More Polls

Read Jerome Armstrong, who did some digging into the crosstabs. Essentially, the polls that show a close race are overestimating male turnout ... men haven't been 45% of the primary electorate in any state this cycle (Missouri was closest at 44-56).

I wouldn't call a 10 point win for Hillary Clinton a 'blowout'; the majority of contests so far have ended with 10-point or larger margins for one candidate or the other, and Clinton can in theory get as high as 58% of the vote but end up with only a 4 delegate gain. In practice, though, I think 8-10 delegates is more likely, which of course means Clinton will have to win an even higher percentage of remaining pledged delegates on April 23rd than she did on April 21st.

Comments

[ikl]

The gender gap isn't that huge in PA, so this is only worth about 1% if true. In other words, a lot less that the margin of error. I'm more interested in the regional crosstabs which should give an indication of where CD delegates are in play.

[ikl]

By contrast, the difference in African-American turnout forecasts explains 2% of the difference between ARG and Mason-Dixon. And 15% vs. 19% doesn't account for the full range of possibilities here given turnout patterns in prior primaries.

[Nicholas Beaudrot]

The crosstabs showed that Philly is fine, the Philly suburbs are dead even, "The T" he's down about 37-54.

So I think this means he'll win the 14th, 1st, 2nd, 7th, and maybe the 13th. Clinton will net 2 delegates in the 10th and 3 in the 11th, and 1 each in all the Western PA districts, and everything else is a tie. I think that means Clinton nets 6 delegates, but that feels like I'm lowballing it.

[ikl]

Yeah, that seems optimistic to me. I doubt he wins 13. 7 and 14 are possible but he'd have to have a good night. I am optimistic that Obama keeps 11 at 3-2 with Casey's help. I fear that Murtha's district will go 4-1 for Clinton. Obama might surprise in 10 and keep it 2-2 but I fear that he will only get around 35% there. Optimistically assuming 5-2 and 7-2 splits in Philly, that means that I am projecting a +7 for Clinton in CD delegates. But that assumes big Obama wins in Philly and State College (to keep 5CD at 2-2) and narrow Clinton wins in 14 and 7. So I might revise my guess based on the next wave of polls . . .

[Shock Mouse]

"Read Jerome Armstrong..."

No.

(Regardless, I'd be curious how many other polls right before the primary elections over-estimated male turnout, and if this is the explanation for why Clinton typically outperforms the last-day polls).

[low-tech cyclist]

I followed the link, but I don't see where Jerome discusses the polls' assumptions about M-F turnout.

At any rate, I don't expect single digits. If it happens, great.

[low-tech cyclist]

I see Richard Mellon Scaife's newspaper has endorsed Hillary for the Democratic nomination.

[David W.]

Might there be higher male turnout in urban areas, which could favor Obama?

[low-tech cyclist]

SUSA has Clinton up 50-44 in their final PA poll.

No crosstabs on SUSA's site yet. They generally hold off for a few hours so the paying customers can feel like they got their money's worth.

[Shock Mouse]

PPP's final polls has Obama ahead by 3
http://www.publicpolicypolling.com/pdf/PPP_Release_042108.pdf

Unbelievable I know, but the cross tabs show 58% women and 42% men. Which kinda belies your thesis statement.

[Ron]

My take is the results are only interesting if Obama wins (or maybe even if he loses by 5% or less) or if Clinton wins by 20% or more. If it's the roughly 10% Clinton win many (including me) expect, that won't change anything. Clinton will gain a very small number of delegates and tens of thousands of more popular votes all of which will be wiped out 2 weeks later in North Carolina. This has long since ceased being a state-by-state delegate race. Obama has already won on that score. The only thing left is for Clinton to provide some dramatic evidence of the Obama weakness she has been claiming by winning blowouts or winning states Obama is expected to win (NC, Oregon, Montana, South Dakota). If that doesn't happen (and I don't think it will), the race will end sometime in mid-June when the vast majority of the uncommitted superdelegates endorse Obama.

[Nicholas Beaudrot]

Yeah. And the Quinnipac poll implies something like 43% men, and he's only down 7 there. It looks like it's just MSNBC which had more African-American voters and fewer women.

PPP comes in at the high end of plausible African-American turnout, and it also says he's ahead in South-Central PA, and I don't think anyone believes that.

[ikl]

Based on internals, Clinton should win 14, but Obama may win 7 (best suburban SE Philly CD for him). So I may flip flop on that to agree with Nicholas on CD 7.

CD 11 will be a swing district - it is close to 4-1 Clinton. I'm interested in the SUSA internals. PPP has Obama just under 30% there which is a serious danger zone. A tie in CD10 looks out of reach, which is too bad.

[low-tech cyclist]

PPP has joined ARG and Zogby on my "don't believe this pollster without corroboration" list.

Their radical swings (Obama going from +1 to +21 in NC in a week, Obama going from -26 to +2 in PA in two weeks) say that either the primary electorate is far more volatile than anyone else is measuring, or they're wrong by definition at least half the time.

[low-tech cyclist]

SUSA is estimating a 45-55 M-F split.

[ikl]

PPP changed methodologies between those polls. They were pretty upfront about it. Their poll has been very consistent in PA with the new methodology. Doesn't mean that they are right, but they definately should not be lumped with ARG (which sometimes gets close to the final result but usually takes a strange path to get there - see WI or maybe PA this time).

[low-tech cyclist]

Linky, ikl? I'd read that they changed methodologies between their two most recent PA polls, both of which had Obama leading by 3. But these were in March.

[ikl]

I couldn't verify the PPP thing from their blog but that is my memory from the time. In any case they have now issued 4 polls for PA that show the race within 3%. They have also been pretty consistent with NC. They just about nailed OH and TX and had the best WI poll, I believe. So they could be really off here (and I think that they are off by around 9% or so), but this is not ARG we are talking about here. They come in second to SUSA in Poblano's ratings . . .
http://www.fivethirtyeight.com/2008/03/pollster-ratings-updated.html