Friday, June 19, 2009

Not That You're Reading Too Much into the PA Senate Polling, but...

I take issue with some of the "wide lead" talk concerning Arlen Specter's position in the Democratic primary polling relative to Joe Sestak. This isn't a new development: that I have an issue or that the media is talking up the numbers without digging terribly deeply into them. And for the record, Political Wire is technically right. It is a wide lead.

But is that what we should be focused on at this point in the race?

The margin isn't what matters. At this point, Specter's position in the polls relative to the 50% mark is what's important. And the Republican-turned-Democrat is hovering just over that point currently. The other thing to eye is the fluctuation in the level of undecideds in this race. That number is important because of a few things that are likely to keep the number higher [than they would be minus these factors]. First, this race involves a Republican-turned-Democrat. Secondly, Sestak has not "officially" entered the race. And finally, it is very early in the process.

So early in fact, that polling wasn't conducted nearly so soon in the cycle the last time an incumbent Pennsylvania senator was challenged in a primary. And for that information you have to stretch all the way back to 2004 when a political unknown, Arlen Specter, was challenged in the Republican primary by Pat Toomey. What pattern can we glean from that data?

First of all, polling on the Specter/Toomey race did not begin until the fall of 2003 before the April 2004 primary. Polling in May and June of 2009, then, precedes that point in the senate electoral cycle. The starting point is largely the same for the candidates in the polls, though. You can see the trendline here (see "Matchup Poll Graph" on the right side). But what OurCampaign provides is the polling without verification of the sources and without that undecided number. So let's look at the polling data and a better graphic of the trends from the fall of 2003 through primary day in Pennsylvania in late April of 2004.

The thing is that Specter jumped above the 50% mark in a few polls but for the most part was stuck just under 50% throughout. All the movement, not to mention momentum, was with Toomey across the five months of polling in the campaign. The more undecideds decided, the more Toomey gained on Specter among likely (Republican) voters in the closed Pennsylvania primary.

[Click to Enlarge]

If we contrast that with the average Pollster has for the six polls conducted in the last month and a half on this hypothetical Democratic primary race, we see that Sestak has already cut further into Specter's advantage without having even formally announced his intention to run. The 17 point advantage Specter now holds is more than half of what it was in the week after his switch to the Democratic party and all the Sestak talk began (The average of the three polls conducted during the first week in May had Specter up by 41 points.). The kicker is that that is with less than ten points having been cut off the undecideds value (The average undecided mark in those same three polls mentioned above was 21 points with the latest Rasmussen poll showing 13% undecided). In other words, Sestak is taking away from Specter more than he's picking up undecideds.

And it's still early (for polling in this race and for the levelling of wide lead charges).

Recent Posts:
State of the Race: Virginia (6/18/09)

How Not to Emerge as a 2012 GOP Darkhorse

A Week Later, Deeds Still Leads, but...


Robert said...

Very convincing, particularly when you present the data with Toomey. However, usually when there are undecideds and a candidate-to-be-named later, the undecideds break more evenly. It is easier to be wary of someone when they could be running against an unknown candidate or someone who has not officially announced. Relatively unknown candidates have trouble getting traction. Obama did it against Clinton in 2007/2008, but Fred Thompson did not do so well.

Jack said...

Well, Obama was a much better candidate than the former senator from Tennessee.

Josh Putnam said...

...but Thompson was/is an actor!

Thanks for the vote of confidence, Rob. On many levels, this is an apples and oranges comparison (2010 to 2004), but there are some parallels.

The obvious generalizable relationship here is that the polling will tighten as election day approaches (all things being equal and sans economic collapses). You also have several other factors to control for. I've already alluded to a couple. Specter's imperiled this time in a completely different way. He's confronted with having voters of a different party get used to him (That's why that "flight risk" comment was ingenious.). And also that Sestak isn't even formally in the race yet.

But you two have hit on something that I've left off (ah, peer review): candidate quality. It is more than being unknown; it's what you bring to the table. Both Sestak and Toomey are/were in the House, so they are on a level playing field in terms of how political scientists usually define quality (past offices held).

The bottom line is that Sestak is already in a better position now than Toomey was in March of 2004. And truth be told, much of that is due to Specter's switch raising awareness of the race.

Jack said...

On a slightly unrelated note, I'm sure the Democrats would be much happier if it was Torsella rather than Sestak in the race. This way there would be a primary challenger to keep Specter in line without risking a House seat. (Why did I think Torsella was a former congressman?)