Inspired by this thread, but more narrowly focused. The Bradley Effect is the supposed tendency of white American voters, in a race pitting a white vs. an African-American candidate, to vote against the black candidate even after telling the pollsters they intend to vote for him/her. How real is this, how important, and how, if at all, can it help us interpret poll results in the current presidential race?
I doubt there is any way to really quantify this. I think many white Americans have a sort of latent racism embedded in them. Latent in that they do not recognize it in themselves and/or will not admit to it.
I actually see this in my 78 y.o. mother. Over the years I have spotted a distinct racism in her language and views. It is not overt but it is there (such as referring to a person as “black” even when such information is actually irrelevant to the discussion…she does not say someone is “white” in the same context). A few times I have pointed this out to her and she gets very angry. She vehemently denies any racism on her part and I honestly believe that intellectually she wants to believe that. It is more a subtle, in-built trait of hers she herself does not even recognize.
At the end of the day no one knows what lever you pull in the polling booth and yet you can come out and say you pulled the lever for Obama to protect your self image as presented to the world as someone not affected by racism.
I suppose exit polls compared to actual results may shed some light on this effect but it would still be hard to say. But if 90% of the people leaving the polls say they voted Obama and he only gets 40% of the vote that would certainly be telling that either the polling methodology is seriously screwed up or we are seeing the Bradley Effect.
It’s not useful as a tool to predict the outcome of an election, but neither are polls where the candidates are within a few percentage points of each other. I’m sure that the bradley effect exists, but pointing it out and talking about it sounds an awful lot like accusing the people who are voting for the other guy of racism.
I find it is highly overrated. If there was an affect Obama would have performed worse in the primaries than polling would have suggested. There is no evidence that this happened. In fact, I believe on average, he outperformed the polls.
I think the lack of an effect is caused by two main factors. First off the people who wouldn’t vote for Obama because he is black are generally not people who would vote for a democrat anyway. Secondly, I think there are plenty of who are able to disassociate their feelings on blacks in general from their opinion of Obama specifically. A lot of racism (or any other isms) is based on stereotypes. Once you get to know a person it is difficult to maintain those racisms. Thus, people think blacks behavior in a certain matter, but know that Obama doesn’t, so they are comfortable voting for him, while not necessarily changing their opinions on blacks in general. I believe an inconsequentially small number of voters will tell pollsters that they are voting for Obama and not follow through due to the Bradley effect.
I’m pretty skeptical about the polling in general this year, not only because of the Bradley Effect, but also because of what appears to be underpolling of black and younger white voters and because of the “likely voter” determination. A couple of weeks ago on the Diane Rehmshow, two pollsters got into a heated discussion about polling methodology, with David Moore (former Sr. Editor at Gallup) casting a lot of doubt on the ways most polls define and weigh the likely voter (link is to an audio page – no free transcripts as far as I can tell). Moore also argued against the practice of pushing for a definite answer rather than allowing voters to claim “Undecided”. ISTM that Moore was pretty skeptical himself of most of the polls that are being done today.
Regarding the Bradley Effect specifically, the widepsread used of computerized polling (replacing human-conducted polls) might dampen the effect. It appears that people are more likely to tell the truth to a computer rather than a human, at least in some circumstances. I have a textbook on HR practices which quotes a few studies of varying ages (1969, 1989, and 1995) that claim that people are more likely to disclose information that might reflect negatively on them when the questions are asked by a computer. The studies found that people reported their GPA, desire to stay with the company, etc., more honestly to a computer than to a live person. These studies looked at job interviews conducted at a computer, not at political polls done by phone, but maybe the same effect would work in both cases. If so, the Bradley Effect might not be quite as strong this year.
Ultimately, of course, I have no freakin’ clue. As an Obama supporter, I get nervous about the polls showing a close race. But I just don’t have a lot of confidence in any of these numbers, one way or the other.
One refreshing thing about Southern racism is that it’s not especially covert. Most southerners vote Republican anyway even when it’s two white guys and even when it’s a white northern Republican v. a white southern Democrat, but those who wouldn’t vote for Obama because he’s black, while they might shield it in political speech (I don’t really know what he stands for…) never claim to be voting for him to begin with.
I wonder if the “Bradley Effect” can be explained a little differently?
Perhaps there are alot of Dems and Moderates who so much want to elect a black person to a high office to make themselves feel better about the state of race relations… but when it comes down to it, just can’t rationalize it.
I believe that’s how Obama rose to the top in the first place… and I’m sure that’s how a defeat will be explained in November.
Eh? You’re describing a process which, hypothetically, would have caused him to lose the nomination.
I think there might be a bit of a “reverse Bradley effect” this time around, particularly in areas like mine where racism is pretty common and out in the open. A lot of people want to and will vote for Obama but don’t want to be especially open about it.
