Pollster.com

Articles and Analysis

 

POLL: PPP Pennsylvania Dems 4/19-20


Public Policy Polling (D)
April 19-20, n=2,338 likely Democratic primary voters

Pennsylvania
Obama 49, Clinton 46
Release, results

 

Comments
Mike_in_CA:

So basically its a duel between the more "traditional" Quinnipiac-ish polls or the "new model" PPP types.

Quinnipiac has a very good track record, and PA is common ground for these pollsters -- i.e. it's in the Northeast, and it seems they would have a good handle on how to poll it.

That said, its hard to gauge how large a percentage of the population is 18-29 in the Quinnipiac poll (in the Suffolk poll its in single digits, highly unlikely!), but I think herein lies the rub.

PPP has 18-29 demo's at 16%, and Obama wins this group handily. It seems they've adopted a model that emphasizes young people. Also, they have Obama winning the "Philly" area by a hair more than Quinnipiac.

When I say "traditional" pollsters I mean cellphone-less. I've heard a lot of reports that say much of Obama's support in PA comes from these young cellphone users.

That said, it appears that Obama's vote % could be as low as 45%, but it could be much higher, depending on the GOTV effort targeted at young people and the Philly area.

I guess we'll all have to wait just ONE more day (Thankfully!)

____________________

This poll is pure junk, as is ARG's C+13. I predict Clinton will win by 3%-10%, with a median call of 6%. It's the obvious call given the tight 5, 6, 7 polling spread we're seeing on most of the polls (notably Mason-Dixon, SUSA, and Quinnipiac). We may have to revise our prediction of a 12 delegate gain for Clinton when we see SUSA's crosstabs.

____________________

damitajo1:

PPP has had very low performance.....

____________________

richard pollara:

Goes to ARG's comment that if you don't like a result trash the poll. In my opinion none of the pollsters have figured out a way to accurately predict likely voters in a Primary. The best insight that I have seen is from Mark's analysis of the undecided. The last few days they tend to break for the safer choice. In CA. NH, OH, and Texas they broke over the weekend for Mrs. Clinton. I expect to see the same thing in Pa. If the spread of 6 is accurate, then the late breakers will move the margin to 10+. But if we have learned anything over the last 4 months (only 4 months!!!) it is that polling is as much art as science.

____________________

PPP's record:

OHIO (actual: 54C, 44O)
PPP (D) 3/1-2/08 1112LV 51C 42O 7U
PPP (D) 2/23-24/08 600LV 50C 46O 4U

TEXAS (actual: 51C, 47O)
PPP (D) 3/1-2/08 755LV 50C 44O 6U
PPP (D) 2/23-24/08 434LV 48C 48O 4U

SC (actual: 27C, 18E, 57O)
PPP (D) 1/24/2008 595LV 24C 19E 44O

WISCONSIN (actual: 41C, 58O)
PPP (D) 2/16-17/08 822LV 40C 53O 7U
PPP (D) 2/11/08 642LV 39C 50O 10U

____________________

jsh1120:

As long as everyone has some result to support an opinion, I'll toss in my admittedly "split the difference" prediction. Looks like a 7-8% victory for Clinton, I think, if the "undecideds" break for Clinton and the "new voters" turn out, both of which looks likely.

____________________

Tulle:

What do you base the PPP low performance on? If you check other primaries you will see them very different from other polls a few weeks out but thier last one before the vote is not bad at all. With that said, it stlll looks like Clinton by 8 to 12% to me.

____________________

damitajo1:

Sean - picking 4 polls out of 40 is hardly a demonstration of PPP's record. Try this: http://www.fivethirtyeight.com/2008/03/pollster-ratings-updated.html

____________________

Uri:

PA has a huge student population that is all for Obama (just tonight, he'll have a huge rally at the university of Pittsburgh basketball stadium, for example). I'm an HRC supporter, but can definitely see this poll as being the accurate one, unfortunately.

____________________

desirous:

Just because a poll is an outlier does not make the poll bad. PPP has performed well when it is the outlier. They were onto something in SC and WI; they could be onto something here.

____________________

Uri:

BTW, How does PPP have more than double the sample size than everyone else?

____________________

Tulle:

damitajo1, I went to the url you posted and it says SUSA was the best and PPP the second best. So what can I make of that in this case ... not much. I'll stick with my silly 8 to 12% win for Clinton even the the second best pollister shows Obama winning.

____________________

seanh:

damitajo1,

Honestly, I was just pulling pollster.com's available polls focusing on highly contentious primary contests. I looked for results in places like MA, CA, IA, NH, but found none listed from PPP. Results from those controversial states might have been more revealing. My selection was far from comprehensive, probably suspiciously favorable, but does provide some context for PPP's latest numbers. If you have more polls from PPP that cast some doubt on their record, please post away; I'd be genuinely interested to the other side of the story.

I'm no polling expert myself, but overall result aside, PPP's PA crosstabs certainly do look a little out of whack. 50C 41O with women, and 48C 42O with 65+?

____________________

damitajo1:

Survey USA is more consistently on to something: http://tpmcafe.talkingpointsmemo.com/talk/2008/04/surveyusa-track-record.php

____________________

desirous:

"I'm no polling expert myself, but overall result aside, PPP's PA crosstabs certainly do look a little out of whack. 50C 41O with women, and 48C 42O with 65+?"

Either it's whack, which may be the case, or it's a Wisconsin scenario where Obama overperforms with a relatively flat age curve.

____________________

Rasmus:

Damitajo1:
538 is a great site, but the pollster ratings there are for General Election, not for Primaries.
For General-Election results, PPP polls are "good but not great" indicators- but that does not mean PPP can�t be an extraordinary good primary pollster (I do not think they are better than SUSA- but when you refer to 538, you should read Poblanos analysis about the likely voter screen and how it influences the poll results- great stuff, I think.)

____________________

Joseph E:

Reposted (Sorry if it is out of date; I was referring to the early post above:


Sorry, I don?t see how you are getting a 6% margin if you consider PPP
"pure junk". They got their result by using a tight likely voter screen and
some assumptions about turnout that happen to be favorable to Obama (I
don?t think they are have any evil motives). But many of the other
pollsters that show lower numbers for Clinton also use tight likely voter
screens.

SUSA, which has done well this season and in the past, does much less
screening and massaging of the data. They have shown a much larger spread
for Clinton (though see the latest poll from today, which shows a somewhat
tighter race: http://www.nbc10.com/politics/15943088/detail.html )

If we give Clinton a few more votes from the undecided folks who show up,
6% seems far too low. It could happen, but I think a double-digit spread is
much more likely.

(I see Richard has said the same thing)

____________________

Alan:

Rasmus,
538.com's ratings actually do include results from primaries. Poblano says this in his explanation of the ratings: "I now have a database of over 150 competitive contests since 2000. This includes essentially every competitive presidential race, and most competitive primary races, Senate races, and Governor races."

That said, this isn't simply a matter of picking the results from the best couple of pollsters.

____________________

killias2:

Am I the only one confused about why damitajo1 trashed PPP's polls then, to back that up, gave us a link showing PPP to have the second best polls around?

Honestly, I thought they weren't that great either, until you showed me that. After some quick research, I'm actually pretty happy with PPP's record.

However, I think this is a slight statistical outlier, and that it doesn't show where the undecideds break. 8 point win for Clinton.

____________________



Post a comment




Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.

MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR