THE BRADLEY EXCUSE
It's named after Tom Bradley the successful African-American mayor of L. A. who ran for California governor in the '80s when polls showed him with a victory but he ended up losing. Some studies concluded that a small percentage of white voters lied to pollsters about not voting for Bradley for fear of appearing racist or prejudiced against him.
Subsequent studies of other similar cases, as well as contradictory cases, where black candidates either lost where polls said they wouldn't, or vice versa, have concluded that either there is a "Bradley Effect" or there isn't. So the science behind this supposition that a certain amount of white voters will get into the voting booth and be unable or unwilling to pull the lever for a black candidate they told pollsters they were voting for, or did vote for, is very iffy,
But notice how the more responsible seeming rightwing Republican commentators (like say William Bennet on CNN) will refer to it often lately, and how Democratic and liberal commentators will too often go along with it, assuming that there has to be a fair amount of prejudice still out there against voting for a partly African-American candidate.
But many studies show that percentage to be really small, almost insignificant in polling terms, and most prejudiced voters to be pretty outspoken with their opinions (there are many yotube videos showing interviews with supporters of McCain and even more so of Palin flat out admitting they won't vote for a "colored" candidate, or just saying that Obama is a Muslim terrorist), but these are not found in any significant numbers in Democratic districts or even in most "swin districts that polls show going for Obama this time.
But one study has projected a possible six percent difference from the "Bradley Effect"—meaning Obama could go into the election with anywhere around a six percent lead and lose and this study will be a ready excuse.
Which is why the rightwing is pushing it so much, and too many in the media are repeating it. Because if the rightwing Republicans attempts to rig the vote is successful—through purging voting roles in Democratic districts, or closely contested districts, of legitimate voters as has been proven they are doing in several states already (their success at keeping the media and the FBI focused on "voter fraud" has led to less attention focused on their voter purges, disqaulifying many legitimate voters who are registered Democrats, or tricking them into voting Republican through rigged or "faulty" voting machines, as has already been demonstrated in early voting in many states where voting machines are owned by a company run by Republicans, including rightwing Republican activists (who have proven since Nixon's dirty tricks team and Lee Atwater's deathbed confessions and Rove's blatant tactics that they are capable of any kind of illegal manipulation of actual votes and voters in order to gain and maintain power) who claim the "faulty" machines were just "calibrated wrong" (which means when you push the Obama rectangle it registers for McCain)—they can claim any percentage difference of ten or less is a result of "the Bradley Effect" and too many in the media and even among Democrats, will accept it (as they too easily did the Florida debacle in 2000 after the phony Rove directed "protest rally" that turned into an invasion and pressure on the vote counters, etc., and as they did the voting machine problems in Ohio that showed great disparity between exit polls and what the machines reported).
The only bright side to such a scneario would be if the Democrats are able to secure a large enough majority in the Senate this time (60 or more) to make it impossible for Republicans to block their votes and actions. Then they could hold investigations to expose and prosecute such vote manipulation. But if the rightwing Republican dirty tricks work, the Democrats may not be successful in capturing 60 seats in the Senate either.
Hopefully the Obama lawyers and poll observers are more ready to contest and fight any McCain win that contradicts entrance and exit polls, which until Florida 2000 had proved to be correct in every previous election.