October 13, 2008
Why the "Bradley Effect" is bullshit
You've probably heard a lot about the "Bradley Effect" and if you haven't you definitely will. The so-called Bradley effect is named after Tom Bradley who was a mayor of Los Angeles that lost an election for governor after pre-election polls had Bradley ahead. In the aftermath some people chalked up the surprise loss to racism.
The theory of Bradley Effect is that a significant number white respondents to political polls told the pollsters that they intended to vote for the black candidate but that racism caused them to actually vote for the white candidate. Obviously with a half-black candidate running for president there's a lot of conjecture about the role of race in this election.
As someone who works with surveys every day I can tell you that there is no such thing as the Bradley Effect. Proponents of the Bradley Effect theory want to impute certain attitudes to voters based on data that was never intended to measure such attitudes. First I'll give you the technical answer and then the common sense answer to why the Bradley Effect is bullshit.
What we know from the Bradley election, and others where the Bradley Effect has supposedly played a part, is that the pre-election polls were wrong. Everything else is purely speculation. What I can tell you is that the most important reasons the political polls get it wrong are sampling errors and faulty turnout models.
The purpose of a poll is to query a small sample of a greater universe in order to correctly predict the behavior of that larger universe. But the universe of voters in a state is not homogeneous. You need to make sure you accurately represent the age, gender, race, party affiliation, etc. in order to have a poll be meaningful.
If I polled 700 Floridians but they were all living in Miami-Dade county it would not be an accurate reflection of the state. I would need to weight my responses based on the actual distribution of voters throughout the state.
The other thing to keep in mind is that word "voters". Just because a certain group represents a certain percentage of the population doesn't mean they represent that percentage of voters. Simply put, not everyone puts the same emphasis on voting. For example, in general we see that older persons are much more consistent about voting. The holy grail for political pollsters is an accurate turnout model.
This is strictly a case of Occam's Razor, the simplest explanation is the most likely. I attribute the incorrectness of the Bradley poll and others where this effect has allegedly taken place to sampling errors and flawed turnout models because frankly these are the culprits for incorrect polling data in most cases where race isn't even an issue (for example two white candidates running against each other).
But I'll leave you with the real reason why the Bradley Effect is bullshit. The polls ask you who you intend to vote for, not why. And even if they did, there's a million plausible reasons a racist respondent could give as to why they were voting for the white candidate if they were embarrassed to admit their racism. There's no reason to tell a pollster he is voting for the black candidate when he knows he's not going to. It just doesn't make sense when he could just say he is voting for the white candidate and leave it at that.
Proponents of the Bradley Effect theory want you to believe that there's a significant number of people out there that are racist enough to vote against a black man they would ordinarily vote for if he was white but too embarrassed to say that they are voting for the white man running against him. I don't know. I'm pretty sure that most racists don't make any bones about their racism and the polls never ask anyway.
Posted by Henry Louis Gomez at October 13, 2008 12:43 AM
Comments
I have respectfully disagree that the "Bradley effect" is total BS. I more than agree that the imputed racism is BS and that even the best of polls are built on data that can be wrong. The reason that the Bradley effect is not total BS is that some number of people will lie to pollsters trying to give the "right" answers. Since polls are perceived to be part of the political establishment and since that establishment is perceived to be unabashedly leftist, the "correct" answer is to vote for the Black candidate. It is the desire for approval from the pollster, even if the pollster has no such bias, that causes the Bradley effect.
I live in California. I remember Bradley. Here in Orange County, we watched him balkanize L.A. He could not have been elected dog catcher. In the Governor's election he got slaughtered here. Yet, if you talked to people before the election, he was favored by about half of Orange County. People didn't want to be seen as racist ( which the media applied to anyone opposed to Bradley ), so they took the easy way out.
Generally, the only racism in the Bradley effect is in those who analyze it.
Posted by: Ken Hahn ![[TypeKey Profile Page]](http://www.babalublog.com/nav-commenters.gif)
at October 13, 2008 01:12 AM
Then what of the hundreds of times that polls are wrong when race isn't a factor? I think it's awfully convenient to pick out a handful of elections and cry racism. There's frankly no evidence for it. Nobody forces anyone to answer a pollsters questions or asks if the reason a respondent is supporting a candidate is race. That's the fact. Trying to read in a racial factor into polls that weren't designed to measure that is a faulty analysis. Period.
Posted by: Henry Louis Gomez at October 13, 2008 01:19 AM
BTW, I could just as credibly argue that supporters of Bradley thought he had it in the bag and just didn't turn out on election day. To me it's just as plausible if not more so based on the low level of political engagement that many of the various base groups that support Democrats have (young people, ethnic minorities, poor people, etc.).
Your anecdote about Orange County is just that, an anecdote. The whole reason to do polling is to have empirical not anecdotal data.
Posted by: Henry Louis Gomez at October 13, 2008 01:22 AM
Have you heard of the "Mulatto Lotto" effect?
Posted by: mulatica at October 13, 2008 02:09 AM
Your argument against the "Bradley Effect" makes sense, Henry. Why would anyone feel compelled to prove they are not racist to a faceless person on the other end of a phone line? Then again, we are talking about liberals here. The same liberals that want everyone to take public transportation and ride their bikes to work to help stop global warming--all from the comfort of their limos and private jets.
Liberals are a screwy bunch. They have this do as I say, not as I do mentality where they want everyone (except them) subjected to taxes, regulations, and inconveniences because it's the "right thing to do." Honestly, I can see a liberal who does not like Obama, for whatever reason, saying they would vote for Obama because they believe that is what everyone else should do. They do not believe they are subject to their own expectations--they just want everyone else to be.
With that said, I agree with you that the polls are seriously flawed and you can see that just by the large variance between different pollsters. Most likely, the "Bradley Effect" comes from these flaws, but I would not discount completely the liberals' penchant for saying what they think should be said and doing the opposite.
Posted by: albertodelacruz at October 13, 2008 07:30 AM
I agree with Henry. The Bradley effect is a product of the 1980's ... not today.
Posted by: Cigar Mike Pancier at October 13, 2008 09:42 AM
Henry the Bradley Effect sounds too much like Social Desirability Bias - where respondents give the response they feel is socially acceptable.
In the case of Obama I feel there may be a number of these.
However a more interesting spin for the Bradley Effect is that since he was leading in the polls before the election, people felt their vote wasn't needed and stayed home. Those lost votes may have cost him the election.
Posted by: La Ventanita at October 13, 2008 10:37 AM
I could see a "social desirability bias" if the survey were specifically about issues such as race. But the fact is that this type of political poll doesn't delve into motivations. There's obviously an infinite number of reasons to vote for or against a candidate and they obviously include race. My big beef with proponents of the theory is that they want to attribute the discrepancy only to racial attitudes. It's impossible to know. What they want you to believe is that there's a group of racists that are SO politically correct that they they'll actually say they will vote for a candidate that they know they won't vote for. It's a stretch in my mind that isn't supported by any real evidence.
Posted by: Henry Louis Gomez at October 13, 2008 11:29 AM
Post a comment
Thanks for signing in, . Now you can comment. (sign out)
(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)
