Tuesday, January 31, 2012

Prognostication #4:
The Florida primary

Once again, all the debates and ads and opinion polls are over and it's time for some Republican voters to actually make their voice heard. There has been no shortage of polling and there is strong consensus on the order (Romney, Gingrich, Santorum and Paul). That said, the distance between the numbers is all over the place.  Of the websites I use to gather polling information, RealClearPolitics puts the data in the nicest format. If you follow the link, you'll see two of the most recent polls say Romney will either will by 7 or by 20. Who should we trust?

My method is to take the recent polls and look at the ones that have the least None Of The Above.  Then I look at the median and average values for all the candidates with the None Of The Above removed and give a weighted average. With the huge number of polls taken, I have seven that I'm using, none older than this weekend.  I round my data to the nearest half a percent.  Here are my predictions, as well as the latest from Nate Silver, founder of the website fivethirtyeight.com and now the poll expert for The New York Times.  Silver's numbers are in parentheses.

Romney 44.5%(44.0%)
Gingrich 31.0%(29.3%)
Santorum 13.0%(13.9%)
Paul 11.5%(11.2%)
Other 0% (1.6%)

As the numbers happen to shake out, I'm bullish on Romney, Gingrich and Paul, while Silver is expecting big performances from Santorum and Other. Our big disagreement is on Newt and for me to do well, he has to pull in at least 30%.
I follow the news but I don't let that influence my reading of the data.  I can't be sure how much effect Santorum's decision to leave Florida because of his daughter's hospitalization this weekend will move his final numbers. I'm a little surprised Paul's numbers stayed as high as they are given that he did not campaign in Florida. Rick Perry ignored New Hampshire and New Hampshire returned the favor.  But if my experience online has any meaning, I can say that Rick Perry's followers are not at loyal as Ron Paul's.

More to the point, I can say that most breeds of dogs are not as loyal as Ron Paul's followers.

By tomorrow morning the numbers will be in and I'll post again, reporting how close we got to right.

By the way, I'm now on Twitter under the name ConfidenceOfVic, short for Confidence Of Victory, my name for the method I use to assign odds to candidates in races with two or three people running.  I could use my method in a race like this one with four people, but I'm not as confident the numbers have meaning when more variables are added. 
I kind of like the shortened name on Twitter.  I've known some men named Victor who were shy, but any guy who shortens his name to Vic, that's a confident guy.


Anonymous said...

Not sure about this one. I think the I-4 corridor here is definitely talking Gingrich. Polls show him doing OK 9 pts off from Mittens in Miami but Romney has pissed off the Latinos down there. So Tampa and the panhandle numbers have to push him over the top. This will be very close and I'll take Newt winning by less than 1%.

Matty Boy said...

Nate Silver puts Newt's chances of winning at about 3%. My way of counting it would make the odds much worse.

Karlacita! said...

Wow, it's not going Newt's way so far.

How long does it take for the final results to be tallied?

Matty Boy said...

It's at 99% complete now at about 8:00 pm on the West Coast, around 11:00 pm in Florida. It's close enough to done to call the game. I beat Silver 90.7% to 90.0%, not a huge win but a win nonetheless. The only place I got a major schooling was by lowballing the amount of None Of The Above votes. I did much better predicting Newt's number and a little better on Mitt's number.

Nice to put another in the win column.