Articles and Analysis


What's Up in Pennsylvania?

Topics: 2008 , Barack Obama , Divergent Polls , Hillary Clinton , Pollster , SurveyUSA

I will admit that, like Pollster reader jsh1120, I am at a bit of a loss about the flurry of recent results from Pennsylvania. In the last week, we have seen surveys released showing everything from an 18-point Clinton lead to a 2-point Obama advantage (for all links, see our Pennsylvania chart and table). Of course, we have had days before this primary season where we saw huge spreads among pollsters in their Democratic primary results.

As before, the most likely explanation involves differences in the kinds of people pollsters are selecting as likely primary voters. In California, Texas and other states with very large minority populations (either African American, Latino or both), the variation in racial composition explained much of the difference. In Pennsylvania, however, the percentage of black voters is relatively low and the Latino population in the low single digits. As such, the mix of gender, age and socio-economic status may help explain the divergent results before us. Unfortunately, only a few pollsters routinely release their composition statistics and most are not asking respondents about their education or income levels. So we are largely left to speculate about the differences.

One way to help clarify the numbers, if only slightly, is to focus on the results among white voters for the recent polls that have tracked and released results by race over the last month. Doing so (see the table below) at least eliminates the differences due to variations in racial composition.


The polls certainly differ, but in many ways, their results are consistent. All four show Clinton with roughly 60% of the white vote in early March. All show modest increase in support (+3 to +7 points) for Barack Obama in recent weeks. They diverge mostly on the Clinton trend. Three of the four show a relatively modest decline in support for Hillary Clinton, while SurveyUSA shows no change at all for Clinton and the smallest increase (+3 points) for Obama.

I am hoping to post more on Pennsylvania -- focusing on the new Quinnipiac University data by race and education -- later today.



Thanks for this analysis!

Don't you think SurveyUSA showing AA making up 14% of the electorate is a bit low considering the massive registration drive organized by the Obama campaign focusing on black and young voters. This passage from the politico is telling:

Another high-stakes voter registration drive just concluded in Pennsylvania, where the deadline to register as a Democrat and participate in the primary was March 24. The Pennsylvania Department of State reports that more than 234,000 voters have either newly registered as Democrats or switched from other parties, and the state hasn't finished counting the new registrations.
"We put together a massive effort," said Hildebrand, saying that the numbers include "over 200k Obama supporters" - an impressive number, and likely more than 10 percent of the total turnout in the primary.

Include over 200k Obama supporters?? I supposed a lot of AA and Young people?? so what do you think Mark?

Link: http://www.politico.com/news/stories/0408/9328.html



Mark - thanks for the quick analysis. That's something I've been looking at, myself. With the polls showing distinct and separate results - what it comes down to is comparing apples with apples. Showing trends within the same pollster does exactly that.



depending on what stats you consider, aa percentage of pa pop is around 9.
here's a quick scan of the cities: and even though philly is almost at half, that does not throw off the over all percentages.

Rank African American Percentage
1. Philadelphia 43%
2. Pittsburgh 27%
3. Allentown 8%
4. Erie 14%
5. Reading 12%
6. Scranton 3%
7. Bethlehem 4%
8. Lancaster 14%
9. Levittown 2%
10. Harrisburg 55%

btw: don't put your trust in college kids... or maybe, you should. they are unreliable and even so, they are not turning out in droves typical for obama. it is much closer.



Sometimes the most simple explanation really is the correct one. Perhaps this poll of C +18 and the PPP poll of O +2 are simply outliers (or rather, the SUSA poll is an outlier and PPP just isn't very good). Eliminate those 2 results, and we are left with the last 2 weeks of polling showing leads for Clinton of 11, 9, 6, 5, 2, and 0--a fairly typical spread for a state primary race that is just heating up in terms of advertising, campaigning, etc.

Now, this is not to say that we should discount the SUSA poll, as too often people dismiss a poll they don't like as an outlier. This could just be the beginning of a trend back towards Clinton in the state now that she is buying more air time than she had been, but given how far off this poll is from other recent results, we should wait for some validation before putting too much stock in it just because SUSA has been the best this primary season. SUSA is still subject to the same 95% confidence as everyone else, right?



Sorry - off topic.

What's up with the Bush approval chart on page 1 of the Pollster.com site? It has not been updated in over a month.

I sent an email to the "contact us" address - questions@pollster.com but it came back as an invalid address.



Thanks for the mention, Mark. I'll treasure it. (Though since it only references my lack of clarity, I may think twice about that.) In any event, your comments are well taken. Though we don't have the data, I'm suspecting that much of the variation may have to do with minority (and perhaps "new voter") representation in the Pennsylvania polls. My guess is that frames for "likely voters" may not have become any easier to determine or for the various organizations to reach consensus about.



Re PA:

Those recent polls with a higher number of undecideds in general are showing a closer race. (Though that does not account for all the difference.)

Could it be that this is reflecting a fluidity in the decision making process, and when pressed undecideds are moving to Clinton. (Which would also suggest that here numbers are softer.)

Mark, one way to help answer that is to look at how the question is asked, and how they deal with leaners. Do you know?




Ismackc said:"btw: don't put your trust in college kids... or maybe, you should. they are unreliable and even so, they are not turning out in droves typical for obama. it is much closer. "

I think that you'll see that student (or at least young adult 18-24 year olds; or 18-29 year olds depending upon the exit survey) are setting records for any Primary campaign. They are consistently turning out in higher numbers than has ever been recorded, often double or triple any prior Primary election.

