Reader Posts
« previous | TPM CAFÉ READER POSTS HOME | next »
The "Bradley effect" debunked by Indiana and NC
Every time an African American candidate for political office receives fewer votes than were predicted by pre-election polls, the poll errors are explained (in the media) by the "Bradley effect", which is an urban myth that white voters are reluctant to admit to pollsters that they are unwilling to vote for an African American candidate.
But there never seems to be any comment among these so-wise media types when an African American candidate outperforms pre-election polls.
In Indiana, pre-election polls had Clinton ahead by at least 4 points. She won by 2. In North Carolina, pre-election polls had Obama ahead by single digits only. He won by 14.
Can we finally lay to rest the idiotic, sensationalist, divisive, Rovian concept of the "Bradley effect"?
Comments (18)
Actually, she won Indiana by 1.4 points.
May 10, 2008 11:45 PM | Reply | Permalink
Thanks, right you are - and that with a significant assist from Rush Lumbaugh's morally bankrupt campaign to sabotage the Democratic primaries.
May 10, 2008 11:50 PM | Reply | Permalink
nope, actually .8 percent, 11,000 votes
May 11, 2008 12:41 AM | Reply | Permalink
Really? I'm happy to be corrected. That's awesome.
May 11, 2008 1:02 AM | Reply | Permalink
You beat me to it, articleman.
Unlike Guam which is very tiny, in IN, a huge state, which was perceived as hers, she barely held her own. That is one of the reasons why IN is such a story.
May 11, 2008 8:58 PM | Reply | Permalink
The average of the poll 3 days before the primary had Hillary up by 4 in IN and Obama up by 9 in NC.
May 10, 2008 11:56 PM | Reply | Permalink
I don't understand the concept of this concept, which is a description of statistics, being divisive. Eugene Robinson used it the night of NH. He's not Rovian.
If the poster thinks that polls pre 2008 in elections involving blacks did not either overstate or correctly state the support for the black candidate, the poster is wrong.
Obama has altered reality positively. Sometimes the polls meaningfully underpredict his performance. Almost always, however, in states with very substantial black populations.
In this cycle, both blacks and younger, landlineless voters have been undersampled, especially in reasonably young states, so the models are all fucked up this year.
The Bradley effect seems to have less general validity now, which is great and I share your desire to move beyond it. But acting like it never happened is wishful, we move beyond it by changing reality, not attacking this description of it. SC primary was the first meaningful instance known to me in our history where the black candidate substantially outperformed the polls. But SC is half-black in the Dem electorate.
These are important issues bearing a lot of scrutiny.
May 11, 2008 12:46 AM | Reply | Permalink
Agreed, maybe the "Bradley effect" has been operative in some past elections, and maybe Obama has transcended it.
My question is how well proven the effect ever was in past elections. It is a divisive concept to spread all over the media if it was never really true. Maybe 99.9% of whites are honest with pollsters, and inaccurate pre-election polls have been due to all the other factors that make polls inaccurate. To invent an explanation for inaccurate polls is unscientific, and to invent an explanation that accuses people of racism they don't actually feel is divisive and counterproductive.
Racism has been and remains pervasive in our society, but we should discourage the media from exaggerating it. I cling in hope to the stories of small towns in the deep South that have elected black mayors, who perform like all other mayors, and after a while the whites in town seem to forget they even have a black mayor - they just see him as a regular guy. Maybe that can be the outcome of this election, that Obama can transcend race on the national stage.
May 11, 2008 7:34 AM | Reply | Permalink
You bring a very good point. Whethere it is a new thing or not.
Now, can someone notify Hillary's campaign and the Media??
May 11, 2008 9:51 AM | Reply | Permalink
I think a lot of white voters were willing to vote for an African American until they became acquainted with Jeremiah Wright. Of course it's easier to see the world in black and white, which is to say that whites won't vote for blacks. But the truth lies somewhere in between. A lot depends on who the African American candidate is, and who he associates with.
May 11, 2008 4:00 PM | Reply | Permalink
Yes, so we will see what happens in Nov.
Will you be voting Dem in Nov, Otto, regardless of nominee?
May 11, 2008 8:59 PM | Reply | Permalink
Otto previously said he would vote for Obama.
May 11, 2008 11:33 PM | Reply | Permalink
White voters in NC were able to see the Wright thing as a media fueled detraction and it rightfully was rejected as a factor in NC. I am proud that NC took the high road while many have used the Wright propaganda as a convenience excuse.
May 11, 2008 11:14 PM | Reply | Permalink
The Bradley effect works both ways. Some AA voters lie and say they're going to vote for the white candidate.
I agree that in the Clinton/Obama battle in NC, we saw much less of a Bradley effect--compared to say Gant-Helms. However the polls in NC were off because they didn't estimate the vote correctly. AA turnout was higher than normal, and the pollsters haven't been able to forecast liklihood of voting correctly anywhere (except for Wisconsin).
May 11, 2008 6:51 PM | Reply | Permalink
By the way, here is a little more meat for discussion:
http://en.wikipedia.org/wiki/Bradley_effect
May 11, 2008 9:00 PM | Reply | Permalink
Thanks for the reference, nice to know there has been some actual research on the subject. Whatever the reason, Obama's campaign seems to have transcended the effect in Indiana and NC.
Maybe the explanation is suggested by what one congressional superdelegate said the day after these two primaries: "Where are all these 20-somethings coming from?!?"
May 11, 2008 9:38 PM | Reply | Permalink
One factor that has confounded some of the pollsters is the inability to predict the newly-registered voter behavior, because so much of polling samples are based upon previous behavior. Newly registered voters don't have previous behavior, and these new voters have been breaking for Obama, cancelling out any possible 'Bradley effect.' I have seen discussed in more than one place the 'Reverse Bradley effect' which is that AA voters actually underreport in the polls and end up voting for the black candidate at higher rates than the polls indicate.
May 11, 2008 10:36 PM | Reply | Permalink
Some of this, especially the reverse effect, is mentioned in the wiki article above.
May 11, 2008 10:39 PM | Reply | Permalink
Post a Comment