When Partisan Primaries Produce Unrepresentative Extremists...

...isn't the obvious remedy to eliminate partisan primaries?

Philip Bump in the Washington Post has an interesting column with a bunch of nifty graphics. They show that each cohort of newly elected Republicans in Congress, both the House and the Senate, has become increasingly more extreme from one decade to the next. For Democrats, not so much: the ideology of newly elected Democrats has remained roughly the same since the Sixties. This difference is what political scientists call asymmetric polarization, and it’s a phenomenon that’s been well-known for some time.

What Bump’s new column adds to the analysis is the observation that the extremity of a newly elected Republican member of Congress is not dependent on the size of the new member’s margin of victory. In other words, these days a newly elected Republican is just as likely to be extreme in a competitive district, with narrow margins of victory, as in an uncompetitive district, with wide victory margins. By contrast, winning Democrats in Congress tend to be extreme only in safe seats, while much more moderate in swing districts.

What to make of this data? Bump’s conclusion: “Republican voters elected more ideological candidates in primaries and those candidates often prevailed even in close elections thanks to partisan loyalty.” He adds parenthetically: “There might be a lesson here for Democrats”—presumably meaning that in primaries Democrats should nominate more extreme candidates, whom they might be able to get across the finish line in November, and thus have members in Congress who are more deeply or intensely partisan.

My conclusion from his data is different. The culprit here is the mechanism of partisan primaries, which, when they function in the way that Republican primaries currently do, offer general election voters a choice between only an extreme Republican and a Democrat. In a district or state that is unlikely to vote Democrat, even if the Republican is likely to win by only a narrow margin, the winner necessarily will be an extreme Republican because a more moderate one wasn’t an available option.

If instead the general election (rather than primary) voters were given a choice between a moderate and extreme Republican along with the Democrat, then it’s possible that the winner would be the moderate rather than extreme Republican. Lisa Murkowski’s victory in 2022 is an oft-cited example of this. She couldn’t have won a partisan primary, but she was able to win a general election against both an extreme Republican and a Democrat.

One way to prevent partisan primaries from blocking general election voters from choosing more moderate candidates who can’t win their party’s primary is to eliminate so-called “sore loser” laws. Murkowski, after all, won in 2010 as a general election write-in candidate after losing her party’s primary (before Alaska changed to its current format of an all-candidate “nonpartisan” primary). But eliminating “sore loser” laws will produce more moderate winners in November only if those more moderate candidates are able to get the plurality of votes in the general election.

Under the conditions of asymmetric polarization that led to the election of extremists in Congress, it’s more likely that either the more extreme Republican or the Democrat will be the plurality winner even if the moderate Republican is permitted on the general election ballot after losing the partisan primary. Adopting a “top two” system like California’s will solve the problem if the more moderate Republican would be among the top two pluralities in the all-candidate “nonpartisan” primary. But, again, asymmetric polarization may cause the top-two pluralities to go to the extreme Republican and the Democrat.

The solution to the problem that Bump’s data identifies is Condorcet Voting. If general election voters prefer the moderate Republican to either the extreme Republican or the Democrat, when each pair of candidates is considered head-to-head, then the moderate Republican will be the Condorcet Winner. Of course, there will be deeply “red” districts where a majority of general election voters will prefer the extreme Republican to either the moderate Republican or the Democrat. Then, the extreme Republican would be the Condorcet Winner. But that’s just as it should be in a democracy: the majority of voters should be able to elect the candidate whom they most prefer, whatever the ideology of the candidate happens to be.

But what’s so important about Bump’s new data is that it shows extreme Republicans winning seats in Congress from districts that are not deep red, but instead much closer to 50-50 purple. In these moderately red districts, the Condorcet Winner almost certainly would be a moderate Republican, not an extreme one. Yet it’s the structure of the existing electoral system, with its partisan primaries followed by a plurality-winner general election, that deprives the majority of voters in the district from electing the candidate whom they most prefer.

If Condorcet Voting were the electoral system in place for congressional elections, then the elected representatives from these more moderate districts would match the collective preferences of the voters in these districts, instead of the mismatch that Bump’s data shows. Moreover, if the Republican members of Congress were more aligned with their constituencies in this way, they would not tend to respond to an autocratic president with the “meekness and sycophancy” that Bump laments.