Double Reverse Coat Tails

Ed Kilgore reports that someone at U Minnesota argues that Republican governors didn't help Romney with their reverse coat tails.  I am honestly surprised at the sign of the alternative hypothesis against which the no reverse tails hypothesis is not rejected.  I comment



If it isn't one thing it's the other.  The problem with data is that it's hard for them to average to exactly zero.

I heard a reverse coat tails hypothesis in 2011.  The idea was that the recognized horribleness of Walker, Scott and Kasich would help Obama.  Ooops the evidence against the 2010 vintage reverse coat tails hypothesis is evidence in favor of the 2011 reverse coat tails hypothesis.  Really my honest guess was that the allegedly unsupported hypothesis would be that Obama would do surprisingly well in states which elected Republican extremist governors in 2010.

In any case, the method of looking at who won states with close vote totals makes no sense.  The states to look at are chosen based on the same variable which is averaged.  I promise statisticians consider this, sorry to use technical terms, a major boo boo.

Rather the Obama vote should be compared to a predicted vote based on Obama 2008 vote and, say, the change in unemployment in the state January through October 2012 (that is what are called fundamentals by political scientists and Nate Silver, by which I mean the number Nate Silver calculated cause do you really think anyone else is going to do much better?).  Look at this Obama surprise by change in party of governor 2008 to 2012 so shifting D to R in many states (including NJ and VA switching in 2009) and R to D in none IIRC.

Well now I should just do it.  Really Obama 2012 - Obama share 2008 on Republican Gov 2012 minus Republican Gov 2008 (note I've dropped the rest of the fundamentals).  Oh crap, that's something I should just do rather than talk about doing it.

Oh hell will be half vast.

Nationwide HuffPo 2012 minus Wikipedia 2008 Obama's share declined 4
          Obamashare12 - 08
Florida     - 1 % Scott nega coat tails = -3%
Iowa        - 1%  Branstead negatails -3%
Indiana     - 6%  Deaver tail 2%
Maine       - 2%  LaMontaine Lameness -2%
New J     + 0.6%  Christie -4.6% -- which side was he on?        
Ohio        -1    Kasich Kost Romney -3%
Penn        -2.5  Corbett Cost Romney 1.5%
Virginia    -1.5  McDonnell ??? negatail -2.5
Wisconsin   -3.4  Walker V Ryan's hope ? about average

It sure seems to me that the nega coat tail hypothesis is supported by the data.  Obama's vote share declined less in all the states where I remember a Republican gov winning in 2009 or 2010 except for Indiana.

Would this be surprising ? Could there be worse advertisement for the Republican party ?