Silly Polls

Did MSNBC really get crosstabs of Pennsylvania voters who bowl? Gun ownership I understand, but bowling? Why not ask who watches Grey's Anatomy?

Of course regular bowlers are probably more likely to be white, non-urban, and working class. So I'm not sure it tells us anything interesting. In general it appears that the undecided voters are more likely to be Clinton voters, but Obama is probably going to pull in the 39% that he needs outside of the metro areas in order to avoid a delegate blowout.

Comments

[low-tech cyclist]

Ah, the depths to which our media have sunk! Bowling, indeed.

Anyway, it seems to me that the main thing is how firm that 48-43 is, because Hillary would have to swing 3/4 of the undecideds just to pull it up to 55-45.

And all 55-45 gets her is enough of a victory to have a reason to keep going in the face of increasingly steep odds, since PA is really her last chance to make a major dent in Obama's leads in pledged delegates and total votes.

IMHO, the ranges look like this:

1) Hillary loses: even Hillary realizes it's game over.
2) Hillary wins by 7% or less: I don't know what Hillary does, but the "pack it in, already!" chorus rises to a crescendo.
3) Hillary wins by 8-13%: Hillary keeps going, the "pack it in" crowd stays quiet, and by Thursday or Friday, the main story on the Dem side is Hillary's next-to-impossible uphill battle.
4) Hillary needs to win by 14% or more to get positive momentum out of PA. She'd better hope SUSA's right, and everyone else is wrong.

I expect I could be off by a percentage point or two, but that's my best guess of how it'll play.

I think it has way less to do with expectations than (a) the grinding reality of the delegate arithmetic, and (b) in view of that arithmetic, what it would take to get wavering superdelegates to commit, wavering donors to contribute, and wavering voters in IN and NC to consider her in a more positive light.

I think 57-43 in PA would do at least some of those things. Any less, and I really doubt it.

[ikl]

This poll projects African-Americans at 19% of voters on Tuesday. That is not impossible, but is at the very high end of the possible range. So I don't think that this polls is as good for Obama as it looks at first glance.

[Nicholas Beaudrot]

ikl: I don't understand this one.

African-Americans are 10.7% of the state's population.
The state is 50% Democratic.
Effectively all African-Americans are Democrats.
Doesn't it follow that African-Americans will make up roughly 20% of the voters? It seems they've been just under twice their percentage of the population in the most recent primaries.

[mijnheer]

Speaking of guns: It is a sad commentary on society that Hillary Clinton felt she could attract voters by announcing that she likes to kill animals for fun. Kerry did the same in 2004. To hell with both of them.

[ikl]

Nicholas, I would have thought so too, but the data doesn't suggest that this is usually the case:

http://www.fivethirtyeight.com/search/label/african-americans

[Nicholas Beaudrot]

I think some allowance needs to be made for the fact that Pennsylvania is a closed primary ... I don't think it will get as high as 22%, but 18-19% seems within reach.

Also the fact that the African-American population is geographically concentrated makes the turnout operation really easy.

[ikl]

These factors would all seem to apply to NY, and CT where African-American turnout was really low. NY might be discounted on the grounds that Obama didn't have a real ground game there. But CT? MD turnout was better but a lot less than 200%. NJ is the only nearby state on the high end of the range and even there it was only 160%.

I asked Poblano about the open primary issue - it seems that it isn't statistically significant. Poblano's model predicts 17.8% for PA, so 19% isn't unrealistic, but it is on the high end.

It might just be a bad sample, but the 2004 exit poll for PA had African-Americans significantly overrepresented at 13% when they are under 11% of the state's population.