When a poll is taken, it is a social encounter between an interviewer and a respondent. For better or worse, all of the unstated understandings, taboos, survival instincts, and hidden rules that apply in other parts of life come into play. Much of the writing about the “Bradley Effect”(where the polling showed the African American Bradley overwhelming winning the California Governorship, but the actual election results saw him defeated) or the possible “Reverse Bradley Effect” misunderstands or oversimplifies this complexity of this social dynamic. When people respond to pollsters, it's not that they are being dishonest; it's that they are following the rules for conversations with strangers that are standard and widely practiced in the United States.

There are many situations where people talk differently to members of one race than to another. In his speech on race, Barack Obama mentioned “That anger may not get expressed in public, in front of white co-workers or white friends.” In fact, it is safe to say that on certain topics, conversations differ depending on the racial identities of those involved.

Recently my students and I looked at this issue in the context of topics not related to voting using the General Social Survey. ^{Data for this article use the following variables from the General Social Survey: INTRACE1, RACE, IMMAMECO and SEXEDUC.
Davis, James Allan and Smith, Tom W. General social surveys, 1972-2006 [machine-readable data file] /Principal Investigator, James A. Davis; Director and Co-Principal Investigator, Tom W. Smith; Co-Principal Investigator, Peter V. Marsden; Sponsored by National Science Foundation. --NORC ed.-- Chicago: National Opinion Research Center [producer]; Storrs, CT: The Roper Center for Public Opinion Research, University of Connecticut [distributor], 2007.
1 data file (51,020 logical records) + 1 codebook (2,552 pp.). -- (National Data Program for the Social Sciences, no. 18).}This survey is an excellent general interest resource that you can analyze online using the facilities of NORC (National Opinion Research Center) to identify patterns in responses and how those patterns are related to the race of the interviewer. My students and I used the data from 2004.

For example we looked at this question:

There are different opinions about immigrants from other countries living in America. (By "immigrants" we mean people who come to settle in America.)

How much do you agree or disagree with each of the following statements?
Immigrants are generally good for America's economy

 Note the almost 8% difference when a White interviewer asks the question "whether immigrants are good for the economy" compared to when a Black interviewer asks the same question.  (For example with a White interviewer – 37.5% of the overall respondents say they agree. With a Black interviewer, 45.6 percent of the overall respondents say they agree.)

Sure enough, people tend to respond to Black interviewers somewhat differently than they do to white interviewers. What makes it even more complicated is that in the case of this question, the impact is more pronounced for Blacks than for Whites.

Here's the table of answers to the same question from White respondents only.

There is about a 6 percent drop in the percent of Whites expressing no opinion when the interviewer is Black compared to when the interviewer is White.

And here is the table for Black respondents only:

For Blacks, there is a whalloping  15 point drop in expressing no opinion when the interviewer is Black, from 40% to 25%.

So, what happens? When speaking to White interviewers, people are more likely to say they neither agree nor disagree, but this tendency is especially pronounced for Black respondents. But both Blacks and Whites tend to respond more supportively toward immigrants when speaking to a Black interviewer. Which is the situation in which they are being “honest”? We don't know.

On some controversial topics, things get even more complicated. For example, this question:

Would you be for or against sex education in the public schools?

Here we see just a small difference overall between the overall responses to Black and White interviewers.

The same overall pattern applies to White respondents as we can see in the table below.

For Blacks though, there is a meaningful difference. Blacks are less likely (by roughly 6 %) to tell White interviewers that they favor sex education. Again, do we know in which situation the respondents are being “honest”? No.

We looked at other examples too, including topics such as religion and government spending, because the General Social Survey offers a lot of different questions to choose from. In examining the data, sometimes it is Whites who switch more; sometimes it is Blacks; sometimes there is no real impact.

So, is there a “Bradley effect” or “reverse Bradley effect”? It's not so simple, but yes, of course race shapes all kinds of social encounters in the United States, and it would be wrong to think that polling is any different.

Discuss this article.

About the Author: Bronxdem is a professor of sociology and a long-time contributor and researcher for ePluribus Media.

ePluribus Media Contributors: cho, Bronxdem's students.

 

Photo credits: © 2008 Winston Davidian, reprinted with permission from IStockphoto

 

If you like what ePMedia's been doing with research, reviews and interviews, please consider donating to help with our efforts.