, we anticipated that self-enhancement would also reduce variability.

Hypothesis 2a was that on average, variability would be lower for self-reports than informant reports. Hypothesis 2b was that self-report variability would be lower than informant-report variability for more evaluative than less evaluative traits. It remains somewhat unclear how visibility and types of informants would result in different self-informant variability ratios. On one hand, an individual might not self-enhance on traits, whether visible or invisible, that are not ego involving. On the other hand, informants (e.g., strangers) might show more variability on invisible traits because they may not have the same access to trait-relevant behaviors as the target (or closely acquainted informants). In addition, other-enhancement effects might reduce variability for intimately acquainted informants, but without intimacy, people may linger much more around the midpoint of the Likert-type scale.

Method

Meta-analytic database and coding

In the current meta-analysis, we compared two psychometric properties of self- and informant reports of the same target’s personality: standardized mean differences and variability differences of the Big Five personality traits and facets. We compiled studies that used the same scale for self- and informant reports of personality by searching online databases, reviewing existing meta-analyses of informant reports of personality, and contacting researchers to request unpublished data. Our search yielded a total of 117 sources (87% published, 13% unpublished), from which we were able to extract data from a total of 152 independent samples (N = 33,033; see the Supplemental Material available online). We coded personality scales according to a working taxonomy of Big Five factors and facets and categorized informants according to the taxonomy in Connelly and Ones (2010). Reliabilities for our coding of all study characteristics were strong.

Meta-analytic methods

As noted above, we focused on two effect sizes: mean differences and variability differences. We examined standardized differences in means between self- and informant reports with the Cohen’s (1988) d statistic:

d = \frac{M_{self} - M_{informant}}{S D_{pooled}},

where M_self reflects the mean for self-reports on a given personality trait, M_informant reflects the mean for informant reports on the same trait (using the same scale), and SD_pooled reflects the standard deviation pooled across self- and informant reports. Thus, positive values of d indicate higher self-report scores on the trait. The variability-difference score (U ratio) was calculated as the average standard-deviation score for self-reports divided by the standard deviation score for informant reports (Schmidt & Hunter, 2015). Often, authors would report statistics for informants after averaging across multiple informants (e.g., mean and standard deviation of three friends’ reports of Agreeableness). Such averaging has no impact on the mean of the informant score, but it does systematically reduce the variance. Thus, we adjusted standard deviations of multi-informant averages to that of a single informant.

For mean differences, we followed Schmidt and Hunter’s (2015) random-effects model for conducting a psychometric meta-analysis of within-subjects d values and U ratios. Random-effects meta-analyses assume that the effect sizes in the studies may come from populations with varying population parameters (in accordance with our hypotheses), whereas a fixed-effects model assumes identical population parameters in the studies reviewed (Schmidt & Hunter, 2015). In addition, it is well documented that (a) imperfect reliability tends to underestimate effect sizes in studies (e.g., Nunnally, 1978) and (b) differences in the reliability of measures across studies introduce variability in effect sizes. As a result, meta-analyses tend to underestimate the average effect size and to overestimate true variability in results across studies. Hunter-Schmidt meta-analyses correct estimates of meta-analytic averages and variability for the disattenuating effects of unreliability. Other key estimates in the current meta-analytic output include the number of independent samples (k), the number of targets (N), the standard deviation of d (SD_d), the mean difference corrected for attenuation due to unreliability (δ), the standard deviation of delta (SD_δ), the ratio of the standard deviation of self- and informant reports corrected for the number of raters in informant reports (U), and finally, the standard deviation of U (SD_U). For the quantitative integration of effect sizes across different studies, we calculated the mean and variance across studies’ U ratio by weighting each study’s effect by its sample size (Schmidt & Hunter, 2015). For additional details on the Big Five taxonomy, collection methods, inclusion criteria, calculation methods, and meta-analytic results by types of personality measures, see the Supplemental Material.

