I've gotten a few questions about the "split decision" narrative, and whether it should change if Ds pull out wins in FL-Sen and AZ-Sen (good chance in both!). This gets to one of my biggest pet peeves about elections analysis and sports analysis. 1/
(This is a pet peeve of @NateSilver538 as well). So I'll start with a sports story: The 2014-15 Kentucky Basketball team (sorry @mghancock). It roared undefeated through the regular season and was pronounced a juggernaut for the postseason. 2/
The thing was, if you looked underneath the hood, things didn't look as impressive. They beat Columbia by 10 points. They beat Ole Miss by 3 in overtime. TAMU by 6 in double overtime. Vanderbilt by 8. FL by 7. LSU by 2. GA by 8. 3/
These were all unranked teams. Moreover, it was a down year for the SEC; they only ranked team during the regular season. Anyway, the point is, people make too much of a deal about close wins and close losses. The analysis was "this team always pulls it out!" 4/
When the analysis could also have been "Huh, this team almost lost to seven unranked teams." I wasn't terribly surprised when they almost lost in the Elite 8 and lost in the Final Four. 5/
The point is that, for *analytical* purposes, it doesn't really matter whether Rs lose AZ/FL narrowly or win them (it matters a LOT for 2020 and policy purposes). They're anomalous in the context of what we saw in House races, along with some of the gubernatorial results. 6/
Put differently, for analytical purposes, differentiating between a 5,000-vote McSally win and a 5,000-vote McSally loss puts way too much weight on the decisions of 5,000 people in a much wider electorate. 7/
For me, the challenge to various Grand Unifying Theories of this election is that, if you'd told me (or most people) over the summer that Ds were going to get 39 House seats or so, we've have predicted the Senate flipping, Ds sweeping GA/FL/OH governorships 8/
Stabenow winning by 20 (as she did in 2012), and so forth. Similarly, if we were told that McSally and Nelson lost by a few thousand votes, and the actual result in the above races, I suspect we would be skeptical about major Democratic House gains. 9/
The challenge of synthesizing those outcomes is the same, whether McSally and/or Scott narrowly win, or whether they narrowly lose. 10/10
