Results for Prognostications #8 and #9: The Michigan and Arizona primaries

While Mitt R-Money won both races last night, Nate Silver and I split in terms of closeness of prognostication.  I lost in Michigan by over-rating Ron Paul by a bunch, while Nate lost Arizona by predicting the None of the Above vote would be non-existent.  Here are the numbers.

Michigan

Romney 41.1%(Both of us low, Silver gains by 1.0%)

 Santorum 37.9%(I'm low by 0.4%, Silver is 0.2% high. Silver picks up 0.2%) 

Paul 11.6%(Both of us too high. Silver picks up a whopping 1.8%)

Gingrich 6.5%(Again, both of us high. I beat Silver by 0.4%)

None of the above 2.9%(Both of us low. I'm 0.2% closer)

 

Final prognostication score: Silver 91.6%, Hubbard 89.2%

 

Arizona 

Romney 47.3%(Both of us low, Silver gains by 0.4%)

  Santorum 26.6%(Both of us high. Silver picks up 0.2%)

Gingrich 16.2%(Again, both of us high. I beat Silver by 0.7%)  

Paul 8.4%(Both of us too high. I gain 0.2%)

None of the above 1.5%(Both of us low. I'm 1.1% closer)

 

Final prognostication score: Silver 89.0%, Hubbard 90.4%

 

If I made a mistake, it was discounting the polls with a lot of undecided in Michigan too much, which made my Ron Paul prediction there stand out like a sore thumb.  After seven match-ups, I'm ahead of Nate Silver 4-3.

So next week is Super Tuesday, with primaries in seven states (Georgia, Massachusetts, Ohio, Oklahoma, Tennessee, Vermont, Virginia) and caucuses in four (Alaska, Idaho, North Dakota, Wyoming).  There is also a caucus in Washington state on Saturday.  I am going to make an arbitrary cut-off point of at least three polls in a state reporting after Friday and not all three polls by the same company for me to put up a prediction.  I don't know Silver's threshold for enough data. I expect some of the caucus states are not going to have many polling companies interested in hunting down those tiny numbers of caucus goers, but there's a good chance Silver and I will be going head to head in at least six contests next week.  The only change to my model will be to up the standard None of the Above vote to 1.5%.

Mmmm... fresh data!  So very fresh. Mmmmmm.