In addition their %age of the total electorate is up when compared to the total voting electorate. They have usually been about 10-14% of the total Democratic Primary composition but have been about 20% average this campaign.

What many people forget is that the sample for this age cohort is smaller than other cohorts. These usually have 5, 10, or more years added to their scalar. Thus they should be expected to comprise a larger %ile of the voting public than 18-24 year or 18-28 year old classes.

In any case, the number of younger, college age individuals are significantly up this year...even if this is obscured somewhat by somewhat similarly large turnouts in other demographics. Still, when thoise other pools are up dramatically, and the youngsters still make substantative increases thats not something to pooh-pooh!



I am a volunteer with the Obama campaign. I make phone calls, I wear Obama gear and am stopped on the street for conversations because of it, I politic people I meet in my daily life.

I know that pollsters make an attempt to cover "all registered voters" or "likely Democratic voters", etc. but their sample size is often less than 2000 people, sometimes less than 1000.

In my little corner of the world (Chester County, PA) -- I personally talk to at least 100 people a week about the election. In working voter registration drives the last 3 weekends before the cut-off date, that number was far higher. So over the past month, I have spoken to at least 1000 people.

So here's my take -- first, usually when I talk to people about the election, I ask if they have been polled this year. I haven't found anyone yet who says "yes".

The people I meet are certainly random, as they are just going shopping in a strip mall (voter registration), or are on a phone list of registered Democrats, but the person answering the phone might be an Independent or Republican spouse or family member.

My take is that Hillary's support, while strong amoung about 20% of the people I speak with, is mushy amoungst most of her supporters. They are wavering BECAUSE they think that Obama might be the eventual candidate, and they want to get on to the D vs. R fight instead of the C vs. O contest. They are unnerved by Tuzla and that sense of "we should have believed the bad things about the Clintons back in the 90's". A lot of them WANT to support a woman, but do not want it to be "this" woman.

I discount the enthusiasm from the Obama supporters I speak with -- they often say they know I'm busy -- get back to changing minds, not talking to the dedicated, and it may be that their support is not as strong as I'd like to believe it is, but I hope it is!

So that's my take -- my "personal polling" indicates that Hillary will win PA by 7 - 8 points, of which 4 are directly related to having the entire "machine" behind her, but that the delegate count won't buy her much because of the congressional district split.


On the Poll:


I am an Obama supporter too, but in the interest of accurate polling, I have to take issue with the thrust of your post.

You equate your discussions with around a 1,000 people with the 1,000 people sample size of typical polls. Big mistake.

First, your time period is a month. Polls typically have to take place in a matter of days to get an accurate snapshot of people's attitudes.

But more importantly--and a key issue for accuracy--is the manner in which you are meeting people.

You say yourself that you are an Obama volunteer and you wear Obama gear. You may feel that your discussions are random, but you are self selecting voters who are more likely to be in agreement with you.

Having worked on many campaigns and worn plenty of candidate gear, I can tell you that you are more likely to have supporters of the candidate who's button you are wearing come up to you and give you a good word or pat on the back while opponents are likely to keep quiet and keep walking.

Plus, being in Chester County, I assume you are talking mostly with people in that region of the state. Again not a representation of the entire commonwealth.

Pollsters go thru great lengths to ensure true random samples. You can read the details of how those sample are determined thru these pages.



On the Poll --

I understand what you're saying. I really do. And while I certainly agree that "Chester County" is not random, I read the polls -- A LOT.

And when you have two "random" polls that come out on the same day with one saying Obama plus 1 and the other saying Clinton plus 8, I think, HHHMMM.

Certainly, I understand that the pool can be different, the questions can be different, and there ARE certain pollsters who "trend" a certain way.

And while polls take "a snapshot", what I am interested in is WHO IS GOING TO WIN, because like everybody else, I'd love to be able to see into the future.

But I think my interactions are more random than you think. Back in 2000, while wearing Gore gear, I had people curse at me for being "so wrong". In 2006, wearing a Rendell button, I had Santorum supporters tell me I was going to rot in hell, and Rendell wasn't even running against Rick.

This year, I've spoken to McCain supporters who were reregistering as Democrats to vote for Clinton JUST to keep the race alive, and McCain supporters who hate Democrats, and Clinton supporters who tried to convert me.

But whether you polled me on February 5th, or April 9th, my position didn't change. A LOT of people are like me.

I admit it can be a "7 blind men and the elephant" phenomenon. And certainly, because of where I live, I meet far more of the Obama demographic of high income/college educated.

But we'll see -- I put out my number of a 7 - 8 point Clinton win. I hope I'm wrong because I want Obama to win by one vote -- but come the night of the 22nd, we'll know.....




What you need to realize is that while the AA% in Pennsylvania is @10%...the more critical element is what share of the DEMOCRATIC constituency do they consist of...especially in the cities/areas you list.

In places like Levittown the number of AA's may be very low...but so may the number of Democrats. So just rattling off a list isn't very useful. To precisely predict the number of delegates influenced by AA voters one needs to estimate the proportion of AA voters, the number of those individuals likely to vote in the Democratic primary and make an estimate of AA turnout in those districts.


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.