March 18, 2008

Can Polls Capture Race and Gender Bias?

Topic A today is obviously Barack Obama's speech on race and the Jeremiah Wright controversy and by extension, the larger issues of race and gender and how they affect perceptions and vote preferences in the presidential election. I have several reactions from a pollster's perspective (though I'm not sure I have sufficient time to blog them all this afternoon). Let's start with some fascinating data that CBS pollster Kathy Frankovic included in her online column yesterday.

The analysis helps answer a question posed yesterday by my colleague Amy Walter, editor of The Hotline, raised in her a Hotline TV segment yesterday: "Can we ever really capture voters perceptions about race or gender in polling?"

Frankovic begins with the same question: "Will Americans admit to bias?" She reviews this history of questions posed by national surveys since the 1930s by Gallup and other pollsters, as well as the potential measurement problems this topic poses. Then she shares some data from some recent exit polls that I have not yet seen tabbed by race or gender elsewhere:

Polls in recent Democratic primaries suggest that something might be going on under the radar. We conduct exit polls on paper, so there is no interaction between an interviewer and a respondent, and therefore less opportunity for socially desirable answers. And in fact, in Ohio, one in five white men -- and nearly as many white women -- said the race of the candidate mattered to them. Those voters voted nearly four-to-one for Hillary Clinton, a much higher ratio than white voters who did not say race mattered. But even more black voters -- about one in four -- said race mattered to them, and nearly all of them voted for Obama: even more than black voters who said race didn’t matter. Fewer voters admitted gender was a factor, but men who did were more likely to vote for Obama than those who did not; women who said gender mattered strongly supported Clinton.

But this experiment in asking about race and gender had a different impact in last week’s Mississippi primary. There was almost no gender gap in Mississippi, but the racial divide was enormous. Ninety two percent of African-American voters supported Obama, while just 26 percent of white voters did. Even though nearly four in ten black voters said that race mattered to them, it would have been almost impossible for them to be more pro-Obama than blacks as a whole. However, when whites admitted that race mattered to them (and 24 percent of them did) their votes were more anti-Obama than white voters overall. Only 10 percent of them voted for Obama.

Incidentally, Frankovic's CBS polling unit also just released results from survey questions on Obama and Reverend Wright completed on Sunday and Monday nights. The question numbers in the PDF release imply that the results are part of a much longer survey, one that will presumably continue over the next few evenings, capturing reactions to today's speech (the last several national CBS News surveys had a field period of 4 to 6 days).

-- Mark Blumenthal

March 18, 2008 in Measurement Issues, The 2008 Race

Comments

Post a comment

URL:

Comments: (you may use HTML tags for style)

Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button.

Please email us to report offensive comments.

See our comment policy here. Note that we require commenters to share their email address via Typekey. We will never share your email address with anyone without your explicit permission.