Rating level of evaluativeness and visibility of the Big Five trait domains

To determine the evaluativeness and visibility of the Big Five domains and facets examined in our meta-analysis, we solicited ratings from experts in personality psychology (mostly professors, except for one postdoctoral researcher). Although most researchers agree on the structure of the Big Five, they are divided on the characteristics of the Big Five traits and facets. Nineteen individuals in total provided their responses online and separately rated the desirability and observability of each item on a 7-point Likert-type scale, ranging from 1 (very undesirable or very difficult to observe) to 7 (very desirable or very easy to observe; adapted from John & Robins, 1993; for the calculation of the evaluativeness and visibility scores and results, see the Supplemental Material). The anxiety and depression facets were rated to be highly undesirable, and the intellectual-efficiency, positive emotions, nonmanipulative, and dependability facets were rated as highly desirable. That is, these were rated to be highly evaluative. Furthermore, the depression, depth, and openness-to-sensations facets were rated as the least visible, and the activity, dominance, orderliness, and sociability facets the most visible. Of the traits, Agreeableness and Conscientiousness were rated as most desirable. Furthermore, Extraversion and Conscientiousness were rated as highly visible at the trait level, consistent with previous research and findings.

Results

Mean difference in effect sizes across the Big Five trait domains

We first conducted meta-analyses for each of the Big Five trait domains and facets with all categories of informant raters included (see Table 1). Contrary to Hypothesis 1a, results showed that the average meta-analytic effect across the broad groupings of the Big Five domains was nearly 0 (average δ = −.038), suggesting that there is not a general trend of systematic errors that inflate or deflate self- or informant reports. The only Big Five domain that produced even a modest self-enhancement effect was Openness/Intellect (δ = .247). Contrary to Hypotheses 1b and 1c, our results showed that there was no elevation in self-reports beyond informant reports, but the largest effect toward self-effacement was observed for Emotional Stability (reversed Neuroticism; δ = −.145), followed by Conscientiousness (δ = –.123). However, we noted that there was considerable heterogeneity in effect sizes (SD_δ) when clustered at the domain level, suggesting potential room for moderator effects. Thus, we separated out effect sizes in terms of (a) specific facets of Big Five traits (see Table 1) and (b) informants with particular types of relationships with targets (see Table 2).

Table 1.

