Tuesday, February 28, 2012

Prognostications #8 and #9:
The primaries in Michigan and Arizona

After three weeks of waiting, the crystal ball is back.  The voters in Michigan and Arizona go to the polls and there have been plenty of polls to give us an idea of an idea of how things are going, so I'm making a prediction of the final numbers and so is Nate Silver of fivethirtyeight.com and The New York Times. There has been more data published about Michigan than Arizona, including two polls in the last twelve hours, but both have been well covered, so here are the numbers, with my prediction first and Silver's in parentheses.

Romney 38.0% (39.0%)
Santorum 37.5% (38.1%)
Paul 14.0% (12.2%)
Gingrich 9.5% (9.9%)
None of the Above 1.0% (0.8%)

Romney 43.0% (43.4%)
Santorum 27.5% (27.3%)
Paul 10.0% (10.2%)
Gingrich 18.5% (19.2%)
None of the Above 1.0% (-0.1%)

We have no disagreements about the order. Both of our models say Michigan will be very close and Arizona will not. 
In Michigan, our big difference is about Ron Paul.  There were some polls that had Paul comfortably in third place and others that said he and Gingrich were neck and neck.  Most of the latter had much larger Undecided votes than made sense this late in the game, so I did not favor those polls, regardless of freshness.

In Arizona, the big disagreement is about None of the Above.  If it's anywhere near 1%, this will give my model a big advantage. If it is under 0.5%, Silver will be in better shape.
In the real world, this is really about delegates, though there is also value in getting good press by winning or making a good showing. Michigan's race is multiple contests, two delegates per each congressional district and two more for the overall winner in the state. Silver has some polling information that one district might be won by Dr. Ron Paul - the candidate the Daleks hate most - while the rest are battles between the front runners. Arizona is winner take all and R-Money's lead looks insurmountable there. Expect the news coverage to focus on the closeness of Michigan and ignore that Mitt's lead in delegates will increase.

I'll have a new post tomorrow giving the results. So far in the five races where each of us has predicted an outcome, I'm ahead 3-2.


Anonymous said...

I'm thinking Little Ricky will pull this one out in MI with 40%.

dguzman said...

How much effect do you think that pRicky-voting democrats will have on the final outcome in MI?

Matty Boy said...

Z&M: Both Silver and I thought Mitt would win, but your guess was definitely possible this time.

dg: To me, this story was another news item without data, like when I thought Buddy Roemer might beat Rick Perry. I don't chase that stuff anymore.