The Bradley Non-Effect

I have had several recent discussions with friends and relatives about the so-called "Bradley Effect"; that is, the under-reporting in polls of voters who will vote against a certain candidate-- in this case, Barack Obama-- because they're racists but do not want to admit it to a pollster. Typically, these people will respond "undecided" to a poll rather than say they're voting against someone, but once the "curtain falls" in the voting booth (do any of them still have curtains by the way?), they'll vote their prejudice. Most of those who have approached me have real fear of the vigor of the Effect and its implications for the current race for president of the United States. Does Obama leading in the polls really mean that he is going to win? Does he have to have a 10-point lead in order to be considered "ahead?"

While no one really knows what, if any, Bradley Effect there will be on this race, most of what I have read shows that there is no longer any effective Bradley Effect in play. I say "effective" because I think that there may very well be some of that input into the polls, but it seems to have decreased according to many studies at some point in the mid-90s, presumably as an older generation became itself marginalized and as the civil rights era of the 1950s and 1960s became more distant in time. I write as a "young" 35 year-old who grew up in the South, incidentally, someone to whom, while I was growing up, the various race riots, sit-ins, speeches, protests, both peaceful and violent, and segregation in cities across America was something that I read about in history books and seemed scarcely real, or at least no more or less impactive than the Civil War. (Of course, over time and reading and meeting many people who lived in that era, I have personally come to know more about the struggles of the age and also to be struck by the parallels between the struggle for equal rights for African Americans and women and gay/TG rights. But I digress.) There are a number of factors as to why the Bradley Effect seems to have disappeared, faded away, or even become a "reverse" Bradley Effect.

(Incidentally, the Bradley Effect was named for Tom Bradley, a black man running for governor of California in 1982; while the polls seemed to show him with a significant lead prior to the election, he nevertheless lost).

First, and I think I have already made this argument in a certain way already, persons under about 45 don't really have the same built-in prejudices or exposures to institutional prejudice that those older than that age do. They just don't have the same "us versus them" kind of context and the whole of history to deal with. We grew up with equal rights a matter of fact and segregation as something obviously wrong. Now, there are always exceptions, but I think that it is apparent and clear that society has decisively changed for the better in this respect. If nothing else demonstrates this fact, we all seem to have that grandfather or grandmother who spouts off those utterly embarrassing racist declarations every once in a while-- but note that this is your grandmother and grandfather, not your mother or father (again, with exceptions, but the exceptions prove the rule here I believe).

Second, empirical evidence suggests that the reverse Bradley effect is greater than the Bradley effect. The strongest evidence of a Bradley Effect hurting Barack Obama would have been found in the primaries. Analysis shows that while there was some Bradley effect in certain states, mostly Appalachian states like West Virginia and Kentucky, but also New Hampshire, the reverse Bradley Effect-- that is, people voting for Obama because of his ethnicity for whatever reason or motivation-- was greater; more states were affected, including all of the South and most of the Northeast. If there is a Bradley Effect, you would have expected it to show up in the Democratic primaries, where people who nominally should have supported Obama just wouldn't-- with the addition of Republicans, people who mostly won't vote for him anyway, the impact of racism is minimized. There is not much of an effect if a bunch of hardcore Republican racists do not vote for Obama, because they would not have voted for him anyway.

Third, and I think this has been overlooked a bit, most polls now are done via computer. That is, a person responding to a poll does not have the same kind of peer-pressure to respond a certain way to a question. Who cares if the computer is disgusted with your reason why?

There is more to say about this topic, but I refer my readers to the post of Nate Silver, an excellent statistician and analyzer of polls, to cover the more statistical details and points.