Meta-Analyses of the Big Five Traits and Facets

			Mean difference				Variability ratio
Trait	k	N	d	SD_d	δ	SD _δ	U	SD_U
Emotional Stability	132	27,466	−0.125	0.231	−.145	.140	1.047	0.190
Global emotional stability	107	23,744	−0.159	0.247	−.183	.183	1.033	0.180
Lack of anxiety	26	6,269	−0.126	0.204	−.145	.105	1.011	0.178
Lack of uneven-temperedness	16	2,791	−0.069	0.223	−.080	.070	1.034	0.200
Lack of skepticism	2	1,488	0.003	0.014	.003	.000	0.894	0.027
Lack of depression	23	7,108	−0.156	0.189	−.181	.113	1.025	0.099
Self-esteem	10	2,832	−0.225	0.168	−.261	.000	1.169	0.195
Lack of negative affect	19	5,295	−0.100	0.263	−.116	.233	0.986	0.119
Extraversion	130	28,217	−0.083	0.214	−.096	.108	1.006	0.189
Global extraversion	110	23,476	−0.086	0.188	−.100	.000	1.011	0.194
Sensation seeking	2	1,059	0.050	0.161	.058	.121	1.040	0.000
Sociability	22	5,026	−0.114	0.181	−.132	.000	0.974	0.141
Positive emotions	24	6,944	−0.093	0.286	−.108	.270	1.001	0.120
Dominance	21	5,043	−0.234	0.215	−.272	.128	0.932	0.129
Activity	19	5,999	−0.187	0.135	−.217	.000	0.992	0.066
Openness/Intellect	113	25,406	0.206	0.274	.247	.236	1.022	0.184
Global openness/intellect	103	23,431	0.229	0.270	.275	.231	1.019	0.181
Curiosity	4	682	0.102	0.127	.122	.000	0.844	0.145
Depth	6	1,792	0.267	0.221	.320	.176	1.097	0.252
Aesthetics	14	4,965	0.101	0.127	.122	.000	0.999	0.113
Openness to sensations	3	1,156	0.194	0.086	.233	.000	1.077	0.041
Intellectual efficiency	18	4,899	−0.017	0.221	−.021	.167	0.986	0.149
Nontraditionalism	9	2,272	0.201	0.253	.241	.215	1.018	0.175
Agreeableness	113	25,017	−0.028	0.259	−.033	.210	0.927	0.196
Global agreeableness	111	23,529	−0.044	0.278	−.053	.237	0.921	0.203
Nurturance	8	2,924	−0.055	0.192	−.066	.145	0.891	0.109
Nonmanipulativeness	8	3,989	−0.067	0.154	−.080	.000	0.924	0.146
Cooperation	10	2,030	−0.137	0.146	−.164	.000	0.877	0.110
Modesty	10	2,951	−0.041	0.204	−.049	.144	0.887	0.153
Conscientiousness	120	25,480	−0.101	0.271	−.123	.227	0.940	0.178
Global conscientiousness	110	22,724	−0.098	0.272	−.119	.226	0.944	0.186
Achievement	13	3,770	−0.225	0.315	−.272	.321	0.931	0.107
Cautiousness	5	1,749	−0.172	0.252	−.208	.242	0.921	0.115
Dependability	18	3,750	−0.145	0.307	−.175	.284	0.857	0.119
Orderliness	15	3,572	−0.124	0.174	−.149	.000	0.948	0.125

Note: For each variable, k refers to the number of independent samples; N indicates the number of targets; d indicates the standardized mean difference between self- and informant reports; δ indicates the standardized mean difference between self- and informant reports, corrected for attenuation due to unreliability; and U refers to the standard-deviation ratios of self- and informant reports. For each measure of effect size, a standard deviation is also given. Corrected mean-difference scores (δ) and variability-difference scores (U) are presented in boldface for emphasis.

Table 2.

Meta-Analyses of the Big Five Traits Across Different Types of Informant Reports

			Mean difference				Variability ratio
Trait	k	N	d	SD_d	δ	SD _δ	U	SD_U
Emotional Stability
Family	38	7,218	−0.096	0.288	−.111	.233	0.984	0.102
Friends	65	12,926	−0.176	0.192	−.203	.000	1.039	0.147
Colleagues	7	917	−0.175	0.377	−.202	.328	0.984	0.126
Strangers	22	2,300	0.017	0.377	.020	.297	1.152	0.271
Extraversion
Family	34	6,805	−0.107	0.164	−.124	.000	0.976	0.060
Friends	62	13,212	−0.126	0.223	−.146	.128	0.979	0.113
Colleagues	8	1,040	−0.049	0.147	−.057	.000	0.962	0.087
Strangers	23	2,400	0.270	0.343	.314	.232	0.972	0.297
Openness/Intellect
Family	31	6,322	0.105	0.305	.126	.278	0.963	0.108
Friends	55	11,741	0.182	0.224	.218	.132	0.989	0.109
Colleagues	6	873	0.126	0.219	.151	.000	1.042	0.093
Strangers	20	2,429	0.655	0.542	.784	.563	0.967	0.221
Agreeableness
Family	31	6,322	−0.132	0.313	−.158	.289	0.877	0.098
Friends	57	11,586	−0.069	0.195	−.082	.000	0.873	0.087
Colleagues	6	873	0.008	0.196	.009	.000	0.870	0.133
Strangers	19	2,123	0.272	0.506	.325	.512	0.945	0.199
Conscientiousness
Family	31	6,322	−0.146	0.288	−.177	.251	0.895	0.084
Friends	62	11,825	−0.152	0.239	−.183	.148	0.906	0.091
Colleagues	6	873	−0.099	0.130	−.120	.000	0.912	0.076
Strangers	20	2,364	0.284	0.368	.343	.311	1.026	0.348