I don’t think it will be strong enough to counter the actual Bradley effect, but I don’t think either will be big enough to make much difference.
I didn’t have a name for it, but this idea crossed my mind on Labor Day weekend. I went home to Pittsburgh and in talking to a bunch of my friends (mostly middle-class and working-class white guys…the classic Reagan Democrats) a lot of them said that they thought Obama was going to lose, especially because when white Pennsylvania voters get into the voting booth, they’re not going to vote for a black guy.
Frankly, I’m really afraid that it’s all too true and I’m sure some of those friends are going to vote McCain or not vote for just that reason. And please don’t tell me I need to get better friends.
Here is a discussion from Nate Silver at 538.com that spells out the primary results and explains why he considers the Bradley Effect to be an outdated idea. He also references the “reverse Bradley effect” discussed above, but applies it to the fact that polls are tending to show something like 80-20 splits in AA votes for Obama when primary results showed something much closer to 90-10 (and above).
http://www.fivethirtyeight.com/2008/08/persistent-myth-of-bradley-effect.html
The key point is that the Bradley effect only correlates to covert racism - those voters that claim they’ll vote for a black candidate and then don’t. Overt racists don’t count because they are plenty happy to tell pollsters that they won’t vote for a black man. Also, it doesn’t apply to exit polls which is what many people are quoting when they claim that Obama underperformed his poll numbers in the primaries.
I think that by the time the election rolls around, pretty much everyone who just doesn’t trust a black guy to be President will have found some other reason to to mistrust him. I mean, if you’ve been looking for some bullshit to blame your McCain vote on in case anyone asks, you’ve probably found it by now. I think whatever racism there is in this election either already is, or by election day will be, reflected in the polls. I’m not sure what the elections where the Bradley Effect was observed were, but I doubt that they parallel this one much more deeply than the whole “white guy vs. black guy” part.
Then again, I suppose the race politics of this election simply have no useful precedent, so all bets are off.
Win or lose… I’m describing a process by which a voter would, as you put it, “vote against the black candidate even after telling the pollsters they intend to vote for him/her.”
But did you read the numbers in my link above that indicated that this most certainly did not happen in the primary? That is, there were not significant numbers of voters that voted against Obama after indicating they would vote for him. In fact, in the average, more voters voted for him after indicating they would not. The only region that Obama over-polled in was the Northeast, of which only NH should be competitive in the general election.
Perhaps because the most liberal Dems will have no qualms about voting for Obama and proudly (ie saying so to pollsters).
The test will be in the general when the former HRC supporters tell the pollsters that they support Obama but decide to vote for the abundantly more experienced candidate in the booth.
I remember a Newsweek article at the time Andrew Young ran for governor of Georgia and Doug Wilder ran for governor of Virginia. Pollsters for both campaigns figured they needed a 5-6% cushion in their polling numbers to account for folks who said they favored the black candidate but would never in a million years vote for him once the voting-booth curtains were closed. That was almost 20 years ago; I hope we’ve grown up a bit as a nation since then.
But, he did win the nomination. Why would you expect the general election to play out differently?
ETA: Never mind, covered in post #15.
I just looked up the 2006 senate race in Tennessee. Just before the election MSNBC polled Corker at 50% and Ford at 38%. The actual vote was Corker 51% and 48% for Ford.
Just one example, but Bradley was just one case as well. I think the effect is dwindling in general, and in this election which has had so much mud slinging for so long anyone not wanting to vote for Obama would have plenty of non-racial reasons to use for cover.
Well, there are other examples used, notably Douglas Wilder and David Dinkins, where their lead almost, but not entirely, disappeared overnight before the election. And four years later, when Dinkins lost, the polling was accurate. But I don’t think I’ve seen any clear indication of the phenomenon since the mid-nineties.
Here in Massachusetts, Deval Patrick won by about a 21 point margin, and the only polls I’ve been able to Google are a Zogby Interactive poll that had him up by 24 (well within ZI’s margin of error in the 2006 gubernatorial races, by the looks of it), and a poll from early October that had a much smaller lead. And if there was race calling out for the Bradley effect, that was it. By election day, Kerry Healey was so unpopular* that you kinda did look like a racist if you admitted to voting for her. Well, that might be a bit much, but she did a pretty good job of hiding whatever appeal she might have to voters.
- My favorite Kerry Healey campaign idea: Having supporters dressed as inmates carrying signs reading "Prisoners for Deval, parading in front of Deval Patrick’s house while his young son was home alone. Really kinda skipped a few steps beyond “effective negative ads,” past “obvious smear job,” and landed in “mentally unhinged nastiness.”
The memory of that fine campaign still brings a nostalgic tear to my eye. 