Some nervous Democrats, watching as Barack Obama's lead in the polls slowly grows, may be concerned that a lead of five points or so may not be enough, thanks to the notorious "Bradley Effect"--the phenomenon, named for Los Angeles Mayor Tom Bradley's losing gubernatorial campaign in 1982, whereby white voters lie to pollsters about their willingness to vote for an African-American candidate.

They should probably relax. FiveThirtyEight's Nate Silver, who's long argued that the Bradley Effect seems to have expired in recent years, offers a recapitulation of his arguments (now reinforced by Bradley's own pollster from 1982), along with speculation about why this has happened.

All along, Silver's argument has been very simply that during the primary season, Barack Obama generally outperformed his poll standings, which shouldn't have happened if the Bradley Effect was operating. In retrospect, it's clear that much of the talk about the Bradley Effect was spurred by the one primary state, New Hampshire, where Obama narrowly lost despite pollster predictions that he was ahead. Absent any fresh evidence--and the ability of pollsters to "push" respondents for honest answers should have produced some by now if it existed--it's time for Democrats to stop worrying that racist voters will revert to type in the privacy of the voting booth. Racism, of course, still exists, and may be hurting Obama, but not in the sort of secretive, poll-refuting manner that is suggested by the Bradley Effect. As Silver points out, there are plenty of more socially acceptable reasons voters could offer for deciding to vote against Barack Obama, no matter what's actually going on in their hearts and minds.