California Republicans face an interesting ballot this November. The first office they'll see is the presidency, and their presumed nominee is greatly unpopular and possibly facing a convention insurrection. The second office, US Senate, features only two candidates, both of whom are Democrats. How will the typical Republican vote in that election? Chances are a good many of them won't.
The peculiar Senate race is the result of California's top-two election system, which voters approved in 2010. For any partisan race below the presidency, voters simply pick their favorite among all the candidates regardless of party, and the top two vote getters go to the November runoff. The Senate contest, between state Attorney General Kamala Harris and US Representative Loretta Sanchez, is the first California statewide race to feature a same-party runoff.
One of the early hopes of the top-two system was that it would boost voter turnout. Since any registered voter of any party, or even no party at all, could now participate, in theory many of them would. Plus, candidates would recognize the potential contributions of unaffiliated voters and would be incentivized to cater to them.
These hopes have largely not panned out. Voter turnout in California's 2014 elections was at record lows. But what about in same-party contests like this year's Senate election? These contests present an unusual situation. In theory, the more moderate candidate might have an advantage, since independents and minority-party voters might join forces to stop the more extreme candidate.
To get a sense of this, I examined all the runoff races in California's 2014 general elections for Congress, Assembly, and state Senate. I omitted the ones that had just one major-party candidate running. I was mainly interested in observing ballot roll-off. This is a common effect in which people vote for the contest at the top of the ballot but decline to participate in one lower down. In 2014, the top contest was the gubernatorial race between Democratic incumbent Jerry Brown and Republican challenger Neel Kashkari.
In down-ballot races where a Democrat was running against a Republican, the typical roll-off was about 2.8 percent. That is, 2.8 percent fewer people cast a vote in a congressional or state legislative race than in the gubernatorial contest. For a same-party runoff contest, though, the roll-off was more than three times that — 9.7 percent.
In the chart below, I have broken down those roll-off figures by office level, looking at the 80 Assembly races, 20 state Senate races, and 53 US House races up in 2014. While overall roll-off is understandably lower in US House races, with their higher spending and visibility, the gap between cross-party and same-party runoffs is pretty similar across races. Far fewer people tend to vote when their choice is between candidates of the same party.
Further analyses by Eric McGhee and Jonathan Nagler demonstrate that it's mainly the minority-party voters who are declining to participate. Given a choice between a Democrat and a Democrat, many Republicans stay silent.
It's possibly in the minority-party voters' interests to figure out which of the candidates is more moderate and more likely to deliver policies they want, but that's actually very difficult for even well-informed voters to do. And in today's very partisan environment (particularly in a polarized state like California), the benefits one would derive from voting for one slightly more moderate officeholder are vanishingly small.
In addition to some of the notable problems of the top-two system (such as members of one party picking the nominees of another party), we have this significant turnout problem. The promise of moderation coming from same-party runoffs only works if voters from the minority party participate. Competing for those voters is supposed to force the candidates to the center. But it's looking like a lot of those voters are just letting those contests go on without their input. They don't like the options they're being given, so they're choosing not to decide.