Senate Minority Leader Mitch McConnell of Kentucky probably has less to fear from his Democratic challenger.  Photographer: Win McNamee/Getty Images
Senate Minority Leader Mitch McConnell of Kentucky probably has less to fear from his Democratic challenger. Photographer: Win McNamee/Getty Images

With the last potentially competitive Republican Senate primary in Tennessee today, Nate Silver looks at the numbers and declares that “the GOP still has plenty to worry about. There have been far more close calls to its incumbents than usual.”

Silver reports that Republican incumbents this year are averaging only 73 percent of the vote in their primaries, down a bit from 2010 and 2012, and substantially lower than the 89 percent Republican incumbents were drawing in the three cycles before that.

So what does this mean for future incumbents? Silver takes a straightforward position: We’re dealing here with a continuous variable, so an era in which senators typically win by 40 points is a lot more treacherous for them than one in which they typically win by 80 points. As he says, that’s absolutely true in sports: A baseball team that wins a lot of blowouts is a much better team than one that wins a bunch of one-run games, and can safely be projected to win more in the future. Following that logic, the fact that several senators fell below 65 percent in their primaries this year suggests that Republican senators in general are in trouble, and it was just luck that all of them this year (so far?) have survived.

Well, maybe.

It’s possible that Silver’s logic will hold. In particular, it’s almost certainly correct to think of Senator Thad Cochran of Mississippi’s very narrow survival this time as a chance result -- which should hardly reassure any of his colleagues.

On the other hand, I’d be very wary of overinterpreting some of the diminishing gap.

Think about general elections in districts dominated by one party. Sometimes, the out-party won’t even bother fielding a candidate, and the incumbent will receive 100 percent of the two-party vote. Other times, a certain loser can be found to at least show the flag. Depending on the district, that hopeless candidate might be destined for, say, 30 to 40 percent of the vote. If a systematic change takes place that makes it far more likely for hopeless losers to file in those districts, the average incumbent margin of victory is going to be much lower -- but the chances of incumbent victory won’t change.

That might be what’s happening in Senate primaries. Whether it’s campaign finance, or institutional changes within the Republican Party, or some other change, it’s possible that it's easier to enter races for hopeless losers who nonetheless are capable of running something resembling a real campaign, and therefore of winning a solid share of the primary vote. Without, in fact, making it any more likely that incumbents will lose.

There’s another piece of this. In baseball, average margin of victory is a good indicator of underlying quality, partly because teams basically always attempt to maximize runs scored and minimize runs allowed.1 It’s probably not true, however, that incumbent senators in primary elections always attempt to maximize their vote percentage. An incumbent with a 60 point lead, or even a 25 point lead, four weeks before Election Day may well save scarce resources for the general election.

I’m not saying that Silver is definitely wrong, just that elections can be complicated, and average margin of victory -- while interesting -- isn’t necessarily meaningful.2

One additional complication. It matters how often incumbents actually lose. It also matters, however, how much they fear primaries. To determine that, sophisticated quantitative analysis of past elections is utterly irrelevant; what we need to know is what senators think their greatest threats might be. So we need to listen to what those senators say -- and even more, we need to watch what they do. Do they act as if their main concerns are primary challenges or general elections? After all, whether it’s easier or harder to knock off an incumbent senator, most of them will be renominated and most of them will be re-elected. But as I always say, politicians are paranoid by nature; what we need to know is what these particular politicians are paranoid about.

1 Yes, in a blowout, a team might give the starting lineup some rest, or scrape the bottom of the bullpen; yes, there are other ways that it’s not quite literally true that every team in every situation is attempting to maximize the runs scored/runs allowed ratio. But we have plenty of evidence on this, and it all shows that we can model baseball as if teams were trying to score every possible run and prevent every possible run.

2 Another example: House incumbents opened larger average margins of victory in the 1960s and 1970s, leading some analysts to conclude they were increasingly safe. Later research found that those average margins of victory were accompanied by larger swings between elections, and that therefore fewer close elections didn't mean safer incumbents.

To contact the writer of this article: Jonathan Bernstein at

To contact the editor responsible for this article: Max Berley at