Furthermore, even when we separated out the traits into more specific facets, the magnitude of self-informant differences remained small (most δs ≤ .27). Within the Openness/Intellect domain, self-enhancement was most pronounced for the depth, nontraditionalism, and openness-to-sensations facets (δs = .320, .241, and .233, respectively). Note that these facets align more strongly with the Openness aspect of the domain. In contrast, facets from the Intellect aspect (where we hypothesized self-enhancement effects would emerge) showed minimal self-enhancement effects (for intellectual efficiency: δ = −.021). Thus, there appears to be some self-enhancement on the Openness/Intellect factor, but our findings do not suggest that this is uniquely driven by the aspect tied to intelligence. Beyond the Openness/Intellect domain, several facets high in visibility (e.g., dominance, orderliness) showed small trends toward self-effacement—dominance and activity (from the Extraversion domain): δs = −.272 and −.217, respectively; achievement and orderliness (from the Conscientiousness domain): δs = −.272 and −.149, respectively. Self-esteem (from the Emotional Stability domain) also showed slight tendencies toward self-effacement (δ = –.261). In general, however, focusing on specific facets within the trait did not generally change the magnitude of the self-informant differences or reduce the variability around these estimates (SD_δ). These findings may suggest that people are generally accurate but somewhat self-effacing when rating their own personality traits. Alternately, informants may be more likely to provide favorable reports of the target (other enhancement). Importantly, initial analyses also revealed moderate variability in the SD_δ values among the Big Five trait domains and facets. Such variability in effects suggests some potential for moderators to affect elevation accuracy.

To test the effects of trait visibility and evaluativeness on elevation accuracy (Hypotheses 1b and 1c), we conducted a multilevel analysis in which we regressed study-level effect sizes (standardized mean differences) on corresponding standardized evaluativeness and visibility scores (ratings made by raters), clustered by study. Thus, the regression coefficients from this model reflect the changes in the mean differences for an increase in evaluativeness or visibility of 1 standard deviation. Although the coefficients were significant given the large number of effect sizes analyzed (N = 1,321), the effects of evaluativeness (b = −0.04, 95% confidence interval, or CI = [−0.06, −0.03]) and visibility (b = −0.03, 95% CI = [−0.05, −0.01]) were quite small and for evaluativeness in the opposite directions as hypothesized. Overall, trait evaluativeness and visibility do not appear to markedly impact elevation accuracy (although these findings do not speak to their effects on other forms of accuracy).

Next, we conducted separate meta-analyses on the basis of the relationship that the informant had with targets (see Table 2 and Fig. 1). Consistent with our Hypothesis 1d, results showed that mean differences in effect size were more positive in studies using strangers as informants than studies using family members or friends. In particular, self-reported Extraversion, Agreeableness, and Conscientiousness showed moderate self-enhancement relative to stranger reports of those traits (δs = .314, .325, and .343, respectively), with this effect being even more pronounced for Openness/Intellect (δ = .784). Interestingly, the self-enhancement effects that appeared for studies using strangers were not mimicked by studies using work colleagues. Rather, the magnitude and direction of the effects for colleagues generally followed that of family members and friends in producing very modest self-effacement for Emotional Stability, Extraversion, Agreeableness, and Conscientiousness.

Fig. 1.

Mean difference in effect size between self- and informant reports for each of the Big Five traits, separately for each type of relationship that informants had with targets.

Variability differences

As summarized in Tables 1 and 2, the U ratios for the overall meta-analyses and for the meta-analyses across potential moderator variables were very close to 1.00, suggesting generally similar degrees of variance between self- and informant reports of the Big Five traits with only a few minor exceptions. The standard deviations of Agreeableness tended to be slightly smaller for self-ratings than informant ratings, particularly those of close acquaintances. This finding suggests that tendencies to self-enhance by individuals on the low end of the Agreeableness continuum may be balanced by tendencies to self-efface by individuals on the higher end. Discrepancies in variability were observed in some other facets, such as skepticism, self-esteem, curiosity, cooperation, modesty, and dependability. It is important to emphasize that the U ratios were very close to 1.00, paralleling our previous findings of small mean differences. Furthermore, no consistent variability differences were observed by informant type.

