David A. Kenny
April 4, 2021

List of Topics

This page consists of a range of topics related to self-perception. Click on a topic to view it.

Self-Enhancement and Effacement: Do people see themselves as better or worse that others?

Self-Accuracy vs. Other Accuracy: Who knows you better, yourself or others?

Self-Consistency: What is the consistency of self perception across traits, time, and interaction partners?

Self and the PERSON Model: How does the PERSON Model treat self-perceptions?

Chapter 7 of
Interpersonal Perception: The Foundation of Social Relationships pages 186-193.

Self-enhancement is the extent to which a people rate themselves more favorably than they rates others. Using Laing notation, it is symbolized as A(A) > A(B) and A(A) > B(A); that is a people see themselves as better than they see other and better than how they are seen by others.  As reported in 2019 by Michael Dufner and his colleagues, this bias has been variously called overconfidence, self-serving bias, self-favorability, overclaiming, self-deception, narcissism, better-than-average effect, comparative bias, bias blind spot, mnemic neglect, overoptimism, unrealistic optimism, optimistic bias, positive illusions, self-aggrandizement, self-arrogance, arrogance, self-inflation, self-love, and social desirability.

As shown by Kwan et al. (2004), to measure individual differences in self enhancement or self-effacement, the person's perceiver and target effect should be removed. This can be done by subtracting the person's perceiver and target effect from the self-perception. Better, but more complicated, is not to assume that the self weighs the perceiver and target effect to the extent as do others. One can empirically estimate the weight of the perceiver effect or k and the weight of the target effect or q in self perception. Note that what remains can be interpreted as the relationship effect for the self.

For western samples, we usually find self-enhancement: People see themselves as better than others. The one exception is the Big Five factor of emotional stability: People see others as more emotionally stable. Very recently, in 2019, Hyunji Kim, Stefano Di Domenico, and Brian Connelly have questioned whether self-enhancement even exists. They demonstrate that, although self-enhancement clearly exists in judgments of strangers’ personalities, the effect is virtually nonexistent when the targets are not strangers. They do replicate the finding that the lowest level of self-enhancement occurs for Neuroticism versus the other Big Five factors. It is well known that close others are idealized and seen overly positively, for instance, the 1996 study by Sandra Murray, John Holmes, and Dale Griffin. Thus the level of self-enhancement or effacement very much depends on
exactly which others are examined.

To measure properly individual differences in self-enhancement, one must do a double correction of removing perceiver and target effects, as recommended by Kwan et al. (2004). What remains is in essence, how the person uniquely perceives the self, i.e., the self-relationship effect.

An extensive study of k and q is the 2010 meta-analysis that I conducted with Tessa West of 24 studies with a total of 118 variables and 2,992 participants. In that analysis, we found the average value of k was 0.79, and of q was 0.64. Both values are statistically lower than 1, indicating that the perceiver and target effects are weighted less and have less importance in affecting self-perception than they do in the perception of others. This finding is consistent with the findings that self-other agreement is lower than consensus and assumed similarity is lower than assimilation.

Tessa West and I found hardly any evidence that k and q were moderated by visibility or evaluativeness. Such a result is consistent with the point made by Jüri Allik and colleagues (2010) that variance differences affect self-other agreement. Both self–other agreement and assumed similarity are correlations, and correlations are affected by variance: Generally, the larger the variance, the larger the correlation. However, both k and q are regression coefficients and should be less affected by variance differences. Thus, it appears that much of the effects of visibility and evaluation are mediated by changes in variance of ratings.

Two studies partitioned self-perceptions into three parts of perceiver, target, and relationship effects, and then correlated each with some external measures. The first is a 2008 study by Jan-Erik Lönnqvist and colleagues, which measured the association of perceptions with psychological adjustment. A sample of 199 Finnish military cadets showed that a negative relation between adjustment and personality ratings was more due to the target effect than to self-enhancement

The second is a 2014 study by Niels van der Kam and colleagues, who examined the effects of transformational leadership perceptions on the quality of their behavior with the others in the group, something called leader–member exchange, or LMX. They found positive effects for the target but negative effects for both the perceiver and relationship (i.e., self-enhancement) effects. They also demonstrated that the negative effects of self-enhancement were moderated by the Extraversion of the target: The more extraverted the target, the more negative is the effect of perceiver’s self-enhancement on the quality of the perceiver’s interactions with the target.

Allik, J., Realo, A., Mõttus, R., Esko, T., Pullat, J., & Metspalu, A. (2010). Variance determines self–observer agreement on the Big Five personality traits. Journal of Research in Personality, 44, 421–426.

Dufner, M., Gebauer, J. E., Sedikides, C., & Denissen, J. J. A. (2019). Self-enhancement and psychological adjustment: A meta-analytic
review. Personality and Social Psychology Review, 23, 48-72.

