Results of a new survey I conducted in conjunction with the Associated Press and Stanford University suggest that racial attitudes will likely hurt President Obama’s reelection prospects in 2012. In general, we found that the prevalence of anti-Black attitudes increased between 2008 and 2012. Predicting the influence of racial attitudes on the expected vote shares for the two major party candidates, we suggest that the influence of anti-Black attitudes is likely to be between four and five percentage points and that the net influence of both pro- and anti-Black attitudes will reduce Obama’s proportion of the vote by around two percentage points. The Associated Press is reporting on the story in today’s news:

Associated Press Story on our Results

Associated Press Description of Methodology

The analytical strategy was presented in an article we published in Public Opinion Quarterly.

Here is the full report for the current study:

[gview file=””]


I am quoted on the issue in the Toronto Star.



10 Responses

  1. Very interesting study. Thanks for this important contribution to our understanding of American politics and cultural attitudes. I see that your analysis considered census region. Did you examine the role of location at any other geographic resolution? For example, did you look at survey responses by State (e.g., Mississippi vs Montana, etc)? Thanks again! Best, Justin

    • Justin — The controls for region were simply to ensure that what we were measuring was not a regional effect that we were interpreting as a racial one, but you also raise a notable question about whether the slopes might be different in different parts of the country. Unfortunately, we could not examine either question on a state by state basis because there were not enough people within each state to do an analysis.

  2. Josh – thanks for the reply. It might be worth looking at Census Divisions rather than Regions ( I’m not wild about the Census Division geography but I would guess that this resolution would provide a sufficient sample size and lead to a better glimpse of regional variability. Are the data used in the study available for download?

    • The data will be available in the future, but I don’t know exactly when — the AP has a standard system for distribution. As for the geographical analyses, I think they fall a little outside of the scope of the present study, but census divisions could be a good lens. I’m a little swamped at the moment, so I worry that I probably will not have a good opportunity to look at these questions for a while. Thanks so much for the suggestions though.

  3. So do you truly think that Americans who believe that African-Americans (blacks) should not deserve special favors are actually racists? How do you know if the attitude you are measuring has anything to do with race, for perhaps your respondents might respond the same to people of any race?

    • Jason,

      There is a long literature on this question I suggest reading. Notably, none of the questions alone can be taken as sufficient evidence of anti-Black attitudes, but there is considerable evidence that people who are high on a number of the Symbolic Racism measures behave in ways that reflect racial bias. Take a look at this review piece if you are interested in reading a bit more on the issue: Over thirty years later: A contemporary look at symbolic racism.

      • I hear you as I’m an academic myself with degrees in both Political and Behavioral Sciences, but isn’t it a stretch to equate small changes in a couple questions as the equivalent of a more “racist” nation, as many of the media stories did? I fully admit that the instruments you used show racial biases, but I feel that the implications were exaggerated.

        • Jason,

          I both agree and disagree. On one hand, there was a significant change in both measures in an anti-Black direction (albeit not huge). From that perspective I think that it is important to note that President Obama’s first term was associated not with racial progression, but with a slight regression. Conversely, I agree that noting that a majority of Americans could be considered racist was probably not the best way to point out this fact. I think the media story might have been better served by highlighting the fact that racial attitudes seemed to account for differences between the President and his challenger in similar ways in both 2008 and 2012.

          • Of course it would have been better served in the story to have highlighted what the data was actually appearing to say as I believe that would have been more intellectually honest. Unfortunately it also probably wouldn’t have been as exciting and ground breaking. I think sensationalism in journalism played a role in some of the headlines I saw. Nevertheless we have a correlation here between people who score highly on Symbolic Racism measures and people who engage in behavior reflecting a racial bias, but again how do we know this equals causation? I read the review but also read criticism of “symbolic racism” as a concept. Wouldn’t it be better to admit that this theory is still debatable and hence emphasize the word “SEEMED”? That last question isn’t directed specifically at you, but rather at any professional (academic or media) who discusses the issue.

Leave a Reply