Dozens of studies have correlated students’ personality with their GPA (Grade Point Average). The reason is simple. Students are easily accessible, and GPA is an easily measured objective outcome variable. The main finding of these studies is that personality predicts GPA. This is not very surprising to the average reader or students themselves, but in psychology, it was once common to dismiss personality as an explanation for behavior and behavioral consequences. Personality matters!
The more interesting question is which personality traits predict GPA and how much of the variance in GPA they explain. Study after study and meta-analyses of these studies show that Conscientiousness (being orderly, planning ahead, persistence, and low impulsivity and procrastination) is the strongest personality predictor of GPA (O’Connor & Paunonen, 2007; Poropat, 2009). Effort matters, and some students are more able and willing to work hard for a higher GPA. Nobody claimed that psychology is rocket science.
However, it may be surprising how weak the relationship between Conscientiousness and GPA really is. The typical meta-analytic estimate is a correlation of approximately r = .24 (O’Connor & Paunonen, 2007; Poropat, 2009). This means that Conscientiousness explains only about 6% of the variance in GPA.
To make this correlation more meaningful, we can use it to compute the probability that a student who is above average in Conscientiousness also has an above-average GPA. Assuming a bivariate normal distribution, the probability is only about 58%, compared with 50% if Conscientiousness and GPA were unrelated.
This blog post uses data from a published journal article (McCredie & Kurtz, 2020) to examine three additional questions. First, do studies that rely on self-ratings underestimate the true relationship between Conscientiousness and GPA because self-ratings are imperfect measures of personality? Second, are informants, such as parents and peers, better judges of students’ Conscientiousness than students themselves? Finally, do evaluative biases in self-ratings reflect a real personality characteristic that predicts higher GPA, or are they merely rating biases?
The use of multiple raters also makes it possible to examine another interesting question: Does having a positive view of oneself predict higher GPA?
This possibility was suggested by Biderman et al. (2019), who examined evaluative factors in Big Five and HEXACO personality questionnaires. They found that a general evaluative factor predicted GPA even after controlling for the individual personality domains.
[Insert image of Biderman et al. (2019)]
However, their analyses relied exclusively on self-ratings. This makes it difficult to determine whether the evaluative factor represents a substantive personality characteristic or whether it captures valid personality variance that is not adequately separated from evaluative bias.
A multi-informant design provides a stronger test of this question.
Results: How Valid Are Self- and Informant Ratings?
I used structural equation modeling to reanalyze the open data. The model distinguished common personality factors from evaluative factors that were specific to each rater. Here, I summarize the key findings.
All three raters contributed to the measurement of Conscientiousness. The standardized loadings on the common Conscientiousness factor were:
- Self-ratings: .79
- Parent ratings: .75
- Peer ratings: .64
These coefficients indicate how strongly each rating reflects the common Conscientiousness factor. Their squares suggest that the common factor accounted for approximately 62% of the variance in self-ratings, 56% in parent ratings, and 41% in peer ratings.
Thus, self-ratings were at least as strongly related to the common factor as parent ratings, whereas peer ratings showed somewhat weaker agreement.
These findings are broadly consistent with the literature on self–other agreement in personality judgments (Connelly & Ones, 2010). There is no compelling evidence that informant ratings are consistently and substantially more valid than self-ratings. Both methods provide useful information, and their relative validity may depend on the trait, the informant, and the circumstances.
How Strongly Does Conscientiousness Predict GPA?
The common Conscientiousness factor predicted GPA with a correlation of approximately r = .31 (SE = .09).
This estimate is somewhat higher than the typical meta-analytic correlation of r = .24 based on observed personality scores. The difference is consistent with the expectation that measurement error attenuates correlations between observed personality scores and external outcomes.
However, the difference is not dramatic. Even a correlation of .31 explains less than 10% of the variance in GPA.
Thus, correcting for measurement error strengthens the relationship between Conscientiousness and academic performance, but it does not fundamentally change the conclusion: Conscientiousness matters, but most of the variance in GPA remains unexplained.
Do Positive Self-Evaluations Predict GPA?
One of the advantages of the multi-informant model is that it can distinguish common personality variance from rater-specific evaluative variance.
Personality ratings are influenced by the desirability of traits. People generally prefer to describe themselves and others as conscientious, agreeable, extraverted, emotionally stable, and open to experience. As a result, ratings of different personality traits may be positively correlated simply because a rater has a generally favorable impression of the person being rated.
This phenomenon is commonly called a halo effect.
The model identified clear evaluative factors in all three rating sources: self-ratings, parent ratings, and peer ratings.
If these evaluative factors represented a substantive personality characteristic, we might expect them to show positive agreement across raters. For example, people who evaluate themselves especially favorably might also receive especially favorable evaluations from their parents and peers.
However, the correlations among the three evaluative factors ranged from −.45 to .09, and none was statistically significant.
There was no consistent positive agreement across raters. This pattern is consistent with the interpretation that the evaluative factors primarily reflect rater-specific tendencies rather than a common personality characteristic.
