There's no beating around the bush about it- our poll in NY-23 was way off the mark. I made a poor judgment call by not just killing it when Scozzafava dropped out and then endorsed Owens. Like any pollster we will make mistakes, we will just try not to make the same ones twice. Obviously the bizarre machinations in the race last weekend made it extremely volatile and if we ever a poll a race this strange again we will make better decisions (or maybe make the most prudent decision to just sit such a contest out!) But I apologize for putting bad information out there, it is very embarrassing.
Beyond that though I'm pretty proud of how we did tonight. There was a lot of skepticism about our New Jersey and Maine numbers and we did really well in both places. I'm not going to lie, it feels very good to beat most of the New Jersey pollsters in New Jersey, especially since it was our first time polling the state. We did a good job of chronicling the end of Corzine's gain in momentum in mid-October and the shift back toward Christie over the last couple weeks.
In Maine I think our accuracy is an indicator that IVR is superior to live interviewers when polling on sensitive issues about sexuality and probably race. I've argued that since we put out our first poll there.
In Virginia we were pretty good, I'm very proud of the sum total of our polling there over the last two years. We were also very good in Chapel Hill, correctly predicting a one point race and that the Council race would be pretty bunched up but with the incumbents winning. In Charlotte we were within the margin of error on our final prediction but underestimated the strength of Anthony Foxx's turnout operation and ability to get Obama wave voters back out- hats off to them.
I will note that with the exception of Virginia all of our 'error' overestimated the prospects of Republican candidates, which should just go to show that much more that our polling is above board and non biased, not that I expect that to stop Republican whining when they don't like our numbers.
Thank you to everyone who's followed our 2009 election cycle polling.
Great job Tom.....
ReplyDeleteCongrats on a good night.
ReplyDeleteYou can't win them all, and it's refreshing to see a pollster own up to his/her mistakes. It is not obvious that PPP really did anything obtusely wrong-- it is likely that Rep.-elect Owens really was 17 points behind before Scozzafava's endorsement sank in with voters. However-- I agree that, under the circumstances, it may have been best to pull the poll.
PPP remains my favorite polling outfit-- see you in 2010!
Tom,
ReplyDeleteI wouldn't take the NY-23 numbers too hard, that was a wild and crazy race and even the best of pollsters have an occasional crazy race. Plus, you guys are probably among the few who actually own up to your mistakes, there are way too many people who don't do that!
On NY-23, told ya so. Would you like to know how I knew?
ReplyDeleteChapel Hill was a three-point race though. But yeah, you did pretty well, and I don't think that anyone (well, except die-hard partisans) is going to hold NY-23 against you.
ReplyDeleteWell, you knew all the facts about N-23 and yet you were making a strong prediction that Huffman was going to win big (17 points) as late as Monday. Well he lost by 5%, which means you were way off the mark by 22% points. That's one of the worst results I have ever seen from any pollster in recent memory. You should seriously re-consider your methodologies ( on Monday you claimed your polling found no change after the R candidate endorsed Owen).
ReplyDeleteAs far as NJ and Main, those were close races and the end result was close to the margin of error with most pollsters, so even though you predicted them right there is nothing to be proud about here.
It wasn't a weird night for PPP, it was a very bad night. No way to sugarcoat it.
So does this mean that all of your negative comments about Palin are out the window as well?
ReplyDeleteThank you, too. I was all hopeful for Corzine until I saw the split between IVR and live-interviewer polling, and IVR turned out to be right. As for NY-23, nobody called that one correctly: not really a black eye for PPP.
ReplyDeleteWere your turnout predictions correct in Maine? I read that the Maine SoS was expecting 35% of RVs and got 50%; maybe you were expecting 50% as well.
The Charlotte race suggests that the Obama turnout machine really might reappear in 2010, at least if we can point to accomplishments by then. If not, and if the economy stays flat on Main Street (as seems likely), then 2010 will indeed be a rough year for Dems.
"The lesson there is that the new voters will come back out- with a good candidate. Foxx was a good candidate. Creigh Deeds in Virginia and Jon Corzine in New Jersey were not."
ReplyDelete"good candidate" or black candidate?