If it was unfamiliar before the titanic primary clash between Hillary Clinton and Barack Obama, the so-called Bradley effect no longer is: It is the disconnect between the higher percentage of people who, in the interest of giving the socially acceptable answer, tell pollsters they will vote for a minority candidate and the (lower) share who, in the privacy of the voting booth, actually do.
Political scientists today who study what's called the "new racism" say, however, that while the underlying stereotypes remain, what's new is how they are expressed (or rather not expressed) publicly. Because social scorn attends public expressions of overt racism, neoracists are more subtle: For instance, they may talk about "welfare spending" and support "school choice" as a substitute for saying "Let's not give public funds to black people"; or they trumpet "family values" as code for expressing their belief that minorities don't have them.
The downside of all this is that the underlying sentiments persist -- although I like to think and truly believe that the share of Americans who harbor such ideas decreases with each day, and as each new generation replaces its parents and grandparents. One in every 15 new marriages in America is mixed race, and the offspring of those marriages are thus mixed-race kids; we can take heart that things can only improve as we move forward.
If there is an upside, however, it is that the Bradley effect is disappearing, as scholars have demonstrated. The reason this is an upside is that there should, in theory, be fewer surprises on Election Day -- fewer disconnects between the final polling numbers and the voting results. (This presumes that the polling methods and sampling are accurate in other ways, of course, which is not always true.) And Democrats, in particular, should take heart in the disappearing Bradley effect this cycle because, well, because the party's presidential candidate is an African-American.
Now, having said all of that, I think we are still in terra nova here. This is the first major-party black presidential nominee, and if there's a residual Bradley effect at work it makes logical sense that it would adversely affect a black Democrat. So I'm not convinced that there will not be a small difference between the final national polls and Obama's total. (Of course, his superior field campaign could neutralize and, thus, mask these effects; indeed, my suspicion is that, relative to final polling numbers, Obama's field campaign may be worth a 2-point bump, and the Bradley effect may account for a 2-point loss.)
The point is this: People who say Obama had better be up by 8 or 10 points in the polls fret too much because the Bradley effect ain't what it used to be.