bq. The present study examines the relationship between racial prejudice and reactions to President Barack Obama and his policies. Before the 2008 election, participants’ levels of implicit and explicit anti-Black prejudice were measured. Over the following days and months, voting behavior, attitudes toward Obama, and attitudes toward Obama’s health care reform plan were assessed. Controlling for explicit prejudice, implicit prejudice predicted a reluctance to vote for Obama, opposition to his health care reform plan, and endorsement of specific concerns about the plan. In an experiment, the association between implicit prejudice and opposition to health care reform replicated when the plan was attributed to Obama, but not to Bill Clinton—suggesting that individuals high in anti-Black prejudice tended to oppose Obama at least in part because they dislike him as a Black person. In sum, our data support the notion that racial prejudice is one factor driving opposition to Obama and his policies.
That is from a forthcoming paper by Eric D. Knowles, Brian S. Lowery, and Rebecca L. Schaumberg. A gated version is here. I cannot find an ungated version. This study is not yet published but is already getting some attention and, from Mike Munger, some criticism.
The study drew on a non-random group of subjects — e.g., it is 68% female — and gathered data on four occasions. In late October 2008, subjects’ level of “explicit racism” was measured via a series of questions. A measure of “implicit racism” was also constructed, using a version of the Implicit Association Test. In early November 2008, subjects answered a series of questions about Obama. In mid-November, they indicated their vote choice.
Then, almost a year later (Oct 1-3, 2009), a random half-sample of subjects was asked several questions about health care reform. The second half-sample was given a description of a health care reform proposal; half of these were told it was Obama’s and half were told it was Bill Clinton’s. Then they indicate their attitudes toward the proposal.
The measure of implicit racism predicted attitudes toward health care, primarily via its apparent effect on attitudes toward Obama himself. The effect of implicit racism was notable even when controlling for explicit racism — a finding that is of interest to some psychologists, although I will not dwell on it here. Moreover, in the Obama-Clinton experiment, the effect of racism predicted support for Obama’s plan but not Clinton’s plan (even though, again, the plans were described identically).
What should we make of this? The study’s major limitations are two. First, the sample is non-random and thus it is difficult to know how much can be generalized from it.
Second, support for health care is modeled only as a function of explicit and implicit racism, with attitudes toward Obama as a mediating variable. There are no additional control variables — party identification, ideology, and so on. All of these other factors could be associated with racism, attitudes toward Obama, and support for health care reform. The article’s estimated effects of racism, both direct and indirect, are suggestive but , it seems to me, far from conclusive. There is a long-standing debate in political science about the relative importance of racial prejudice and race-neutral political values, one which this article does not engage.
Even the experiment isn’t conclusive in this regard. An Obama frame makes racism a more significant predictor, but this could be spurious. Perhaps the Obama frame would prime other factors as well — e.g., partisanship, since Obama could be a more polarizing president than Clinton; or ideology, if people perceive Obama as more liberal than Clinton. I would prefer to see a model that includes more factors besides racism and then an analysis of whether the Obama and Clinton frames affect any or all of these factors.
A final point, which has to do with interpretation rather than research design: an individual-level effect of racism on attitudes does not necessarily translate into a large aggregate effect. If one estimated a model of health care attitudes that included multiple factors, my guess is that simulating the complete end of racism would not make health care reform much more popular. Any people who are prejudiced will also have other reasons for opposing reform. In this earlier post, I simulated the “end of racism” by making whites with a negative opinion of blacks relative to whites evaluate whites and blacks equivalently. This simulation added a single, solitary point to Obama’s share of the vote. The same could be true here.
I do believe that racial prejudice is likely to affect attitudes toward Obama and his policy proposals and am, like the authors, deeply skeptical of the Democracy Corps finding that racial prejudice was “almost beside the point” when it came to explaining opposition to Obama. However, this article provides only tentative evidence against that conjecture.