Kenny, D. A., & West, T. V. (2010). Similarity and agreement in self- and other perception: A meta-analysis. Personality and Social Psychology Review, 14, 196–213.

Kwan, V. S. Y., 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.

Lönnqvist, J.-E. (2008). Issues in socially desirable responding and personality research. Unpublished doctoral dissertation, University of Helsinki, Finland.

Murray, S. L., Holmes, J. G., & Griffin, D. W. (1996). The self-fulfilling nature of positive illusions in romantic relationships: Love is not blind, but prescient. Journal of Personality and Social Psychology, 71, 1155–1180.

van der Kam, N. A., van der Vegt, G. S., Janssen, O., & Stoker, J. I. (2014). Heroic or hubristic?: A componential approach to the relationship between perceived transformational leadership and leader–member exchanges. European Journal of Work and Organizational Psychology, 24, 611–626.


Chapter 7 of Interpersonal Perception: The Foundation of Social Relationships pages 193-196.

To what extent does a person's self-rating correlate higher with the person's behavior than does the rating of an observer. To be fair, if there are multiple observers, the correlation needs to adjusted down to a single observer.


Simine Vazire developed the influential Self–Other Knowledge Asymmetry (SOKA) model in 2010. According to SOKA, the self should be more accurate than others for traits low in visibility, whereas others should be more accurate than the self for traits high in evaluativeness (e.g., intellect). This was shown in a study Vazire conducted with Matthias Mehl in 2008. They found evidence that for some behaviors the self was more accurate, and for others a peer was more accurate. Thus there is not a simple answer to the question of who is more accurate.

In perhaps the most extensive analysis of the relative accuracy of
self- versus other-perceptions comes from the previously discussed meta-analysis conducted by Connelly and Ones in 2010. They reviewed how personality (e.g., the Big Five) predicts life outcomes. This study is not an accuracy study per se but, rather, a validity study. They examined the areas of academic achievement and job performance. In nine of the 10 tests, Big Five ratings by another person were a better predictor of both types of performance than self-ratings. The one exception was Agreeableness and Academic Achievement, which had a trivial correlation of .06 for self-ratings versus a .02 correlation for the ratings of another person.

Connelly, B. S., & Ones, D. S.  (2010). An other perspective on personality: Meta‑analytic integration of observers’ accuracy and predictive validityPsychological Bulletin, 136, 1092-1122.

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

Chapter 7 of
Interpersonal Perception: The Foundation of Social Relationships, pages 194 and 196-197.

As done for judgments of others, the consistency of self-ratings across several modes can be addressed. This is done for traits, times, traits, and interaction partners.

It is possible to examine the consistency of the relationship effect in self-perceptions, which is self-enhancement. Here the consistency of the self-relationship across different traits is examined using the data from the 1997 study done by Bernadette Park, Sue Kraus, and Carey Ryan of University of Colorado dormitory residents. For example, the correlation between the self-relationship effects for Agreeableness and Culture were computed for the three waves (.12, .28, and .24), and the median of the three values (.24) was used as the correlation between the two variables. The average of all of the 10 possible pairs of Big Five traits is only .03, with three of the 10 correlations being negative. The one large correlation, .65, is between Agreeableness and Conscientiousness factors. It should be noted that, because of the complex method of measuring k and q, these correlations are unstable, and so this conclusion is quite tentative. That said, the weak and even negative correlations are consistent with the theorizing of the 2017 paper by Thomas Lösch, Augustin Kelava, and colleagues, who argue that self-judgments are more highly differentiated than the judgments of others. They base their argument on the 1986 internal–external frame of reference model of Herbert Marsh.


To examine consistency over time, again the 1997 Park and colleagues
study is used. Over-time correlations of self-relationship effects between adjacent waves were computed, and the mean of the 10 correlations was determined. That median stability was .80, which is somewhat greater than the stability for relationship effects (r = .75) but lower than that for target effects (r = .96).

Based on the pioneering work by Micha Strack (2004), it is relatively simple to use the SRM to measure self-perceptions as a function of interaction partner. One simply asks the person for a self-rating when they are interacting with different people. For instance, Elton rates how friendly he feels after he interacts with Bernie and Barry. In this way, self-judgments are dyadic variables whose variance can be partitioned into SRM components. The SRM components of perceiver, target, and relationship have the following interpretations:

  • Perceiver: how a person sees her- or himself across all interactions,
  • Target: how a person “makes” others view themselves when they interact with him or her, and
  • Relationship: how a person perceives her- or himself differently depending on the particular interaction partner.

To illustrate the variance partitioning for self-judgments, consider self-ratings of Friendliness. A large amount of perceiver variance would reflect the tendency for some individuals to believe that they are friendly across all of their interactions and for others to believe that they are unfriendly across all of their interactions (i.e., there is variability in the extent to which people report being friendly across all interaction partners). A large amount of target variance indicates that some individuals elicit self-ratings of Friendliness from others, whereas other individuals do not elicit vibes of Friendliness. To clarify, when people interact with Elton, they tend to see themselves as friendly, but when they interact with Bernie, they see themselves as unfriendly. Finally, a large amount of relationship variance reflects that individuals see themselves as friendly with particular interaction partners and as unfriendly with certain other particular interaction partners.

