“The Claim That Personality Is More Important Than Intelligence in Predicting Important Life Outcomes Has Been Greatly Exaggerated”, 2022-05-01 (; similar):
We conduct a replication of et al 2016.
We show personality as less important than intelligence in predicting life outcomes.
For pay the predictive validity of intelligence twice as high as this of personality.
For educational attainment and grades it was 4.4× and 5.2× as high.
This finding contradict et al 2016 who argued that personality is more important.
We conduct a replication of et al 2016 who suggested that personality is more important than intelligence in predicting important life outcomes.
We focus on the prediction of educational (educational attainment, GPA) and occupational (pay) success, and analyze 2 of the databases that et al 2016 used (the NLSY79, n = 5,594 and the MIDUS, n = 2,240) as well as 4 additional databases, (the NLSY97, n = 2,962, the WLS, n = 7,646, the PIAAC, n = 3,605 and the ADD health, n = 3,553; all databases are American, except for the PIAAC which is German).
We found that for educational attainment the average R2 of intelligence was 0.232 whereas for personality it was 0.053. For GPA it was 0.229 and 0.024, respectively and for pay it was 0.080 and 0.040, respectively.
[Keywords: intelligence, personality, the Big-Five, life outcomes, educational attainment, income]
… et al 2016 approach of comparing the predictive power of intelligence and personality was straightforward. They compared the correlations between intelligence and important life outcomes to the correlations between personality and these outcomes. In our analyses we closely follow this approach. We focus on the Big Five personality dimensions as measures of personality, because they are central to et al 2016 work as well as to personality research in general, and because unlike the other personality measures used, which are specific personality traits, together the Big Five provide a full description of personality, and are commonly available in representative databases that measure life outcomes. We analyze 2 of the databases that were analyzed by (the NLSY79 and the MIDUS), avoiding the analysis of a third dataset, the BCS, because it did not include measures of the Big Five (a 4th data base et al 2016 analyzed, the Stela Maris dataset, did not include life outcomes). Instead, we added to our analyses other 4 large, nationally and internationally representative datasets—the NLSY97, the WLS, the ADD Health and the PIAAC.
In our analyses we focus on educational and occupational success as dependent variables representing life outcomes. Although et al 2016 included in their analysis, in addition to these outcomes, other outcomes such as depression, physical health, mental health and life satisfaction. One reason was that these 4 latter outcomes are assessed by subjective measures and therefore their correlations with personality are prone to biases stemming from social desirability, participants’ subjective interpretation of the questions, and common method variance associated with the use of rating scales in measuring both the dependent (the four latter outcomes) and the independent variables (the Big Five). In particular, common method variance may inflate the relationship between measures of personality and measures of subjective outcomes, since both are measured by self-reported rating scales.