Articles and Analysis


Day Before Pennsylvania Sensitivity Update

Topics: Barack Obama , Hillary Clinton , Pollster , Trend lines

The Pennsylvania race has turned slightly toward Clinton over the weekend, with her lead now at an even 6 points in our standard trend estimate. If you believe in taking more chances with random noise, the sensitive estimator has a 6.4 point Clinton lead.

In the rush of new polling over the weekend, it is also good to check how much any of them may be affecting our estimates.

Dropping any single pollster makes only a bit of different to our estimates. The Clinton trend ranges from 48.5% to 49.6%, while Obama ranges from 42.6% to 43.5%. So dropping your least favorite pollster can, at most, account for the difference in a 5 point race and a 7 point one.

And note that we still have about 9 percent undecided. I wonder what they will do?

Cross posted at PoliticalArithmetik.com. See also previous "sensitivity" update for Pennsylvania.


Joseph E:

Remember, if the "undecided" voters turn out along the same lines as the rest of the electorate (that is, 5 for Clinton, 4 for Obama) we will end up with a 7.5% margin in Clinton's favor, according to the polls.

But if we assume instead that the undecided voters are mostly leaning weakly toward Hillary, or that they will settle on the "safe" candidate with more name recognition, then we can assume a 6 to 3 split or even a 7 to 2 split of the undecided voters in Clinton's favor.

This would increase her margin to 9.5% or 10.5%.

I am sticking with my prediction: 11% in Clinton's favor. We will see if these voters turn out for her or not.


Joseph E:

Also recall that dropping PPP (the biggest outlier in Obama's favor) gets us to 49.6 Clinton, 42.6 Obama. With the more sensitive estimator that ought to be 50/42 approximately. (Leading to a 11% Clinton win after undecideds are sorted by my assumptions)

Though to be fair, I should calculate the result with ARG thrown out as an "outlier" on the other side, if I'm going to alter the data.



Hillary, based on today's poll results should do quite well in PA's primary tomorrow, possibly even scoring double digits. This is especially a strong possibility in light of the fact that in the past primary races, she has faired quite well with the undecided vote.



who cares? the "race" is over - nothing left to do except for hillary to go into more debt and waste dems' money. what a joke she is.



Wasn't it generally held that Clinton needed a massive win here in Pennsylvania...not just a moderate one? The talk was that she'd need something in the 60%-40% range, not a bare double digit lead?

I know her camp has been downsizing expectations...which actually shows that the oomph is not there...to win enough delegates, or popular votes to come close to Obama (and the latter in spite of the fact that many of his wins came in Caucuses, where the turnout is suppressed in numbers).



If Clinton can manage a 10+ pt. win (which is seeming more likely if PA undecideds go the way of OH's), especially having been outspent 3 to 1 by Obama, the media will all her the "comeback kid" (again), they'll say the "momentum" has "swung" back to her (again), and they'll widely question why Obama still cannot "close the deal" for the nomination (again). Plus the closer media scrutiny of Obama will continue. Her fundraising will be renewed and it'll launch her into a likely win in IN (which would give her 'bragging rights' since it's Obama's neighboring state). He'll still win NC, of course, so at that point there will be absolutely no reason for her to drop out until all the primaries are complete (in June) and the DNC decides how to seat the delegates from FL and MI (which they'll be forced to do one way or another). She may very well have a slight popular vote advantage when all the voting is complete. That, coupled with the fact that she will have one almost every single big blue/swing state that Dems must carry to win the White House - most by double digits), will be her very best argument to the superdelegates (who aggregately care more about just winning the White House than who the nominee is). This is esp. true if the William Ayers story or any other Obama controversies/gaffes get stirred up in the meantime. Don't count Hillary out yet. Remember, the media unanimously declared Bill Clinton OUT in Spring of '92. And guess who's the only 2 term Democratic president since FDR???



Oh, one more point regarding SUSA though...they have pushed most of their undecided. So that means that there has been a steady trend against Clinton since the debate in the "leaning Clinton" group. I would think SUSA would be a good indicator of where the undecideds of other surveys are "heading". And they are now heading away from Clinton.

There is less wiggle-room in SUSA. If that's the favored "predictor" then there can't be a 60-40 pro-Clinton split. In fact, it could very well be going the other way.

The way I see it is that SUSA provides some measure of differential leaning in that it's surveyors ask the undecideds the query "But if you were voting TODAY, who would you likely vote for?" These would be respondants that are left as undecideds in the other polls (except ARG...which has shown a similar big shift).

To me this suggests that the "Undecideds" have strongly shifted to Obama. This could be disasterous for Clinton if true.



Patrick -


If I cared about you, I'd tell you why. But since I don't......carry on with your delusions....

Maybe someone will come along and refute them......I just don't have the time nor inclination.


Jacob S:

I have a comment about the methodology behind these trends. It seems to me that WHICH polls have just come out tends to skew the trend. For instance, Survey USA and ARG consistently seem to favor Clinton. Both of these pollsters have recently released polls. As a result, the trend seems to be in Clinton's favor. However, if you compare the poll numbers CONDITIONED on pollster, the result is just the opposite. For instance, survey USA's recent poll actually shows that Clinton's lead has DECLINED from 54-40 to 50-44.

Thus, I wonder how the trend line would change if you constructed separate trend lines for polls conducted by each pollster and then aggregated those trends together. Correct me if I am wrong, but my understanding is that the current trend is constructed by evaluating all the polls together as if they are independent observations. Of course, they are not independent observations because they relate to each other in terms of who conducted which polls.


richard pollara:

Cinnamonape: I am not a big believer in the accuracy of the polls but I do believe in historical trends. If you can call what has happened over the last 4 months history then it seems likely that Hillary is still viewed as the safe candidate by the undecideds in Pa. As Mark suggested a couple of days ago the undecideds move in the last few days to safety. Absent some major news or event I would be surprised to see this trend reversed.
Patrick: Although you immediately got trashed, I think your analysis is correct.


Jacob S:

It turns out that my suspicion is at least partially correct. The chart in the post after this one illustrates my point. Moreover, here are the changes in the eight most recent polls:

Suffolk--No previous poll to compare to
PPP--No Change
Strategic Vision--O+2
Survey USA--O+8

So, by conditioning the trend on the pollster of each poll, we see that Clinton's lead has slightly increased in three polls and Obama's lead has increased in three polls (two of which show significant gains). The average of these changes is about 1.7 percent in Obama's favor.

While this evidence does not show a Clinton collapse by any stretch of the imagination, it does give us a different perspective than the raw trend lines. I believe the trend lines seem to be in Clinton's favor due to the Suffolk poll. However, Suffolk has no previous history in Pennsylvania, so it really does not provide us any useful information about the trend in the race.


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.