We also conducted separate multilevel analyses on variability ratios using the standardized evaluativeness and visibility scores: b = −0.02, 95% CI = [−0.03, −0.01] for evaluativeness; b = −0.01, 95% CI = [−0.02, 0.00] for visibility. Overall, these results suggest similarities between self- and informant reports of Big Five personality traits in variance.

Publication-bias analysis

We examined funnel plots to visually investigate the relationship between the estimated mean differences and the standard errors, separately for each Big Five trait (see Fig. 2). As shown, the effect sizes were close to zero. Furthermore, the relatively symmetric funnel plot suggests that our meta-analysis did not appear to prioritize consideration of statistically significant effects over nonsignificant effects. We further confirmed this by using the Egger regression method (Egger, Smith, Schneider, & Minder, 1997) and Begg-Mazumdar test (Begg & Mazumdar, 1994), both of which showed nonsignificant p values and low correlations, indicating no publication bias (i.e., larger effects were not disproportionately represented in smaller sample studies; see the Supplemental Material).

Fig. 2.

Funnel plot for each of the Big Five traits, across all studies in the meta-analysis. Each dot represents an individual study and is positioned according to its standardized mean difference (x-axis) and standard error (y-axis). The white triangle indicates where 95% of the data points would lie in the absence of a publication bias. The vertical line represents the average standardized mean difference found in the meta-analysis.

Discussion

The current meta-analytic review of self- and informant-report means across 152 independent samples found strong support for the elevation accuracy of personality ratings. These findings generally held across the Big Five traits. Finding such elevation accuracy is surprising, given the extensive literature on self-enhancement (and its frequent citation as a critique of personality research’s strong reliance on self-report measures). The only trait that showed some evidence of self-enhancement was Openness/Intellect (primarily via the depth, nontraditionalism, and openness-to-sensations facets), but this effect was small. Notably, the other Big Five showed trends toward elevation accuracy or even showed slight trends toward self-effacement, with the largest effects obtained for Emotional Stability. One potential explanation may be that informant reports underestimate targets’ experience of internal phenomena and accordingly describe higher levels of Emotional Stability and higher levels of Openness/Intellect than self-reports (cf. Vazire, 2010). However, our multilevel analyses found that specific facets’ elevation accuracy was not strongly associated with ratings of the facets’ evaluativeness and visibility. These minimal deviations in elevation-accuracy differences were paralleled by the results for U ratios, which indicated minimal differences between the variances of self- and informant reports, suggesting high accuracy in both self- and informant reports.

Self-report means emerged as more positive only when compared with means from stranger reports. Two possibilities may explain this particular finding. First, compared with close acquaintances who may have offsetting friendship biases, strangers may be considerably more critical in evaluating the target. Notably, however, studies using work colleagues (who presumably also have less intimacy with targets and lack friendship biases) did not follow the same pattern as studies using strangers. This suggests that either colleagues are similarly intimate with targets or strangers may simply be particularly harsh in judging other people. Alternately, it may be that the typical thin-slice paradigms used in zero-acquaintance research may suppress target individuals’ expression of the positively valenced pole of the personality. Further research is warranted to identify the mechanism of the difference between self- and stranger reports.

