The history of electoral reform in America is littered with people who eagerly made grand claims about how their preferred solution would cure what ailed American democracy. Usually this involves “liberating” the sanctified voters from the dastardly efforts of politicians and political parties to suppress the “true” preferences of the electorate. Too often, the fruit of these reform efforts has been the creation of electoral processes that reduce voter engagement and maintain the status quo. Rob Richie’s arguments in favor of Ranked Choice Voting (RCV) echo many of the same arguments that reformers have been making for over a hundred years. Unfortunately, pro-RCV arguments are similarly based on faulty assumptions about voters, and too often they ignore or inappropriately minimize empirical research that highlights the potential for negative consequences.
In evaluating electoral reforms, I am guided by a clear baseline principle: changes that expand access to the vote and encourage more political participation tend to be better than those that seek to restrict voting or make it more difficult. Based on the results of my research into the impacts of Ranked Choice Voting in city elections, this particular electoral reform fails on both counts. RCV makes voting more complicated, which leads to several negative consequences for the level and quality of voter participation in elections.
Equality of Electoral Voice: Lower Turnout and More Ballot Errors
Under RCV, overall voter turnout does not increase, and is likely to decrease significantly, especially among those segments of the electorate that are already least likely to participate. I examined voter turnout in five San Francisco elections from 1995 to 2011, the last two of which featured Ranked Choice Voting. My main topic of inquiry was whether variation in turnout across racial lines was related to the adoption of RCV. According to the results of my analysis of over 2,500 precincts across the five elections, turnout declines among African-American and white voters was significantly correlated with the adoption of RCV. Additionally, I found that the adoption of RCV exacerbated disparities in voter turnout between those who are already likely to vote and those who are not, including younger voters and those with lower levels of education.
In additional research, I analyzed voter turnout in nonpartisan mayoral elections in RCV cities compared to similar elections in non-RCV cities. The results show that the impact of RCV on voter turnout depends upon whether elections occur during odd years or even years. In odd-year elections with Ranked Choice Voting, voter turnout decreases about eight percentage points, on average, compared to a non-RCV general election at the same time. Whereas in even-year elections RCV has little or no effect on voter turnout. Even-year elections usually coincide with congressional or presidential elections, and they generally have much higher voter turnout than those that occur in odd years. One explanation for this is that during odd-year elections it is very difficult to get low-propensity voters to the polls. By making voting more complicated, RCV exacerbates this tendency, making it less likely that new and more casual voters will enter into the process.
Another major concern with respect to equality and integrity of the electoral process under Ranked Choice Voting is how it affects the tendency of voters to make errors when marking ballots. Research indicates that when voting is made more complicated, for example through ballot design or the presence of many candidate options, ballot errors increase. My colleague Francis Neely and I analyzed almost two million individual ballots in order to measure the incidence of errors that disqualify a ballot from being counted after the adoption of Ranked-Choice Voting in San Francisco elections. We found that such errors were significantly more common in RCV elections than plurality elections. The rate of errors was comparable to electoral situations that involve either very complex voting tasks or poorly designed ballots. To be clear, it is the additional complexity of voting under Ranked Choice Voting, not RCV itself that causes more ballot errors. Nonetheless, it is highly likely that implementation of RCV will result in higher rates of ballot errors that cause individual ballots to be disqualified. These ballot errors will be concentrated among those portions of the electorate who are already most vulnerable to being underrepresented.
Polarization and Voter Confusion
An underlying theme of Rob Richie’s argument is that Ranked Choice Voting will help to alleviate some of the problems related to partisan polarization. The idea is that RCV will encourage more moderate and/or third party candidates to run for elective office, and that voters will not be forced to choose the “lesser of two evils.” While this argument seems to make a certain amount of sense, it rests on the faulty assumption that voters are capable of consistently and accurately distinguishing the ideological and policy positions of candidates for whom they vote, absent clear partisan and ideological cues to guide them. Unfortunately, implementation of Ranked Choice Voting will most likely increase voter confusion.
My research into racial group polarization in nonpartisan urban elections illustrates the problem. I found that there was a reduction in the polarized preference gap between different racial groups. However, this reduction in racial polarization was most likely caused by voter confusion about how candidates will represent their interests rather than any sincere expression of preferences previously suppressed by plurality systems. This result is consistent with analysis by Corey Cook showing that voters’ candidate preference rankings in RCV elections are highly inconsistent, lacking any well-ordered ideological or policy-based structure. Research into the Top Two primary system in California indicates that voters are not able to reliably identify candidates’ ideologies or policies absent strong partisan cues. Because of this, the new system has failed to achieve its supporters’ goals of increasing the chances of ideologically moderate candidates and reducing polarization.
While there is some debate among political scientists about the extent and causes of partisan polarization, there is little doubt that voters are highly polarized from each other. Given the reality of polarized voters, there is little reason to believe that a change to how people vote will result in the election of politicians who are less polarized.
Conclusion: First Do No Harm
The results of my research provide reasons for skepticism about whether the benefits of RCV outweigh the potential costs. However, my research has been limited to nonpartisan city elections, and therefore the findings may not generalize to partisan state or federal elections in which candidates may be better known and voters may be more likely to participate. Additionally, other researchers have documented some positive aspects of RCV. First, voters who experience Ranked Choice Voting tend to express satisfaction with the process and confidence in their ability to understand it. Second, based on high-quality research by other political scientists, it is reasonable to expect that the level of negative campaigning may decrease under RCV.
Despite these positive aspects, and given the unlikelihood that RCV will contribute to a reduction in partisan polarization, I believe that the potential for lower turnout and more ballot errors outweighs the potential benefits of Ranked Choice Voting. Rather than expand the electorate, reduce turnout inequalities, and ameliorate polarization, it is quite likely that the adoption of Ranked Choice Voting for state and federal elections around the country will further entrench the status quo.