In the 1980s, Los Angeles Mayor Tom Bradley twice ran for California governor. The polls showed he would win; he lost both times. That is, the polls consistently overestimated how many people actually did vote for him. The generally accepted explanation for this phenomenon: Tom Bradley was an African-American.
A significant proportion of potential voters tell pollsters they will vote for a black candidate when they know they will do no such thing. This bias has even carried over into the exit polls, leading news media to predict wins for black candidates who ultimately lose. What isn't clear is the extent to which this Bradley Effect reflects actual racism or alternatively, a fear of unfairly being seen as racist.
So much for the history lesson, now let us examine the Obama phenomenon over the past couple of weeks. In the Iowa caucuses where Democrats have to stand up in the presence of their like-minded neighbors and declare their preference for a candidate, Obama won. In the privacy of the New Hampshire voting booths, Obama lost. Why? The Bradley Effect is a prime suspect. Given that most of the primary decisions henceforth are votes rather than caucuses, expect a Clinton victory. At the very least, discount poll results showing support for Obama.
