Phil Keisling, Oregon’s Secretary of State from 1991-1999 and currently director of the Center for Public Service at the Hatfield School of Portland State University, wants mayors to be elected in non-partisan elections, held at the same time as general elections.  He is worried about low turnout in partisan primaries held in odd year elections (in the recent NYC mayoral primary, turnout was 22% overall, only 13% among registered Republicans).  He worries that partisan election systems “relegate minority-party and non-affiliated voters to “observers-on-the-sidelines” status while forcing candidates through the same partisan paces that are driving our national politics into the ditch.”  “(E)fficiently delivering core municipal services or revitalizing downtowns” has little to do with the issues that currently animate party divisions in Washington.  Swich to non-partisan elections, Phil argues, to increase turnout, attract young people to the polls, and revitalize trust in government.  (The full argument is here: http://www.governing.com/columns/smart-mgmt/col-wrong-ways-elect-mayors-partisan-odd-numbered-years-instant-runoff-voting.html)

Phil Keisling has a well-deserved reputation as an election innovator for pushing through vote by mail in Oregon.  He advocated for the top-two primary in Oregon.  And he continues to work to improve civic policies and engage young people in government.

But on this point–as on the top-two primary proposed in Oregon in 2008–Phil and I will have to politely disagree.  Let’s not toss out the party baby along with the dirty bathwater, especially if the bathwater is being generated in Washington, D.C., not in our local municipalities.

Phil’s unhappiness with the direction of the national Republican Party may be blinding him to the positive role that political parties can play in structuring politics not just in the United States, but in every democratic political system yet devised by man.

John Aldrich, a political scientist at Duke, famously asks “Why Parties?“, and his answer is that

parties serve to combat three fundamental problems of democracy: how to regulate the number of people seeking public office; how to mobilize voters; and how to achieve and maintain the majorities needed to accomplish goals once in office.

Parties are a “name brand”, according to another political scientist,”providing credible information about how politicians are likely to act in office.”  Parties serve as training grounds for new political actors, recruit candidates for office, and provide avenues for upward political mobility.  Partisanship among individual voters remains the most important predictor of the vote, and helps voters make order out of a bewildering variety of political claims and issues.

Political parties are obviously not a panacea, and political divisions are deeply problematic in America today.

But let’s not cure the disease by killing the patient.  The problem as I see it is that Phil lumps together different institutional forms–most notably closed partisan primaries with general elections–and concludes that all party labels must be a bad thing for turnout and for voting.

There is a good argument to be made for opening up partisan primaries to unaffiliated voters, as many states do, or perhaps having a “top two” or some other “open” system.  It’s not clear that this will result in substantially increased turnout in primary elections, as Phil claims, but it would allow those voters who don’t want to officially affiliate when registering to vote to participate in the primary.

But to leap from there to non-partisan general elections is a leap too far.  Voting in a non-partisan general election, according to Brian Schaffner and Matt Streb, is like watching a football game where the teams aren’t wearing uniforms.  No one knows who is ahead, who is behind, or who to root for.

Non-partisan elections do not increase turnout–they depress it.  Non-partisan election do not result in more-informed voting, but instead they force voters to replace one cue (party) with others (interest groups), and most problematic, they end up empowering incumbents.

Nor is there any guarantee that a party label assures a victory.  Phil makes the unfortunate error of claiming that winning a partisan primary in New York City is tantamount to victory in the general election.  Michael Bloomberg and Rudy Guiliani would beg to differ.

Once we get beyond non-partisanship, Keisling advocates for a number of positive reforms.  Align local and state elections with the federal general elections?  Absolutely.  Experiment with innovations like instant runoff voting, which avoid the need for partisan primaries?  Great idea.

But abandon political parties?  Unless you want to strengthen incumbents and interest groups and weaken voter control, it’s not a good idea.

So says a new academic study by the team at the University of Wisconsin.

The study is blogged about in detail at the Fact Tank at the Pew Center on the States and will undoubtedly spark some reaction.  The takeaway point for anyone interested in elections and turnout is that a simple minded rational choice model of turnout is, by itself, simply insufficient to understand voting behavior.  (This piece cites none of the “classic” turnout articles, nor should it.)

I’ve been critical of the Wisconsin approach in the past for failing to discriminate among different modes of early voting (they code no-excuse absentee and early in-person the same way) and, at least in past work, for relying only on data from 2008.  This study doesn’t address the first issue but does expand the universe to include the 2008 presidential contest, a significant advancement.

The study also demonstrates the value of peer-review.  Peer review is criticized because it is slow and deliberate.  But peer review makes it a lot more likely that scholars get to the right answer.  In this case, I’ve seen this paper through a number of versions (including acting as a reviewer), and the impact of peer review is very clear in the final product.

There are other pieces circulating that purport to address this question, including at least one that I reviewed recently showing an opposite result, controlling for the number of early voting locations in each jurisdiction.  This question is certainly not closed, but this piece is going to stand as an important marker in the field.