The second finding was equally interesting. The evaluative factors did not predict GPA.
In particular, the evaluative factor in self-ratings was not a significant predictor of GPA (β = −.12, SE = .14).
The sample size limits strong conclusions about small effects. Nevertheless, the results provide no evidence that students who evaluate themselves more favorably across personality traits achieve higher grades after controlling for their substantive personality characteristics.
This finding is relevant to Biderman et al. (2019), who reported positive relationships between evaluative factors and GPA in three large samples. Their standardized regression coefficients ranged from approximately .15 to .21.
The present results do not establish that those findings were statistical artifacts. However, they illustrate an important identification problem in single-rater studies. A general evaluative factor extracted from self-report items may contain substantive personality variance as well as evaluative bias. Predicting an external outcome does not, by itself, establish which component is responsible for the relationship.
The multi-informant model provides a way to separate these sources of variance. In the present data, the common Conscientiousness factor predicted GPA, whereas the self-specific evaluative factor did not.
What About the Other Big Five Traits?
Conscientiousness was not the only personality trait associated with GPA in this sample.
Neuroticism was a negative predictor (β = −.22), suggesting that students who were more emotionally unstable tended to have lower grades. Openness was a positive predictor (β = .15).
Extraversion (β = −.05) and Agreeableness (β = −.01) were practically unrelated to GPA.
These findings should not be interpreted as establishing a universal pattern. The effects of personality traits other than Conscientiousness tend to be smaller and less consistent across studies (O’Connor & Paunonen, 2007; Poropat, 2009). The present sample is too small to draw strong conclusions about these additional predictors.
Why Did Peer Ratings Predict GPA?
Perhaps the most interesting finding was that the unique variance in peer ratings of Conscientiousness was related to GPA.
At first glance, this might suggest that peers are better judges of Conscientiousness than students themselves or their parents. However, this interpretation is difficult to reconcile with the lower loading of peer ratings on the common Conscientiousness factor.
If peer ratings were generally more valid indicators of Conscientiousness, we would expect them to show stronger agreement with the other measures, assuming that the common factor captures the valid trait variance.
An alternative explanation is that peers rely heavily on academic performance when evaluating their friends’ Conscientiousness.
For example, students who receive high grades may be perceived as conscientious even if they are disorderly, impulsive, or unreliable in other areas of life. Conscientiousness is a broad personality trait, and students may be highly conscientious about their academic work without being equally conscientious in other domains.
This would make peer ratings useful predictors of GPA without necessarily making them better measures of general Conscientiousness.
Another possibility is that peers observe academically relevant behaviors that are less visible to parents and are not adequately captured by students’ self-ratings.
These explanations are not mutually exclusive. Facet-level data and more detailed information about what peers actually observe would be needed to distinguish them.
The broader lesson is that unique predictive validity does not necessarily imply superior construct validity. A measure can predict an outcome because it captures criterion-specific information rather than because it is a better measure of the intended broad personality trait.
Conclusions
Three results stand out.
- Self-ratings and ratings by a single knowledgeable informant tend to be about equally valid. Some studies report differences in validity, but there is no consistent evidence that informant ratings are substantially superior to self-ratings. The present results are consistent with this broader conclusion.
- Personality matters for real-world outcomes, but the effects are not huge. Conscientiousness is a reliable predictor of academic performance, and correcting for measurement error produces a somewhat stronger relationship. Nevertheless, most of the variance in GPA remains unexplained. Personality matters, but it does not determine destiny.
- Evaluative factors in personality ratings appear to reflect primarily rater-specific bias. The lack of consistent positive agreement among evaluative factors across raters and their failure to predict GPA suggest that much of this variance reflects evaluative tendencies rather than a substantive personality characteristic. Even the evaluative component of self-ratings, which may reflect positive self-evaluation, provided no evidence of incremental validity for GPA.
As always, results from a single study are never conclusive. The sample is relatively small, and some estimates have substantial uncertainty. Nevertheless, the pattern of findings is broadly consistent with the existing literature on self–other agreement and the validity of personality ratings.
References
Biderman, M. D., McAbee, S. T., Hendy, N. T., & Chen, Z. J. (2019). Validity of evaluative factors from Big Five and HEXACO questionnaires. Journal of Research in Personality, 80, 84–96. https://doi.org/10.1016/j.jrp.2019.04.010
Connelly, B. S., & Ones, D. S. (2010). An other perspective on personality: Meta-analytic integration of observers’ accuracy and predictive validity. Psychological Bulletin, 136(6), 1092–1122. https://doi.org/10.1037/a0021212
O’Connor, M. C., & Paunonen, S. V. (2007). Big Five personality predictors of post-secondary academic performance. Personality and Individual Differences, 43(5), 971–990. https://doi.org/10.1016/j.paid.2007.03.017
Poropat, A. E. (2009). A meta-analysis of the five-factor model of personality and academic performance. Psychological Bulletin, 135(2), 322–338. https://doi.org/10.1037/a0014996