Last updated on 2012.11.11
I had fun giving Bob Mercer heck for his purely speculative September hypothesis that the folks who voted for B. Thomas Marking in 2010 decided that election and could have decided this year's House race. I long maintained that there was no evidence that Marking had any influence on the 2010 outcome and that it thus made no sense to assume those folks would do anything other than split evenly this year between Kristi Noem and Matt Varilek.
But then I looked at a few simple numbers that indicated I was wrong—not as wrong as Bob Mercer, mind you, but wrong on the neutrality of the 2010 Marking vote. Consider:
- In 2010, B. Thomas Marking won 19,134 votes for Congress, 6.0% of the votes cast in that race.
- Marking's percentages in each county correlated weakly but significantly (r = 0.3596, p = 0.003) to Noem's 2010 county percentages. Where Noem was strong, Marking tended to have a little more strength than his average as well.
- Marking's percentages correlated a bit more strongly and negatively (r = –0.4583, p = 0.0001) to Stephanie Herseth Sandlin's. Where SHS was strong, Marking was noticeably weaker.
- This year, Matt Varilek's percentages in each county correlated negatively (r = –0.3899, p = 0.001) to Marking's 2010 numbers in each county.
In other words, these data suggest that, if the Marking 6% were anything other than statistical error in 2010, they were stronger in places where Noem was already stronger than SHS in 2010. An exact percentage is beyond this morning's math, but these numbers lead me to believe that Marking voters did not siphon equal numbers of voters from the party candidates in 2010. Marking more likely dragged down Noem more than SHS. Had he not run, his voters would have leaned significantly toward Noem.
And this year, that tendency held. Noem beat Varilek by 53,851 votes, 14.9% of the electorate. But I'm willing to be that a strong majority of Marking's 6% added to Noem's cushion.
I really don't agree with the idea that Marking voters took from Noem. For one Varilek was a vastly weaker candidate than Noem. Why would they back Varilek? No reasons. In 2010 it was a change election. Either you wanted to change DC or you didn't. If you did want to change DC you voted Noem. Why would people who wanted change throw their votes to someone who couldn't win?
The other thought is that Markings vote total clearly mirrored Lory Stacey's 6% in the SOS race in 2010 and about the 6% Clarke got in 2012's PUC race.
Mike, I wondered about that: maybe Marking and Stacey really were just appealing to the same "what the hell" 6%. But they weren't. When I run the vote totals by county, I find that the correlation between Marking's percentages in each county and Stacey's in the same county was only 0.2348, with a statistical significance not quite meeting the p=5% threshold of confidence. That tells me that you can't look at a county where Marking was strong and conclude that Stacey was strong as well. A Marking–Stacey alignment does not happen with the same frequency as the Marking–Noem 2010 or the Marking–anti-SHS vote. Based on the numbers above, I am willing now to argue that Marking took more voters from Noem in 2010 than he took from SHS, and that if the 2010 race had been a straight-up battle, Noem would have won by a bigger margin.
As for a Clarke–Stacey–Marking alignment...
—correlation of Clarke 2012 percentages by county to Marking 2010: 0.4782.
—correlation of Clarke to Stacey: 0.4359.
...Both around the mild but significant level I not between Marking and anti-SHS votes in 2010, and arguably more noteworthy since they were on separate ballots. But that would likely fit my hypothesis here (and a pretty sensible one, really, fitting with other observable data) that the folks voting for our third-party choices lean GOP when not given a choice. (And seriously, when was the last time we saw a real liberal third-party candidate who could peel away Dem votes?)