A famous quote states that something that cannot be measured does not exist. This is of course not true, but if we want to move from anecdotal evidence to scientific evidence and theories of racism, we need valid measures of racism.
Social psychology has a long history of developing measures of racism and today there are dozens of different measures of racism. Unfortunately, psychologists are better at developing new measures than at validating existing ones. This makes research on racism vulnerable to criticism that racism measures are invalid or biased (Feldman & Huddy, 2005; Zigerell, 2015).
Take the item “Irish, Italians, Jewish and many other minorities overcame prejudice and worked their way up. Blacks should do the same without special favors” as an example. The item is one of several items that is used to measure a form of racism called symbolic racism.
Feldman and Huddy (2005) argue that items like this one have two components. A purely racist component where White people do not see Black people as equal citizens and a purely ideological component that opposes policies that favor any particular group, even if this group is disadvantaged by a history of racism. Whether the latter component is itself racist or not is not the topic of this blog post. My focus is rather on the separation of the two components. How can we separate agreement to the item that is based on racism from endorsement of the item for purely political reasons?
One solution to this problem is to see how endorsement of items with political content is related to items that have no political content. Using a statistical method called factor analysis it is then possible to separate the racial and the ideological component and to examine how much political orientation is related to the two components.
Indirect Measures of Racism
The problem with direct measures of racism is that open admission of racial bias has become less acceptable over time. This makes it harder to measure racism with items like “What about having a close relative marry a Black person? Would you be very in favor of it happening, somewhat in favor, neither in favor nor opposed to it happening, somewhat opposed, or very opposed to it happening?” Respondents may be unwilling to report their true feelings about this issue, especially when the interviewer is African American (Schaeffer, 1980; Schimmack, 2020).
Modern psychological testing with computers has made it possible to avoid these problems by measuring racism with computerized tasks that rely on participants’ behavior in response to racial stimuli. There are several tasks such as the evaluative priming task, the affective misattribution task and the popular Implicit Association Task (IAT). Unfortunately, the IAT has become known as a measure of implicit bias or implicit racism that is distinct from racism that can be measured with self-report measures. I have argued that there is no evidence that people can hide their feelings towards African Americans from themselves. It is more useful to see these tasks as alternative measures of racism that are less susceptible to fake responding. This does not mean that these tasks are perfect measures of racism because the use of computerized tasks creates new problems. Thus, there is no perfect measure of racism, but all valid measures of racism should be positively correlated with each other and the shared variance among these measures is likely to reflect variation in racism. The interesting question is whether political orientation is related to the shared variance among a variety of direct and indirect racism measures.
The data come from a study by Bar-Anan and Vianello (2018). The data have also been used in my critique of the IAT as a measure of implicit bias (Schimmack, 2020). The study assessed political orientation and racism with multiple measures. Political orientation was also measured with the standard and the brief IAT. In addition, participants reported whether they voted Republican or Democrat. Only White participants who reported voting were included in the analysis.
Racism was measured with the standard IAT, the brief IAT, the evaluative priming task, the Affective Missattribution Task, a direct rating of preference for White or Black people, and the Modern Racism Scale. Like other measures that have been criticized, the Modern Racism scale mixes racism and political ideology.
The interesting question is how much political orientation is related to the unique variance in the modern racism scale that is not shared with other racism measures and how much it is related to the shard variance with other racism measures.
The results show two clearly identified factors. The strong relationship between voting and the Republican factor (rep) shows that political orientation can be measured well with a direct question. In contrast, racism is more difficult to measure. The best measure in this study would be the direct preference rating (r_att) that is related .6 with the pro-White factor. But even this relationship implies that only about a third of the variance in the actual ratings reflects racism. The rest of the variance is measurement error. So, there is no gold-standard or perfect way to measure racism. There are only multiple imperfect ways. The results also show that the controversial Modern Racism Scale (mrs) reflects both racism (.444) and political orientation (.329). This shows that Republicans score high on Modern Racism in part because they reject social policies that favor minority groups independent of their attitudes towards Black Americans. However, the figure also shows that Republicans are more racist, as reflected in the relationship between the Republican and Racism factors (.416).
It is important that these results cannot be used to identify individuals or to claim that a particular Republican is a racist. The results do show however, that people who vote Republican are more likely to score higher on a broad range of racism measures whether they mention a political agenda or not.
Critics of racism research by social psychologists have argued that the research is biased because many social psychologists are liberal. The accusation is that social psychologists have created biased measures that conflate liberal policies with bigotry. Here I show that these critics have a valid point and that high scores on scales like the symbolic racism scale and the modern racism scale are influenced by attitudes towards egalitarian policies. However, I also showed that Republicans are more racist when racism is measured with a broad range of measures that have only racism as a common element.
Conservatives may be displeased by this finding, but recent events in 2020 have made it rather obvious that some Americans are openly racist and that these Americans are also openly supporting Trump. The real question for Republicans who oppose racism is how they can get rid of racism in their party.