Cross-posted from Lehigh Valley Independent.
Last month, Scott Kraus over at the Morning Call's Pennsylvania Ave. posted a new poll from GOP Pollster The Tarrance Group that has Charlie Dent (R) leading John Callahan (D) 53-27 about 10 months out from the PA-15 election.
Like many of us, I've been excitedly waiting for a poll to come out on this race -- and have ended up unfortunately underwhelmed.
I guess from having a math degree, I'm a big stickler for polls. And when they don't meet the right criteria for what I think is considered reputable polling, I get marginally frustrated.
First, while it's not exactly uncommon, I don't approve of such a relatively small sample size. The Tarrance Group polled 300 out of the estimated 818,000 people that live in PA-15. For those of you keeping track at home, that's less than 0.04% of the district. So, roughly, based on their numbers, they found 159 people who support Dent (0.02% of PA-15) as opposed to 81 people who support Callahan (0.01% of PA-15).
A 300-person sample size is not outrageously small for a congressional poll, but it certainly is not practical if you want to be reliable. The sample size could also potentially be part of the reason as to why this poll has such a substantially large margin of error, at +/- 6%.
Secondly, they won't release their internals. For those of you that don't following polling too closely, internals are all of the breakdowns of how and hopefully why the pollster got the responses they did. They usually include a breakdown of the regions, questions asked, people sampled, etc. All they did were release their best GOP bullet points.
What I am always most interested to see in internals is: How many Democrats do they think are going to turn up? How many Republicans? How many Independents? Obviously, these predictions shape your poll fundamentally -- especially if you're basing it off of "likely voters," because your poll has to determine who is likely to vote.
Third, I do think that Republican and Democratic pollsters add insight to certain races, but only if they're averaged out with each other and, ideally, against a more weighted non-bias poll. As Ken Petrini notes, "pollsters always seem to skew their polls in favor of whom they are being paid by," which is why I do feel that they can sometimes offer a helpful skewed perspective, but only when they're evened out with other established polls.
In that regard, the Tarrance Group does seem to have a good policy to sometimes average themselves out in needed situations, but unfortunately it doesn't seem like they used it for this poll. From their website:
When circumstances require both a Republican and Democrat perspective, The Tarrance Group often teams up with the Democrat polling firm of Lake Snell Perry & Associates to conduct bi-partisan research.
It's safe to say that I would take this poll much more seriously if Tarrance teamed up with Lake to poll our district.
The good news for Charlie Dent in this polling is that he's above 50%, which is obviously where you need to be. As I've obviously shown, I'm skeptical about this poll until they release their internals, but any time you're above 50% is a good one. However, because of unknown factors in Tarrance's polling, they did end up with a very high margin of error, at +/- 6%, which means that Dent could be as low as 47% -- and that's obviously not where he wants to be.
The bad news for Charlie Dent in this polling is the shockingly high presence of Independent candidate Jake Towne in the race. Tarrance has him pegged at receiving 8% of the vote, which is extremely surprising for an Independent candidate this early on with very low name recognition. That means that out of only 300 people, 24 people not only knew who Towne was, but said that they would vote for him. If those numbers held up, and PA-15 had a 15% turnout, that could mean that Towne would already yield more than 9,800 votes, without having to gain any more name recognition.
One thing that I would love to know, if Tarrance released their internals, would be the percentage of the sample that were familiar with Jake Towne against the percentage of the sample that committed to vote for him.
In this poll, Mr. Towne's numbers were certainly the ones that jumped off the page at me today. Will Dent and Towne end up fighting for the right? According to this poll, Dent has 90% name recognition and Towne's is understandably much, much lower. That means that, based on Towne's relatively unknown campaign, he's already managed to garner 8% of the vote. If Mr. Towne continues on this path, that could entail his numbers continuing to rise along with his name recognition -- and, since he's opposing Dent from the right, Towne could take a decent chunk out of Dent's conservative voters.
Finally, the Callahan campaign tells the Morning Call that they have their own polling and are pleased with their chances.