Two explanations may account for our findings that self-reports of personality are not positively biased. First, given the importance of personality for influencing important life outcomes, perhaps people prioritize accuracy (cf. Kunda, 1990). Furthermore, personality traits vary less in their evaluativeness and controllability, whereas skills and performances (which are used to examine self-enhancement in social-cognition literature) vary largely in their degree. Thus, people may be less likely to self-enhance on personality traits because these are all similarly desirable to have, whereas certain abilities (e.g., athletic abilities) are not necessarily more desirable to have. Indeed, all five traits are socially desirable and evaluative to some extent. Additionally, self-enhancement bias may be more likely to be observed in specific contexts, such as in social comparison. Self-serving biases can be taken as evidence of biased responses to audience presence (Baumeister, Tice, & Hutton, 1989). Finding elevation accuracy is in line with classic (e.g., Rogers, 1951) and contemporary (e.g., Sheldon, 2014) humanistic theories of personality development that posit that people strive for accurate self-knowledge of inner thoughts and feelings. From this perspective, self-enhancement on personality may be less of a common phenomenon and more of an individual difference that has psychological consequences. Indeed, there might be consequences on either end for biased perception; that is, some of the people who self-enhance might be adjusted (or maladjusted; e.g., Carlson, Vazire, & Oltmanns, 2011). A second explanation may be that self-enhancement is present, but people work to ensure that others’ impressions are commensurate with their self-perceptions. For example, the self-verification literature is replete with examples of how individuals selectively affiliate with others who confirm their self-perceptions (Swann, 1983), particularly when nominating an informant for a research study (Leising, Erbs, & Fritz, 2010). In addition, socioanalytic theory has long posited that the impression-management processes that may affect self-reports of personality are likely to be mimicked in targets’ actual behavior (Hogan, Jones, & Cheek, 1985). Thus, accuracy may be a self-fulfilling prophecy in that people work to transform their enhanced self-perceptions into social realities. In such conceptualizations, relegating self-enhancement in personality to the status of a psychometric bias would substantially underrate its origins in identity, reputation, and relationships.

Conclusion

Person perception is an active process and a fundamental aspect of our daily lives. We make personality judgments of ourselves and other people continually. A widely held notion is that self-reports are not as accurate as people think and that people are generally positively biased. The present meta-analytic review found little support for conceptualizing self-enhancement as a motivation that overpowers elevation accuracy. The only instances of modest mean differences between self- and informant reports occurred for Openness/Intellect (higher self-report means). Moderator analyses found moderate discrepancies between self- and informant reports only when informants were strangers, compared with acquaintances. These findings suggest high accuracy in self- and informant reports and indicate that strangers are more likely to be critical in their evaluations of the target’s personality. Future research may be well served by studying self-enhancement in personality as less of a general bias in self-perception and more of an individual difference capable of propagating behaviors critical to life outcomes.

Supplemental Material

KimOpenPracticesDisclosure – Supplemental material for Self–Other Agreement in Personality Reports: A Meta-Analytic Comparison of Self- and Informant-Report Means

Supplemental material, KimOpenPracticesDisclosure for Self–Other Agreement in Personality Reports: A Meta-Analytic Comparison of Self- and Informant-Report Means by Hyunji Kim, Stefano I. Di Domenico and Brian S. Connelly in Psychological Science

Supplemental Material

KimSupplementalMaterial – Supplemental material for Self–Other Agreement in Personality Reports: A Meta-Analytic Comparison of Self- and Informant-Report Means

Supplemental material, KimSupplementalMaterial for Self–Other Agreement in Personality Reports: A Meta-Analytic Comparison of Self- and Informant-Report Means by Hyunji Kim, Stefano I. Di Domenico and Brian S. Connelly in Psychological Science

Footnotes

Acknowledgements

The authors would like to thank the researchers who provided data and additional information about their studies and the experts who completed the survey.

Action Editor

Brent W. Roberts served as action editor for this article.

Author Contributions

H. Kim is the project lead and primary author of this article. She was responsible for developing the aims and hypotheses of the study, compiling the meta-analytic database, coding studies, conducting analyses, choosing the appropriate methods for the analyses, interpreting and writing the results, and drafting and revising the manuscript. S. I. Di Domenico contributed to developing the aims of the study, compiling the meta-analytic database, coding studies, conducting analyses; interpreting and writing the results, and revising the manuscript. B. S. Connelly contributed to developing the aims and hypotheses of the study, compiling part of the meta-analytic database, choosing the appropriate methods for the main analyses, and writing and revising the manuscript. All the authors approved the final version of the manuscript for submission.

