If polarization in politics is a serious problem, is there anything we can do about it? I recently wrote a report for Brookings (PDF) on this topic, addressing some of more popular reform ideas proposed for reducing excessive partisanship in Washington and assessing their likelihood of actually working. (Spoiler: It's got a fairly bleak conclusion, for the most part.) I talk about primary reform, redistricting, campaign finance reform, and a few other proposals I've written about here and elsewhere.
One topic I just briefly investigated was ranked-choice voting (RCV), also called instant runoff voting (IRV). RCV is an election method designed to produce election outcomes somewhat closer to what the public's true preferences are than our usual system of plurality elections does. In a typical RCV election for a single office, voters would get to list their top three candidates for the job. When the results are tabulated, the candidates are ranked according to the number of first-place votes they received. Then the lowest vote-getter is dropped and her votes are redistributed among the other candidates based on the second-place votes listed on her ballots. This process continues until only one candidate remains.
The ones who are more easily able to learn the new voting rules tend to have higher incomes and more education.
This isn't some pie-in-the-sky idea for reform—it's actually being used in several cities today, including Berkeley, San Francisco, and London, as well as in presidential elections in Ireland and parliamentary elections in Australia. (It was also used in elections for the senate of the Associated Students of the University of California at Berkeley when I ran and lost many years ago.)
One advantage of such a system is that candidates have an incentive to be civil to each other, even in the heat of an election. You may really want to defeat your opponent, but if you want that opponent's main supporters to list you second on their ballots, you can't be too cruel. And according to a study of San Francisco elections, the adoption of RCV coincided with an increase in the racial diversity of winners.
These all sound like positive developments. And while such an election has, to my knowledge, not yet been implemented in an American party primary, the possibilities for creating a more diverse set of party nominees through RCV sound promising.
However, Jason McDaniel throws some cold water on these claims with a study of five San Francisco elections conducted between 1995 and 2011. As McDaniel notes, the RCV ballot is a lot more confusing than a traditional "pick one" ballot. It requires some modest instruction for voters to know how to cast a vote. The resulting confusion has costs associated with it. Voters make more errors under RCV systems, with a disproportionate number of ballots spoiled due to over-voting (casting too many votes on one ballot).
Voter confusion also tends to lead to lower participation, but not distributed evenly across the electorate. The ones who are more easily able to learn the new voting rules tend to have higher incomes and more education. The ones who don't participate are more likely to be poorer and less well educated, and are also more likely to be African American or Latino. In other words, RCV seems to exacerbate some of the inequalities that already plague our political system.
None of these studies should be considered the final word on RCV; it has not been employed in many cities for very long and available data are scant. The results we've seen so far may also be spurious—it is not clear that RCV itself is causing the shifts we've seen. But we should at least be cautious as we consider using it in other types of elections, and recognize that even if it brings about a desirable decline in polarization, it could bear many other notable costs.