Memo to all pundits analyzing the New Hampshire returns: Barack Obama’s surprise loss should not be attributed to "the Bradley Effect."
In 1982, Tom Bradley, then the mayor of Los Angeles, seemed poised to win election as California’s first black governor, building a solid lead in pre-election polls, only to inexplicably lose on Election Day to Republican George Deukmejian, a verdict that was chalked up to the unwillignness of many white voters to side with a black candidate in the privacy of the voting booth.” Since the final pre-primary polls showed Barack Obama comfortably ahead of Hillary Clinton—and in light of the predominance of white voters in the state—some pundits are now suggesting that the Bradley Effect reared its head in New Hampshire last night. The Washington Post‘s Eugene Robinson even talked up the possibility on MSNBC last night.
But the data does not support this theory. Hillary Clinton’s unexpected win can be attributed almost entirely to women voters. In the final primary polls before the primary, she and Obama ran virtually even among women. But in the primary itself, Hillary won among women by nearly 15 points, accounting for her overall two-point victory. And there is a reasonable explanation for why women flocked to her in the end: her much-publicized tearing up episode on the primary’s eve, which seems to have reinforced among women (especially in light of Saturday night’s debate) the idea that she was being ganged up on and unfairly targeted by her opponents and the press.
Moreover, this same phenomenon did not happen in Iowa, an equally white-dominated state where Obama and Clinton ran even among women on caucus night–thus accounting for Obama’s win in that state.
To believe that the Bradley Effect played a role last night, you’d have to believe that Obama’s race gave pause only to a narrow group of voters—late-deciding women—and that it only affected this group’s thinking in New Hampshire, and not at all in Iowa. And you’d have to ignore a very plausible and reasonable alternate explanation—Hillary’s extraordinary, open display of emotion—for why these women broke against Obama at the very last minute. On top of that, you’d have to explain why the Bradley Effect didn’t materialize among independent voters in Tennessee in 2006—where African-American Senate candidate Harold Ford’s actual vote total was no worse (and in fact, slightly better) than his showing in the final pre-election polls—but did among Democratic-leaning women in New Hampshire in 2008. It just doesn’t make that much sense.