ORCID iD

Hyunji Kim

Declaration of Conflicting Interests

The author(s) declared that there were no conflicts of interest with respect to the authorship or the publication of this article.

Funding

This research was supported by a Social Sciences and Humanities Research Council Insight Grant awarded to B. S. Connelly.

Supplemental Material

Additional supporting information can be found at

Open Practices

All data and materials have been made publicly available via the Open Science Framework and can be accessed at osf.io/snfjx. The design and analysis plans for the studies were not preregistered. The complete Open Practices Disclosure for this article can be found at https://journals-sagepub-com.web.bisu.edu.cn/doi/suppl/10.1177/0956797618810000. This article has received the badges for Open Data and Open Materials. More information about the Open Practices badges can be found at .

References

Alicke

M. D.

(1985). Global self-evaluation as determined by the desirability and controllability of trait adjectives. Journal of Personality and Social Psychology, 49, 1621–1630.

Alicke

M. D.

Klotz

M. L.

Breitenbecher

D. L.

Yurak

T. J.

Vredenburg

D. S.

(1995). Personal contact, individuation, and the better-than-average effect. Journal of Personality and Social Psychology, 68, 804–825.

Baumeister

R. F.

Tice

D. M.

Hutton

D. G.

(1989). Self-presentational motivations and personality differences in self-esteem. Journal of Personality, 57, 547–579.

Begg

C. B.

Mazumdar

(1994). Operating characteristics of a rank correlation test for publication bias. Biometrics, 50, 1088–1101.

Brown

J. D.

(1986). Evaluations of self and others: Self-enhancement biases in social judgments. Social Cognition, 4, 353–376.

Carlson

E. N.

Vazire

Oltmanns

T. F.

(2011). You probably think this paper’s about you: Narcissists’ perceptions of their personality and reputation. Journal of Personality and Social Psychology, 101, 185–201.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Connelly

B. S.

Ones

D. S.

(2010). An other perspective on personality: Meta-analytic integration of observers’ accuracy and predictive validity. Psychological Bulletin, 136, 1092–1122.

Cronbach

L. J.

(1955). Processes affecting scores on “understanding others” and “assumed similarity.” Psychological Bulletin, 52, 177–193.

10.

Dunning

Griffin

Milojkovic

J. D.

Ross

(1990). The overconfidence effect in social prediction. Journal of Personality and Social Psychology, 58, 568–581.

11.

Edwards

A. L.

(1957). The social desirability variable in personality assessment and research. New York, NY: Dryden.

12.

Egger

Smith

G. D.

Schneider

Minder

(1997). Bias in meta-analysis detected by a simple, graphical test. British Medical Journal, 315, 629–634.

13.

Funder

D. C.

(1987). Errors and mistakes: Evaluating the accuracy of social judgment. Psychological Bulletin, 101, 75–90.

14.

Funder

D. C.

(1995). On the accuracy of personality judgment: A realistic approach. Psychological Review, 102, 652–670.

15.

Funder

D. C.

West

S. G.

(1993). Consensus, self-other agreement, and accuracy in personality judgment: An introduction. Journal of Personality, 61, 457–476.

16.

Harris

M. M.

Schaubroeck

(1988). A meta-analysis of self-supervisor, self-peer, and peer-supervisor ratings. Personnel Psychology, 41(1), 43–62.

17.

Hogan

Jones

W. H.

Cheek

J. M.

(1985). Socioanalytic theory: An alternative to armadillo psychology. In Schlenker

B. R.

(Ed.), The self and social life (pp. 175–198). New York, NY: McGraw-Hill.

18.

Hooper

A. C.

(2007). Self-presentation on personality measures in lab and field settings: A meta-analysis. Unpublished doctoral dissertation, University of Minnesota, Minneapolis.

