This piece, hits one of my buttons. McArdle tries to make the argument that 2016 will very likely be a poor year for Democrats. There are very sound reasons why someone might think that. Personally my opinion is we don’t yet have the kind of data we’d need to make any remotely useful predictions, especially with the unusually large number of variables at stake (economic recovery, ideological friction in the GOP, etc…). I have long been a devotee of the unpopular theory that a Hillary Clinton nomination, even an unsuccessful Hillary ATTEMPT at the nomination, would be a disaster for the party, but that’s a different post for a different day.
But what bugs me is the history argument:
Since the Civil War, only two Democratic presidents have been succeeded by another Democrat. Both of them–FDR and JFK–accomplished this by dying in office.
These kinds of arguments always remind me of this xkcd comic. The fact is that there have only been 43 presidents*. Only 14 of those Presidents were Democrats, and two of them, as McArdle helpfully points out, died in office and so are arguably not really comparable for analyzing transition questions. Which leaves us with 12. And since the party system has changed so thoroughly over time (unless you believe the party is under some kind of curse, which I guess cannot be entirely excluded), it’s really only fair to start counting at the end of WWII, which leaves us with… six. Oh, and since we’re talking about the end of Obama’s turn and that hasn’t happened yet, we really should be at five.
You know what you can do with a sample of five? Nothing. Nothing much at all.
On the same note, McArdle also observes that only four presidents since WW2 of either party have been succeeded by co-partisans. The problem with this is of course that since WW2 there have only been 13 Presidents (including FDR, who really shouldnt be included, but McArdle seems to have counted him so I will here). 30% of Presidents being succeeded by members of their own parties sounds like a good number for her point, until you think about it as 4 vs 9. It doesn’t take much tinkering to shift those numbers significantly either: A full-Florida recount in 2000 may well have given the Presidency to Al Gore, and those numbers would be 5 and 8, for example.
There really are some meaningful statistical things you can do with US election results. See Nate Silver. But these kinds of simple precedent-bases analyses fall down repeatedly, as the xkcd comic details, because even if your analysis counts every President, you’re still dealing with a sample so small that in any other field all noses would be instantly thumbed in its direction.
*Those playing the home game will remember that Obama is the 44th President. But that’s because Grover Cleveland served non-consecutive terms and so the official enumeration counts him twice.
The Viking Dutchman 