Share this:

Chris Bonneau 80x108Damon M Cann 80x108Nonpartisan elections—in which candidates are not endorsed by a political party and their party affiliation does not appear on the ballot—have been criticized as depriving crucial information to voters, making it difficult for them to vote for candidates that represent their beliefs. Chris W. Bonneau and Damon M. Cann tested the impact of nonpartisan election conditions using both a laboratory experiment and data from the Cooperative Congressional Election Survey. They find that there is no significant difference between voting behaviors in partisan and nonpartisan election formats.

Scholars have shown for years that the most important determinant of vote choice is the partisan affiliation of the voters:  Democrats vote for Democratic candidates and Republicans vote for Republican candidates.  However, a large number of elections in the U.S. are conducted in a nonpartisan format where candidates are not endorsed by political parties and their political party affiliation does not appear on the ballot.  What explains voters’ choices in these races?  We argue that these races are nonpartisan in name only.  While it is true that there are no ballot cues for voters, candidates’ campaigns send clear partisan signals to voters, which in turn enables voters to bring their partisanship to bear on their voting decisions.

Nonpartisan elections were designed to wrest control over elections from party machines and return it to the voters.  They are quite popular in local and municipal elections, where dozens of offices are elected, some of which have no partisan character to them (e.g., coroner, recorder of deeds, etc.)  Advocates of nonpartisan elections argue that they allow the officeholder to be accountable to the people, and not to the political party responsible for nominating him/her to the office.  On the flip side, opponents of nonpartisan elections counter that providing the partisan affiliation of the candidates provides an important informational cue to the voters, and depriving voters of this information is making it more difficult for them to participate and participate meaningfully in elections.

Credit: Steve Rhodes (Creative Commons BY NC SA)

Credit: Steve Rhodes (Creative Commons BY NC SA)

Our venue for this study is state supreme court elections.  We select these elections for two reasons.  First, state supreme court justices are important policymakers in the U.S. states, having the final word on interpretation of state laws and state constitutions.  Second, the way these justices attain (and retain) their jobs varies by state.  For our purposes, the most relevant distinction is that some justices run in partisan elections while others run in nonpartisan elections.  Using state supreme courts, we are able to hold the office (and its importance and duties) constant while evaluating voters’ choices in partisan versus nonpartisan election formats.

Our empirical analysis consists of two steps.  First, we conduct a laboratory experiment where we randomly assign subjects into a partisan treatment or a nonpartisan (control) treatment.  We provide subjects with a vignette that describes two hypothetical candidates for the state supreme court and ask subjects to indicate for whom they would vote (or if they would abstain).  In the partisan condition, we also provide subjects with the candidates’ party affiliation.  In the nonpartisan condition, we omit this information.  Everything else about the two vignettes is the same.  The results show a strong effect of party identification on vote choice, but there is no statistically significant difference in the magnitude effect of party identification on vote choice between partisan and nonpartisan ballot formats.  This means that in a tightly controlled setting where candidate information is held constant but only half of subjects were given explicit partisan cues, voters in the nonpartisan treatment condition were still able to draw conclusions about the partisanship of the judicial candidates and bring their partisanship to bear on their voting decisions.

While the experimental evidence allows us to uncover the effects of providing partisanship on vote choice, experimental results inherently have limited generalizability.  To bolster the external validity of our results, we supplement our experimental results with results based on observational data from the 2010 Cooperative Congressional Election Survey (CCES).  In this survey, we asked respondents for whom they voted in the judicial elections in their state (omitting individuals who did not vote).  Our key results are presented in Figure 1.

Figure 1:  Discrete change in the predicted probability of voting Democrat for a change in party ID by ballot format

Bonneau Fig 1

Consistent with our experimental data, while rates of partisan voting are lower in nonpartisan elections, the difference between partisan elections and nonpartisan elections is not statistically significant.  Nonpartisan elections are not effective at shielding the partisanship of candidates from voters.

What does all this mean?  At least where nonpartisan elections involve reasonably high amounts of campaign spending and involve important policymakers, they are likely to be just as partisan as a regular partisan election.  Candidates have an incentive to signal to voters what their partisan affiliation is, and there is ample opportunity in a highly competitive environment like state supreme court elections for candidates to do exactly that.  Our results may not hold for more “sleepy” nonpartisan elections, but the more contentious the race, the less likely nonpartisan elections are to be effective institutions.

This article is based on the paper, “Party Identification and Vote Choice in Partisan and Nonpartisan Elections,” which appeared in the October 2013 issue of Political Behavior.

Please read our comments policy before commenting. 

Note:  This article gives the views of the authors, and not the position of USApp– American Politics and Policy, nor of the London School of Economics.

Shortened URL for this post:


About the authors

Chris Bonneau 80x108Chris W. Bonneau – University of Pittsburgh
Chris W. Bonneau is associate professor of political science at the University of Pittsburgh. He primarily studies American politics with an emphasis on judicial politics and public law.


Damon M Cann 80x108Damon M. Cann – Utah State University
Damon M. Cann is an associate professor of political science at Utah State University. His research interests center around elections (Congressional, Judicial, and Presidential) and education policy methodology.