19.

John

O. P.

Robins

R. W.

(1993). Determinants of interjudge agreement of personality traits: The Big Five domains, observability, evaluativeness, and the unique perspective of the self. Journal of Personality, 61, 521–551.

20.

Kenny

D. A.

(2004). PERSON: A general model of interpersonal perception. Personality and Social Psychology Review, 8, 265–280.

21.

Kenny

D. A.

Albright

(1987). Accuracy in interpersonal perception: A social-relations analysis. Psychological Bulletin, 102, 390–402.

22.

Komar

Brown

D. J.

Komar

J. A.

Robie

(2008). Faking and the validity of conscientiousness: A Monte Carlo investigation. Journal of Applied Psychology, 93, 140–154. doi:10.1037/0021-9010.93.1.140

23.

Kunda

(1990). The case for motivated reasoning. Psychological Bulletin, 108, 480–498.

24.

Kwan

V. S.

John

O. P.

Kenny

D. A.

Bond

M. H.

Robins

R. W.

(2004). Reconceptualizing individual differences in self-enhancement bias: An interpersonal approach. Psychological Review, 111, 94–110.

25.

Leising

Erbs

Fritz

(2010). The letter of recommendation effect in informant reports of personality. Journal of Personality and Social Psychology, 98, 668–682.

26.

Mabe

P. A.

West

S. G.

(1982). Validity of self-evaluation of ability: A review and meta-analysis. Journal of Applied Psychology, 67, 280–296.

27.

Mueller-Hanson

Heggestad

E. D.

Thornton

G. C. I.

(2003). Faking and selection: Considering the use of personality from select-in and select-out perspectives. Journal of Applied Psychology, 88, 348–355.

28.

Nunnally

J. C.

(1978). Psychometric theory. New York, NY: McGraw-Hill.

29.

Paulhus

D. L.

(1991). Measurement and control of response bias. In Robinson

J. P.

Shaver

P. R.

Wrightsman

L. S.

(Eds.), Measures of personality and social psychological attitudes (pp. 17–59). San Diego, CA: Academic Press.

30.

Roberts

B. W.

Kuncel

N. R.

Shiner

Caspi

Goldberg

L. R.

(2007). The power of personality: The comparative validity of personality traits, socioeconomic status, and cognitive ability for predicting important life outcomes. Perspectives on Psychological Science, 2, 313–345.

31.

Rogers

C. R.

(1951). Client-centered therapy: Its current practice, implications, and theory. Boston, MA: Houghton Mifflin.

32.

Schmidt

F. L.

Hunter

J. E.

(2015). Methods of meta-analysis: Correcting error and bias in research synthesis (3rd ed.). Thousand Oaks, CA: SAGE.

33.

Sedikides

Gregg

A. P.

(2007). Portraits of the self. In Hogg

M. A.

Cooper

(Eds.), The SAGE handbook of social psychology (pp. 110–138). London, England: SAGE.

34.

Sheldon

K. M.

(2014). Becoming oneself: The central role of self-concordant goal selection. Personality and Social Psychology Review, 18, 349–365.

35.

Swann

W. B.

(1983). Self-verification: Bringing social reality into harmony with the self. In Suls

Greenwald

A. G.

(Eds.), Social psychological perspectives on the self (Vol. 2, pp. 33–66). Hillsdale, NJ: Erlbaum.

36.

Taylor

S. E.

Brown

J. D.

(1988). Illusion and well-being: A social psychological perspective on mental health. Psychological Bulletin, 103, 193–210.

37.

Vazire

(2010). Who knows what about a person? The self–other knowledge asymmetry (SOKA) model. Journal of Personality and Social Psychology, 98, 281–300.

38.

Wood

Harms

Vazire

(2010). Perceiver effects as projective tests: What your perceptions of others say about you. Journal of Personality and Social Psychology, 99, 174–190.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.58 MB

0.98 MB