So what happened to the Bradley Effect?

[Stranger in a strange land]

I'll admit that I expected there would be at least a small Bradley Effect. Yet it failed to materialize. I didn't think it would be anything like 4, 5, or 7 (lol) points, or that it would appear in places like Iowa, but quite a few people in the media, and many on this forum, speculated, sometimes wildly. It came to be one of the last, best hopes of conservatives, and acquired a near mythic reputation. But yet, on election day, it didn't appear.

Why? Was it all a myth caused by external factors and bad polling? Did it exist in the 80s, but die out as crime, crack, and welfare ceased to be hot-button issues and the older whites most susceptible to it died out? Was it caused by bad polling? In the original Bradley Effect elections, polling was far less frequent and far less precise. Did pollsters figure out a way to measure or negate it? Does it still exist, but was overwhelmed by Obama's superior turnout operation and spectacular campaign? Is it just that, when the economy is this bad, race seems a whole lot less important? Something else? I'm interested to hear your thoughts.

[??????????]

Obama was far enough ahead in the polls for it not to be noticed. Go look at county maps around the south.

[Jacobtm]

The most surprising thing in the world happened.

Pollsters were generally accurate. IN, NC, MO were all as close as can be imagined. VA was slightly closer than polls predicted, a 4 point win rather than 6 or so, but PA was a few points better for Obama than polls projected. So maybe we just chalk those up to MOE?

[MR maverick]

I believe a reverse took place in Ohio on McCain.

[Lief 🗽]

No Bradley effect at all. It's dead. Buried. Let's please never talk about it again.

[MR maverick]

No Bradley effect at all. It's dead. Buried. Let's please never talk about it again.

Maybe so, but I was shocked how Ohio went right away for Obama.

Somebody was lieing to the pollsters because McCain had a lead in that state for most of the election year.

[Хahar 🤔]

No Bradley effect at all. It's dead. Buried. Let's please never talk about it again.

[Lunar]

It's what I've been saying all along, it is just one minor reason why the polls can be wrong.  Methodological errors, other polling bias (cell-only, weighting schemes), and ground game are all more reliable predictors for polling inaccuracy than a knee-jerk, inane, stupid devotion to the Bradley Effect.  Lying to pollsters for any reason is a serious concern and a source of error, but to just "assume" that the actual result is about X more Republican than the results show just because Obama is on the flip side of the ticket is absurd and defies logic.

This is proof how bias interrupts analytical legitimacy.

[Marokai Backbeat]

It may have existed years and years ago, but it is gone now, gone for good. It's possible it never really existed it all, but I'm leaning towards it's died out before now and existed at one point.

[○∙◄☻¥tπ[╪AV┼cVê└]

He didn't do as well as the polls, but that might end up being in the MOE. He was up 7.8 points in the polls, and he's currently up 5.9 points in the returns, and I expect the final results to be over 6, maybe 6.5 or even 7 points.

Obama ranges from 50 to 55, with an average of 52.1
McCain ranges from 42.7 to 46 with an average of 44.3

[Gustaf]

Too many options in this poll. I'm not sure whether it existed or not. There are so many variables involved and polling is sufficiently inprecise that these things are very hard to determine. We still don't have enough data to say much on the issue in a general perspective.

It seems clear though that generally Obama has not done worse than the polls indicated. Given how many polls and elections we've seen at this point, primaries+general, I'm ready to say that with certainty. This obviously indicates that if there was a Bradley effect it was either very small or negated by some counter-force. I don't really know of any such counter-force, so I will assume that it didn't exist for Obama to any great degree.

Whether it ever existed I don't really know for sure. But I have the impression that the 80s was an era where racism was still quite wide-spread but had become unacceptable to voice in public which provides the foundation of a Bradley effect. I don't think there are as many closet racists anymore and more importantly I don't think many of them see themselves as Democrats anymore (which is a factor I suspect would be an important component in the Bradley effect).

Overall, I lean towards saying that it did exist but doesn't anymore.

[J. J.]

Both Gallup, which until yesterday was the polling "gold standard," and ABC/WP called it +11 for Obama, nationally.  I'm now describing both polls as Zogbyesque.  It looks like it was a 1-3 point B.E., which was a little higher than I was suggesting.  That is well out of the MOE.

It did not occur on Rasmussen, that uses robo-calling.

I'm really asking some questions about how Hotline polled.  They were close.

I also want to look at some state polls

[Stranger in a strange land]

Both Gallup, which until yesterday was the polling "gold standard," and ABC/WP called it +11 for Obama, nationally.  I'm now describing both polls as Zogbyesque.  It looks like it was a 1-3 point B.E., which was a little higher than I was suggesting.  That is well out of the MOE.

It did not occur on Rasmussen, that uses robo-calling.

I'm really asking some questions about how Hotline polled.  They were close.

I also want to look at some state polls

or it could be that both made the same error in methodology. The Bradley Effect doesn't seem to have shown up at all in state polling.

[Silent Hunter]

Other.

The white racists just decided to stay home, the polls were more accurate.

There's something else too- Obama is not your typical African-American. He's only half-black, he's Ivy League-educated, he was raised by a white woman (my thoughts are with her family at this time) and he wasn't Jesse Jackson.

He was, basically, a "non-threatening" black guy. If he looked, well, like Dennis Haysbert, things would have been different.

[platypeanArchcow]

But look at Arkansas, Louisiana, West Virginia.  Did anyone expect McCain to win Arkansas by 20 points?  Obama managed to avoid the question of race in places where he tried using the "ground game."  Convincing people that it was OK to vote for him, because everyone was doing it.  Where there wasn't a campaign, there was racial voting -- and there was, I think, somewhat of a Bradley effect, or the related fact of racists staying home.

[© tweed]

it probably never existed.  it certainly cannot be conclusively proven to have ever existed.

[J. J.]

Both Gallup, which until yesterday was the polling "gold standard," and ABC/WP called it +11 for Obama, nationally.  I'm now describing both polls as Zogbyesque.  It looks like it was a 1-3 point B.E., which was a little higher than I was suggesting.  That is well out of the MOE.

It did not occur on Rasmussen, that uses robo-calling.

I'm really asking some questions about how Hotline polled.  They were close.

I also want to look at some state polls

or it could be that both made the same error in methodology. The Bradley Effect doesn't seem to have shown up at all in state polling.

Those are two different firms, either of which is Zogby (Zogby did the same thing), but it's possible.  I want to look at the state polling, but I can't get on the polling pages.  A +6 gap is much different than a +11 gap (in polls that have an MOE of +2).

I really would like to know more about Hotline's methods and weighting.  It looked like it occurred in 3 out of 4 races in 2006 for about 2-4 black candidate overpolling.

[Smid]

Another thing to consider is given the large number of first time voters supporting Obama, and given that often a question pollsters use to determine "likely voters" is "did you vote last election" it's very possible that there was a Bradley Effect that overestimated Obama's vote but also an anti-Obama bias with the exclusion of some first-time voters who turned up. It's possible that the two offset the other.