First the bragging:
-It was said repeatedly in the media that nobody had any idea what turnout was going to be, so the polls couldn't be trusted. Most estimates fell in the 200,000 range. Well we went on the record to Politico as saying we thought turnout would be around 300,000 and it looks like we were pretty much right.
-Our final poll showed Deeds at 40%, McAuliffe at 26%, and Moran at 24%. On the surface it may appear that we were well off because Deeds ended up with 50% (our McAuliffe and Moran projections were exactly correct.) But look at it this way- from our second (Deeds 27, McAuliffe 24, Moran 22) to last to our final poll 80% of the voters who made up their minds had picked Deeds. It looks like that trend continued right into election day with the remaining undecideds.
Second, why I'm not surprised we were on the mark:
This was the perfect race to be polling using the voter list and automated calls.
We called folks who had voted in any one of the Democratic primaries between 2005 and 2008. The reasoning behind that was basically no one who didn't bother to turn out for the Presidential primary was going to turn out for this one. So we were sure everyone who might possibly turn out was included within our sampling frame.
Some folks thought we were being too liberal in our sampling because for the most part it seemed unlikely voters who turned out just for Obama/Clinton were going to come out for the much less interesting Deeds/Moran/McAuliffe match. Well we agree that most 2008 only primary voters weren't going to come out again, but we still thought a decent number of them would.
That's why we said at the start of every poll, if you don't intend to vote in the primary, please hang up now. When you're dealing with an automated poll folks who don't intend to vote don't feel the sort of social pressure they might feel from a live interviewer to participate. So folks who didn't plan to vote didn't bother to answer the poll. No harm done. You don't want a high response rate from people who aren't going to vote. But the sampling we used did ensure that those more casual voters who were going to come out again this year were included.
I have no doubt this is very different from how polling was conducted 20 years ago. But the world is moving forward.
Finally, some notes on the media. Outlets that continue to have stubborn policies about automated polls and put out stories this morning saying the race was a toss up frankly misled their readers. This race obviously ended up being a total blow out, and the final round of polls did reflect that.
It will be interesting to see if the Washington Post writes any follow up to its haughty May declaration, backed up with no facts, that automated polls 'take less care than polls conducted by live interviewers.' One thing's for sure: PPP and Survey USA took more care in polling this race than the WaPo did, because they didn't have enough confidence in their operation to take a crack at it. We're not afraid to poll difficult contests, and because of that we'll inevitably screw up sometimes. But we did pretty well on this one.
Congratulations to Creigh Deeds on a remarkable campaign!