October 2008

« Obama Ad: 'Steel In His Spine' | The RCP Blog Home Page | McCain Jumps on Biden Gaffe »

Analyzing the 'Bradley Effect'

In looking at the "Bradly Effect" and whether we'll see it this year, Jim Geraghty points to this study by two UC-Irving researchers, ""Ashamed Not to Vote for an African-American; Ashamed to Vote for a Woman: An Analysis of the Bradley Effect from 1982-2006."

Perhaps my colleague Jay Cost could take us through all the mathematics, but, skipping ahead, here is the authors' conclusion:

Those who claim that the Bradley Effect is an artifact of the past are partially correct. ... Our findings suggest that black candidates are truly susceptible to problems with pre-election polling.

Most pre-election polls predicted Barack Obama to win in the 2008 New Hampshire Democratic Presidential Primary. On election night, Hillary Clinton surprised everyone and defeated Obama. This was déjà vu for many political commentators and some began to argue that the Bradley Effect had reappeared. While several political
pundits called this the Bradley Effect, our study provides new insight into this case. Perhaps it was not that the polling was not only overestimating Obama's support, but also underestimating Clinton's. This seems to be consistent with the findings from this election. One of the major findings in this paper seems to be that female candidates are the victims of the Hillary Effect. White women candidates like Hillary Clinton tend to do worse in pre-election polls than in actual elections.