Jarret Crawford and I recently had our paper “Answering Unresolved Questions about the Relationship between Cognitive Ability and Prejudice” accepted at Social Psychological and Personality Science. [Preprint & SOM]
Abstract: Previous research finds that lower cognitive ability predicts greater prejudice (Onraet et al., 2015). We test two unresolved questions about this association using a heterogeneous set of target groups and data from a representative sample of the United States (N=5914). First, we test “who are the targets of prejudice?” We replicate prior negative associations between cognitive ability and prejudice for groups who are perceived as liberal, unconventional, and having lower levels of choice over group membership. We find the opposite (i.e. positive associations), however, for groups perceived as conservative, conventional, and having higher levels of choice over group membership. Second, we test “who shows intergroup bias?” and find that people with both relatively higher and lower levels of cognitive ability show approximately equal levels of intergroup bias, but towards different sets of groups.
This paper aims to get a better grasp on how cognitive ability is associated with prejudice. We borrowed a strategy from our prior work on political ideology and prejudice and used many different target groups. Doing so allowed us to find a few new things about the cognitive ability-prejudice association that we didn’t know before.
First, we found that the association between the two variables varies a lot from target group to target group. Prior research consistently found that people with lower levels of cognitive ability express more prejudice than people with higher levels of cognitive ability. However, prior research only focused on small number of target groups. When we use many different targets of prejudice we find that sometimes the association is negative (low cognitive ability –> more prejudice) and sometimes the association is positive (high cognitive ability –> more prejudice).
Second, we found that the size and direction of the association between cognitive ability and prejudice depends on perceived characteristics of the target groups. We asked a second sample of people to rate these 24 groups on three characteristics: the groups perceived ideology (we included by political ideology and conventionalism), status, and choice over being a group member. We then correlated these characteristics with the correlation between cognitive ability and prejudice.
Here we found that the association between cognitive ability and prejudice tended to be negative for groups who were perceived as low choice, low status, and liberal/unconventional. We found that the association was positive for groups who were perceived as high choice, high status, and conservative/conventional. When we put these characteristics into the same model (to control for the correlation between the characteristics), we found that choice and ideology remained predictors, but the effect of perceived status went to just about zero.
Third, we found that the size of the prejudice was about the same for people low and high in cognitive ability. The pattern of positive and negative correlations could be consistent with a spreading interaction (it looks like a > or an open alligator mouth), where people with high cognitive ability like all groups about the same, but people with low cognitive ability dislike some groups and like others. The pattern could also be consistent with a cross-over interaction (it looks like a X), so that both people high and low in cognitive ability like some groups over other groups. To check this out, we looked at the prejudice towards groups with the 6 most positive and 6 most negative correlations between cognitive ability and prejudice. When we plot this data, we find what looks like a cross over interaction.
By looking at a well known and consistent correlation, the correlation between cognitive ability and prejudice, using many more target groups than prior research we were able to find out some new information. Most importantly, I think our data suggest that it is important to be careful making conclusions about what predicts prejudice when, espcially if research is only based on data using a small number of target groups.