Serious question for Bradley Effect believers

[J. J.]

The one question with Bradley Effect is "Who is effectively lying to the pollster."  It could be that people that are more likely to vote for a Republican candidate are more likely to lie to a pollster in a race where one candidate is black.

That said, I really expected three things:

1.  McCain would underpoll nationally by 1-2 points (I think he more greatly underpolled, but not by much).

2.  The underpolling would not be even state to state (It wasn't).

3.  While Obama would carry PA, McCain would underpoll strongly here (on that I was dead wrong).  PA and IA were the two stated that stunned me in terms of under/over polling.

[J. J.]

The one question with Bradley Effect is "Who is effectively lying to the pollster."  It could be that people that are more likely to vote for a Republican candidate are more likely to lie to a pollster in a race where one candidate is black.

That said, I really expected three things:

1.  McCain would underpoll nationally by 1-2 points (I think he more greatly underpolled, but not by much).

2.  The underpolling would not be even state to state (It wasn't).

3.  While Obama would carry PA, McCain would underpoll strongly here (on that I was dead wrong).  PA and IA were the two stated that stunned me in terms of under/over polling.

It is kind of cool the way the "underpoll" effect largely erased the Bradley Effect, except of course where there was no Bradley Effect, the location of the perimeters of which may have been nowhere, somewhere, or everywhere, depending on the scope and reach of the "underpoll" counter riptide. It is sort of like positive and negative mass; the relationship between the two is hard to describe.

And there you have it. 

Actually, the underpolling is the Bradley Effect, though it has diminished over time.  Nationally it looked like 2-3 points.  I was expecting 1-2 points.

Two points: 

1.  It occurred in placed where I didn't expect it (IA) and didn't occur in places where I expected it (PA). 

2.  Obama underpolled (meaning he did better in voting that the polls showed) in states with a high Mexican descent population.  (Now that isn't exactly good for the GOP.)

[J. J.]

1.  It occurred in placed where I didn't expect it (IA) and didn't occur in places where I expected it (PA). 

And this is why everyone laughs at you. Why would it happen in Iowa? Why was race such a big factor in Iowa but not in other states with histories of racial tension?

If you have ask about "racial tension" then you are just too stupid to understand.  It has to do with the person being polled thinking, "If I this answer, [whatever "this answer" is], the pollster will think I'm racist.  I don't want the pollster to think I'm racist, so I won't this answer."

You must be logged in to read this quote.

We didn't have this in the last election, not to this extent.  Now, isn't a giant effect, and it's diminished over the years, but it was still present in 2008.

[J. J.]

WHY THE HELL IS EVERY CASE OF UNDERPOLLING AUTOMATICALLY THE GODDAMN BRADLEY EFFECT YOU HACK???

I'd still like an answer from J.J.

Because it seems to be more prevalent in cases where one candidate is black.

Generally, a good polling firm will get it right (okay, within the MOE).  We don't seem to be getting this with two white candidates.

[J. J.]

OK so why is this more likely to happen in Iowa than other states? There has to be a reason. You basically seem to think the Bradley Effect is because of some Polling God throwing darts at a map and then cursing each state that he hits with it. It doesn't happen that way. Unless one can explain why Iowans would be more likely to lie for racial reasons, it seems more likely that something else was at work as Lunar has explained.

That I don't know, and as I've said it was a surprise.   So was PA, where Obama underpolled.  So was MN where Obama overpolled.

Yeah because polling has never been off in an election involving white candidates before. And of course there never was any poll off in Obama's favor (Nate Silver pointed out that the error in many southern states in Obama's favor was greater than the polling error in favor of Hillary in New Hampshire.)

Please forget primary races.  Did we have these kind of errors in the last presidential election?  Not that I recall.

And, we saw this in the national polls as well.

There is one other factor as well.  There was a pattern to Obama's undercounting.  He did better than he polled in three states with a high Mexican origin population, NM, NV, and CO.