The view that self-perceptions do not differ as a function of interaction partner or social context implies a substantial amount of perceiver variance and relatively little target and relationship variance. That is, people see themselves the same way across the context of interaction partners. The presence of substantial partner and relationship variance would suggest that a self-rating is not fixed but is rather a very social phenomenon. In the Kenny and West (2008) review of six studies that measured self-ratings with different interaction partners, the bulk of the variance, about 44%, is at the level of the perceiver: People see themselves in essentially the same way across different interaction partners. Target variance is very weak, averaging only 3%. Thus we do not find that some people elicit the same self-perceptions across partners. Only one study, that by Micha Strack in 2004, separated error variance from relationship variance. Therefore, relationship variance can only be examined in this study. The Strack study found nontrivial levels of relationship variance, 11% of the total variance. Self-perceptions did change with interaction partner, but the amount of change is relatively small relative to the amount of perceiver variance.

Lösch, T., Kelava, A., Nagengast, B., Trautwein, U., & Lüdtke, O. (2017). Perspective matters: The internal/external frame of reference model for self- and peer ratings of achievement. Learning and Instruction, 52, 80–89.

Kenny, D. A., & West, T. V. (2008). Self-perception as interpersonal perception. In J. Wood, J. Holmes, & A. Tesser (Eds.), Self and relationships (pp. 119–138). New York: Psychology Press.

Marsh, H. W. (1986). Verbal and math self-concepts: An internal/external frame of reference model. American Educational Research Journal, 23, 129–149.

Park, B., Kraus, S., & Ryan, C. S. (1997). Longitudinal changes in consensus as a function of acquaintance and agreement in liking. Journal of Personality and Social Psychology, 72, 604–616.

Strack, M. (2004). Sozialperspektivitat: Theoretische bezuge, forschuungsmethodik und wirtschaftpsychologische praktikabilitat eins beziehungsdiagnostischen konstrukts [Social perspectivity: Theoretical references, research methodology and economic practicability of a relationship-diagnostic construct]. Unpublished doctoral dissertation, University of Göttingen, Göttingen, Germany.



Chapter 7 of
Interpersonal Perception: The Foundation of Social Relationships pages 198-202.

When the perceiver and target are the same person, presumably the number of acts, or n, is very large. People have observed thousands of their own acts. With n so very large, the self-impression would be the sum of P and O, and the other four components would essentially vanish. Thus self-perception within the Person Model is P + O.

What, then, is the meaning of P and O in self-perception? The definition of P is that it is the average of all the scale values of all possible acts for a given target across all possible perceivers. Note that this definition gives no special status to the self as a perceiver, nor does it allow for the fact that some “behaviors” might be accessible only to the self. Those behaviors are not included in P. Moreover, based on the work of Geeza (2010), it was argued that P varied by context, and so different perceiver groups, for example, friends, family, and coworkers, have a different P. Presumably, the P in self-perception is the cross-context P, that is, the P averaged over all perceiving contexts, which was denoted as Pg. Given these arguments, the correlation between the P or Personality components for self- and other-perceptions is then not 1, but rather .89.

What is the meaning of O for self-perception? Because behaviors to which only the self has access are not represented in P, some of O contains the special access that the self has. Thus part of O contains a potentially valid piece. However, it is also well established that self-perceptions are subject to several biases that the perception of others is not. Thus it seems reasonable to expect that there is more O variance for self-perceptions than for the impressions of others. The suggestion on page 199 of the 2020 Interpersonal Perception book is to set the O variance for self to 2.2 (a value greater than the 1.5 value for the impressions of others).

The conclusion that self–other agreement increases as a function of acquaintance is consistent with the forecasts of the PERSON model, showing that accuracy increases with acquaintance because, in many studies of accuracy, self-ratings are often used as a measure of the truth. If consensus increases only slightly with acquaintance, but self–other agreement increases, then the expectation would be for more consensus than self–other agreement when others are not well acquainted with the target, but a narrowing gap when they are well acquainted. The evidence from the two 1993 John and Robins studies would seem to support this conclusion because the gap between self–other agreement and consensus was much smaller when the participants had known the targets on average for 18 years than it was when participants were only moderately familiar.

Geeza, A. A. (2010). A meta-analytic examination of consistency in informants’ perspectives across contexts. Master’s Thesis, University of Connecticut.

John, O. P., & Robins, R. W. (1993). Determinants of interjudge agreement on personality traits: The Big Five domains, observability, evaluativeness, and the unique perspective of the self. Journal of Personality, 61, 521–551.

Go to the next Interpersonal Perception page.

Go to the Interpersonal Perception menu.