Cognitive ability and party identity: No important differences between Democrats and Republicans Yoav Ganzach Faculty of Management, Tel Aviv University, Israel abstract article info Article history: Received 22 January 2016 Received in revised form 14 May 2016 Accepted 31 May 2016 Available online xxxx We examine the association between cognitive ability and party identity in the United States on the basis of two large databases. Contrary to recent findings (Carl, 2014a, 2014b) we find that when socio-economic status and race are controlled for, there are very few associations between the two. © 2016 Elsevier Inc. All rights reserved. 1. Introduction In two recent papers, Carl (2014a, 2014b) analyzed data from the United States General Social Survey (GSS) and concluded that indi- viduals who identify as Republicans have higher cognitive ability than individuals who identify as Democrats. In the current paper we replicate his analysis using appropriate controls and show that there is no important difference in cognitive ability between Repub- licans and Democrats. Our emphasis is on the Carl, 2014b paper, which more clearly focused on the Republican vs. Democrat difference, and used several measures of cognitive ability (the Carl, 2014a paper used only one measure). When regressing party identity on Socio-Economic Status (SES) and four measures of cognitive ability, Carl (2014b) found that cog- nitive ability had a direct effect on party identity as well an indirect effect, mediated by SES. Of these two effects, the direct effect is the more interesting one, since it suggests that when keeping SES con- stant, the more intelligent tend to identify more with the Republican party. The indirect effect is not of much interest since it reflects two well-known associations: The association between cognitive ability and SES (people in higher socio-economic strata tend to score higher on IQ tests) and the association between SES and party iden- tity (because of self-interest people from higher SES strata tend to identity more with the Republican party). Thus, the interesting finding in Carl's (2014b) is that even when SES is controlled for, R epublicans are more intelligent than Democrats. In the current paper we re-examine this finding. The major problem in Carl's (2014b) analyses is that he did not con- trol for race in his models. 1 Therefore, it is possible that his results are driven by the fact that Blacks tend to identify more with the Democratic party (because it is more supportive of minorities' rights) and at the same time tend to score lower on cognitive ability tests. That is, Carl's findings may simply reflect the tendency of Blacks to be Democrats and Whites to be Republicans. Indeed, in analyzing the relationship be- tween verbal ability and party identity for whites and non-whites sepa- rately based on some of the data that were used by Carl (2014b) in his paper, Meisenberg (2015) finds that a large part of the difference in cog- nitive ability between Republicans and Democrats “stem from the fact that lower-scoring non-white minorities predominantly support the Democratic Party (p. 143) 2” . In the current paper we examine the possibility that Carl's findings simply reflect the tendency of Blacks to be Democrats and Whites to be Republicans. We examine these two studies. In Study 1 we rely on the same data analyzed by Carl, replicating his analyses with only one change – we control for race. In Study 2 we examine the relationship be- tween party identity and cognitive ability on the basis of an additional database – the Wisconsin Longitudinal Study (WLS) – using a better measure of cognitive ability than the measures used by Carl. Intelligence 58 (2016) 18–21 E-mail address: yoavgn@post.tau.ac.il. 1 Curiously, in an appendix to his paper (Carl, 2014b, Appendix B), Carl reports an anal- ysis of only White participants, which could be viewed as an attempt to control for race. However, this analysis does not control for SES. Thus, while in the body of the paper Carl (2014b) controls for SES but not for race, in the appendix he controls for race but not for SES. (Note that our model treats race as an independent variable and both IQ and SES as mediators of party identity). 2 Meisenberg's (2015) analyses are based only on one measure of cognitive ability (vo- cabulary score). Carl's (2014b) analyses are based on additional three measures (see the method section below). http://dx.doi.org/10.1016/j.intell.2016.05.009 0160-2896/© 2016 Elsevier Inc. All rights reserved. Contents lists available at ScienceDirect Intelligence