I’ve heard a fair amount of chatter recently — especially among my acquaintances of an, um, older persuasion — about the possibility of the Bradley Effect costing Obama the presidency. For the uninformed, the Bradley Effect refers to the 1982 California Gubernatorial election during which Los Angeles Mayor Tom Bradley, an African-American, lost to Deputy Attorney General George Deukmejian, a whitey (Armenian!), despite a solid lead in the polls. The theory states basically that poll respondents falsely stated their support for Bradley in an effort to avoid appearing racist, but within the private confines of the booth, voted for the white dude. The applicability to the current race is obvious, but is there real cause for concern? Not really.
There are two basic and sound reasons to believe there will be Bradley Effect come November 4th (or even as early voting commences now). The first theory, basically, is that the Bradley Effect never actually existed. Read more here, but according to insiders of both 1982 campaigns, internal polling actually showed that Bradley was going to lose and expectations otherwise resulted from polling error. In addition, exit polls failed to account for absentee ballots. In reality, much of the swing could be traced to a shift in strategy within the Deukmejian campaign which produced a methodical closing of polling data. Of course, this doesn’t negate the role race may have played in the election, but it’s not entirely clear the Bradley Effect lost Tom Bradley the election.
The second theory is that the Bradley Effect did exist throughout the 1980s and into the mid 1990s, but that it no longer persists. A paper published through Harvard corroborates the dissipation, and the existence of a Bradley Effect, at least in several elections during the aforementioned time frame, is supported by this Pew Research report. While academics are not particularly sure why the Bradley Effect has disappeared, the evidence supporting this general conclusion is generally considered strong.
Now, Bradley Effect prognosticators in this election cycle tend to point to Barack Obama’s loss in New Hampshire despite having led in the polls by an average of 6.3 points. But, as Nate Silver argues, this myopic view gets the story wrong. In proper context, the variance between polling predictions and actual performance in New Hampshire (8.9 points) represented only the seventh-largest polling discrepancy of the entire primary season. Larger inaccuracies were: Iowa (9.8), South Carolina (14.3), Alabama (15.6), Georgia (21.4), Wisconsin (10.3), and Mississippi (9.1). The statistically undue attention focused on New Hampshire, in large part, can be attributed to the fact that Hillary’s upset changed the dynamics of the primary.
Finally, as with the explanation of Tom Bradley’s loss, none of this means that race will not factor in the election. It simply means that there’s no reason to fear an Obama loss will result from an accounted for cohort of closet racists.