Monday, November 2, 2009

Election 2009 -- A Viewer's Guide

Let's get right down to predictions, the predictions of others and what to look for tomorrow night as we assess the "mini-midterm" elections for Governor in Virginia and New Jersey and the wild special election in New York's 23rd Congressional District.

First, the backdrop: these races do have national significance, at least to the extent that they reflect the mood of pockets of the country (in this case, the purplish state of Virginia, the blueish state of New Jersey and the purple-red New York 23rd), but are overblown in their coverage for no other reason than that we don't get any other actual voter data until next year, when we have 10 times as many Governor's races and 435 times as many House races, as well as 30-some Senate races.

So, let's not overhype this too much -- these elections matter, but they are not life-changing for the parties.

Having said this, let's get down to the analysis.

Let's begin with Virginia, by far the easiest prediction of the night:
Polls by leader: McDonnell 6, Deeds 0, Even 0
Poll Averages with Rasmussen (without Rasmussen)
Unweighted Average: McDonnell +14.0% (+14.2%)
Sample-Weighted Average: McDonnell +13.8% (+14.1%)
Median: McDonnell +13.5% (+14.0%)
Average of Averages: McDonnell +13.8% (+14.1%)
Final Prediction: McDonnell 57.0%, Deeds 43.0%, Likely GOP Pick-up

What Others Are Saying:
RealClearPolitics Prediction: McDonnell +13.4%
Intrade Odds: McDonnell is a 110:1 favorite

The Intrade odds seem a little excessive to me (I would say 40:1 or 50:1 would be fair, but if I were betting, I might lay $10 on Deeds to take a longshot at $1,100), but it should be an early night in Virginia.

Things to watch: Basically this one just comes down to who has the best polling methodology -- is it Survey USA at McDonnell +18%? Or perhaps Research 2000 at McDonnell +10%? Which averaging methodology works best? Does McDonnell win by closer to 13.5% or closer to 14.2%? No drama in results here short of some type of massive disruption at the polls or foul play -- Bob McDonnell becomes the next Governor of Virginia.

Next, let's move to New Jersey.
Polls by leader: Christie - 3, Corzine - 2, Even - 0
Poll Averages with Rasmussen (without Rasmussen)
Unweighted Average: Christie +1.0% (+0.5%)
Sample-Weighted Average: Christie +1.3% (+0.6%)
Median: Christie +2.0% (+0.5%)
Average of Averages: Christie +1.4% (+0.5%)
Final Prediction: Christie - 45.0%, Corzine - 44.1%, Daggett - 10.9%, Lean GOP Pick-up

What Others Are Saying:
RealClearPolitics Prediction: Christie +1.0%
Intrade Betting Odds: Christie is a 1.08:1 favorite

That's right folks, after weeks of predicting that New Jersey would tilt back blue and ultimately go for Corzine, I'm predicting a Christie win, albeit by a seriously slim margin. The scenario of a late departure to Corzine putting him over the top is certainly possible, but we have very good, recent, polling data available here that does not point to a pending surge. All indications are that this one is going to be a squeaker either way and the statistical evidene says that Christie is more likely than not to win, although by no means assured. The betting odds seem about right, although faced with those odds, I'd be inclined to lay $108 on Christie to win $100.

Things to watch:
(1) Voter turnout -- high is good for Corzine, low is good for Christie. Corzine needs all those new Obama voters to show up in an off-year -- the GOP stalwarts almost always show up.
(2) The Daggett effect -- how many voters dump Daggett late for another candidate? History has shown times where Independents get dumped at the last minute when it is clear they can't win. But will these voters abandon Daggett and if so, who will they vote for?
(3) Ballot Burial -- New Jersey's pro-two party balloting laws bury Daggett amidst a bunch of nobody independents that litter the ballot, while the top two candidates get top billing. This could cost him votes, but again, to where will they go?
(4) Suburban South Jersey -- Newark and Camden will vote for Corzine, the rural northern and central part of the State will go for Christie...but where will the greater Cherry Hill area vote? That's the swing part of the state and will probably tip the election.

This is THE race to watch tomorrow night, and I'll be watching closely.

Finally, to New York-23, where we simply don't have good data to make a statistically-based call. Two days ago, Republican Dede Scozzafava suspended her campaign. Yesterday, she endorsed Democrat Bill Owens.

We have one poll, taken today by Siena that took place after both of these developments. It shows Conservative Candidate Doug Hoffman 41%, Owens at 36% and Scozzafava (who is still on the ballot), showing up at 6%. It appears that the Scozzafava voters (who are mostly Republicans and Republican-leaners, after all) are breaking somewhat more for Hoffman than Owens. 6% of people are either protesting or didn't get the memo that Scozzafava is out.

I can't make a mathematical prediction, but from this little bit that we do know, this certainly appears to continue to be a Lean GOP Hold. On the betting odds below, I'd still bet on Owens. I expected Hoffman to win, but 2.3 to 1 is reasonably long odds for such a turbulent race.

What Others Are Saying:
RealClearPolitics: No prediction
Intrade Betting Odds: Hoffman is a 2.3 to 1 favorite (actually "anyone but Owens" is the 2.3:1 favorite, but for all intents and purposes, that means Hoffman at this point.)

If you live in Virginia, New Jersey or New York's 23rd Congressional District or even if you just have local races to contend with, please vote, regardless of your political stripes. I do my best to project races, but in the end, real votes count, not my predictions. Let your voice be heard!

And tune in to this site for coverage of election night 2009.

No comments: