Current Snapshot of RCV Use in the United States

As of April 2020, almost 10 million voting-age adults live in U.S. jurisdictions that use RCV, or have adopted RCV and plan to implement it for their next round of elections.

Since 2004, when San Francisco first began using RCV to elect municipal offices, there have been 367 RCV elections across 18 jurisdictions. The graph below depicts the number of contests in each of the jurisdictions that presently use RCV or which plan to implement RCV in their next elections.

https://e.infogram.com/242bab36-c0f1-4d06-8aeb-933467b8bfa9?src=embedRCV Electionsnoborder:none;allowfullscreen55014280

Of the 23 jurisdictions presently or imminently using RCV to elect offices, 17 exclusively use single-winner RCV (instant runoff voting), while Cambridge (MA) and Eastpointe (MI) exclusively use multi-winner RCV (single transferable vote); Minneapolis (MN) and Palm Desert (CA) use a combination of single and multi-winner RCV, and Payson (UT) and Vineyard (UT) use a combination of single-winner RCV and a form of RCV called preferential block voting. See the glossary for more information on these methods.

Voter Turnout and Participation Under RCV

Greater participation in our democracy is highly desirable so that our government is truly constructed "by the people." This section explores research on the effects of RCV on two key aspects of participation -- voter turnout and voter engagement.

Voter Turnout

By giving voters more meaningful choices and reducing the number of wasted votes, ranked choice voting could increase voter turnout. On the other hand, some argue that RCV could depress turnout because it imposes a greater cognitive burden on voters (ranking rather than indicating a single preference). The answer to this question is still open. Here’s a roundup of available research so far.

• A study by Professor David Kimball at the University of Missouri-St. Louis, and Ph.D. candidate Joseph Anthony, finds that, on the one hand, RCV in American local elections has a limited impact on turnout. More important factors include a competitive mayoral election, other races on the ballot, and the use of even-year elections. On the other hand, the Kimball and Anthony study shows that, when compared to the primary and runoff elections they replace, RCV general elections are associated with a 10 point increase in voter turnout.

Find FairVote’s one-page summary here and a working version of the Kimball and Anthony paper below. 

Kimball, D & Anthony, J. October 2016. Voter Participation with Ranked Choice Voting in the United States.

• A further result from the Kimball and Anthony study showed that turnout disparities between high- and low-income wards were as prevalent before the adoption of RCV as after. While RCV did not ameliorate demographic inequities in turnout, it also did not exacerbate them in Minneapolis.

• San Francisco had a highly competitive special election for Mayor in June 2018, which was combined with statewide primaries for governor and senator. The ballot in San Francisco included an RCV race for mayor, and non-RCV races for statewide offices.  More San Franciscans participated in the RCV mayoral election (250,868 votes cast for Mayor) than in the non-RCV primaries at the top of the ballot (244,137 for Governor and 237,261 for U.S. Senator), demonstrating that a competitive RCV election can drive turnout. 

• FairVote examined turnout in the 6 largest U.S. cities using RCV. Our analysis showed strong turnout in RCV races compared to races before RCV implementation. This analysis did not attempt to control for other factors, such as competitiveness of races on the ballot, which could drive turnout.

https://e.infogram.com/040bd6df-7995-48f7-a5dd-e24be579dd24?src=embedMayoral Turnout Handout 6 Citiesnoborder:none;allowfullscreen6127920

• FairVote maintains a database of turnout in RCV election. Contact us for more information.

Voter Engagement in the Democratic Process

• In RCV cities, candidates are more likely to reach out to voters in-person than in cities that do not use RCV. Additionally, voters in RCV cities were more likely to discuss politics with their families, friends or co-workers than voters in cities that do not use RCV. 

Smith, Haley. June 2016. Ranked Choice Voting and Participation: Impacts on Deliberative Engagement.

Further reading: 

• Elliot Louthen explores the findings of David Kimball and Joseph Anthony's study in Ranked Choice Voting in Practice: Analysis of Voter Turnout in RCV Elections (RCV Civility Brief #8, November 2015).

• A quick one page summary of  David Kimball and Joseph Anthony's study: The Adoption of RCV Raised Turnout 10 Points.

• FairVote, Key Facts about the Use of Ranked Choice Voting in 2014 in California’s Bay Area (April 2015)

• Burnett, Craig M., and Vladimir Kogan. "Ballot (and voter) “exhaustion” under Instant Runoff Voting: An examination of four ranked-choice elections." Electoral Studies 37 (2015): 41-49.

This section examines how voters interact with the ranked ballot, and how many voters end up electing a candidate they had ranked highly.

Consensus Value:

We define "consensus value" as the proportion of voters who ranked the winner as their first, second, or third choice. We use consensus value as a measure of how much support the winning candidate garnered from the community as a whole. This measure is intrinsic to RCV and provides valuable information on how many voters found a winning candidate acceptable. 

• Over 55% of winning candidates had the consensus of two-thirds of voters

• Almost 25% of winning candidates had the consensus of three-fourths of voters.

• In all races for which we have enough data to determine consensus value, 67% of ballots ranked a winning candidate in their top three.

https://e.infogram.com/912863e4-2b19-4f6b-aa14-95f0a49cb66c?src=embedTestnoborder:none;allowfullscreen5505550

For more details on consensus value, see our spreadsheet of consensus values for single-winner RCV races.

Ballot Use

The benefits of RCV are only applicable to the extent that voters use the ranked ballot effectively. Effective ballot use means ranking candidates beyond a first choice where doing so is advantageous, and avoiding errors which result in ballot invalidation. 

All ballot types result in some number of errors in voting. In single-choice elections, ordinarily only overvotes - votes invalidated due to a voter attempting to vote for more than one candidate - count as ballot errors. In RCV elections, voters may make different kinds of deviant marks, including ranking the same candidate multiple times, skipping rankings, or including overvotes at later ranking orders. However, most of these do not impact the vote being counted as the voter intended. Only first-round overvotes can be fairly compared with errors in single-choice elections, since those are the only errors that, in both systems, invalidate the ballot entirely.

This section examines research into voter error as a measure of voter participation in the election process. 

• A 2016 study by Professor David Kimball at the University of Missouri-St. Louis, and Ph.D. candidate Joseph Anthony, assessed the rates of “residual votes”, which include overvotes and undervotes, as a measure of both voter interest and voter error. The study shows that, in the 26 cities studied, the adoption of RCV was not associated with any change in the number of residual votes. Find FairVote’s one-page summary here and a working version of the Kimball and Anthony paper below. 

Kimball, D & Anthony, J. October 2016. Voter Participation with Ranked Choice Voting in the United States. 

• A 2015 study by Francis Neely and Jason McDaniel shows that overvotes (a type of voter error in which the voter selects too many candidates) are often more common in precincts with more African-American, Latino, elderly, foriegn-born, and less wealthy citizens. However, the pattern of overvoting is similar in both RCV and non-RCV contests. This suggests a need for greater voter education in general, rather than a larger cognitive burden stemming from RCV itself. 

Neely, F & McDaniel, J. 2015. Overvoting and the Equality of Voice under  Instant-Runoff Voting in San Francisco.

• A 2015 study by Craig Burnett and Vladimir Kogan showed ballot exhaustion leading to winners elected without securing a majority of first-round votes. The authors correctly note that many exhausted ballots result from jurisdictions limiting the number of rankings to three. 

Burnett, C & Kogan, V. 2015. Ballot (and voter) “exhaustion” under Instant Runoff Voting: An examination of four ranked-choice elections.

FairVote’s own analysis shows that 18% of RCV races in the US have resulted in a winner who did not secure a majority of first-round votes. This drops to 13% when only considering races in which the voter was allowed to rank as many candidates as they choose. FairVote recommends jurisdictions allow voters to rank as many or as few candidates as they like. Read more on majority winners in the Who Wins RCV Races section.

Voter Support and Understanding of RCV

It is absolutely crucial that voters understand how to interact with a ranked ballot, understand their electoral system, and are able to cast a meaningful vote for a candidate of their choice. This section examines how well voters understand RCV and voters’ level of satisfaction with RCV. 

Voter Support for RCV

• Maine used RCV for the first time for statewide elections in 2018. An exit poll after their November 2018 general election showed 60.9% of respondents in favor of keeping RCV or expanding use of RCV. 

• 94% of Santa Fe voters reported feeling “very satisfied” or “somewhat satisfied” with their first use of RCV in 2018, as reported in an 2018 exit poll by FairVote New Mexico.

• In Minneapolis in 2017, 66% of voters reported support for continued use of RCV, compared with only 16 percent who said they do not support its use and 18 percent who were unsure. 

• Find additional details from exit surveys in the following report:

FairVote. 2018. Exit Surveys: Voters Evaluate Ranked Choice Voting.

Voter Understanding of RCV

• Over 90% of Maine voters reported that their experience using RCV was either “excellent” or “good”. This poll was an online survey conducted by the League of Women Voters after the June 2018 primary elections, which was the first time most respondents had used RCV. 

• Santa Fe voters overwhelmingly reported they were not confused by their ballot after their first RCV election. More than 67% of respondents said the ballot was not at all confusing. 

• Voters in Minneapolis have been using RCV since 2009 and continue to report positive results. 92% of respondents said they find voting with RCV “simple”.

• Two cities in Utah used ranked choice voting for city council races for the first time in 2019. Voters overwhelmingly utilized the option to rank candidates, even in their first use of RCV. A majority of voters in both Payson and Vineyard fully ranked their ballots. 

• As part of the surveys conducted by the Eagleton Poll in 2013-2014, likely voters in cities using RCV were asked, "When you voted in the recent election, how easy was it to understand the voting instructions?”. In these two surveys, 90% and 89% of respondents say they found the RCV ballot easy to understand. This level of understanding remained high across demographic and socioeconomic groups. 

John, S. & Tolbert, C. April 2015. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area.

• In the 2014 Eagleton Poll in California, self-reported understanding of RCV compares favorably to understanding of plurality and the top-two primary. 
  • The percentage of voters in RCV cities who understood RCV at least “somewhat well” (84%) was roughly equivalent to the percent of voters in plurality cities who understood plurality (83%). 
  • More respondents (49%) in RCV cities reported understanding RCV extremely or very well than reported understanding the top-two primary extremely or very well (40%).

John, S. & Tolbert, C. April 2015. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area.

• A 2019 study by Todd Donovan, Caroline Tolbert, and Kellen Gracey analyzed survey data to examine whether there are racial disparities in voter understanding. 
  • In terms of understanding voting instructions for RCV, there were “no differences… between whites and people of color.” 
  • There were “no differences in RCV cities in how whites, African Americans, and Latinx respondents reported understanding” the system. 
  • Read FairVote’s highlights here or find the full study at:

Donovan, T., Tolbert, C., & Gracey, K. April 2019. Self‐Reported Understanding of Ranked‐Choice Voting.

Further reading: 

• Andrew Douglas explores the findings from the 2013 survey  in Ranked Choice Voting and Civility: New Evidence from American Cities.

• Professor Caroline J. Tolbert examines voter experiences in the 2013 election season in Tolbert, Caroline J. Experiments in Election Reform: Voter Perceptions of Campaigns Under Ranked Choice Voting vs. Plurality Voting. Paper prepared for the Workshop on Electoral Systems, Electoral Reform, and Implications for Democratic Performance. Stanford University, March 14-15, 2014.

RCV and Representation

Different voting systems may impact the representation of the full spectrum of voter opinions, experiences and interests. This page explores research into the representation of different groups under ranked choice voting and other American election systems.

Representation for Women and People of Color

• A 2018 paper by Sarah John, Haley Smith, and Elizabeth Zack shows that California cities which adopted RCV saw an increase in the percentage of candidates of color running for office, and increases in the probability of female candidates and female candidates of color winning office. (The author refers to the reform as “the alternative vote” or “AV” which is synonymous with ranked choice voting.) Learn more from this write-up by author Sarah John. 

John, S., Smith, H., & Zack, E. August 2018. The alternative vote: Do changes in single-member voting systems affect descriptive representation of women and minorities?

• A 2018 FairVote report on racial minority voting rights shows that people of color hold office at a higher rate under RCV than under the prior system, and that people of color win office more often since the adoption of RCV. 

FairVote. May 2018. Ranked Choice Voting and Racial Minority Voting Rights.

• In 2014, FairVote's Andrew Douglas used ballot image data to show a pattern of racially and ethnically cohesive voting among Cambridge, MA, city council voters. The report demonstrates that at-large RCV benefited candidates from ethnic and political  minority groups. The benefit is attributed primarily to the low electoral threshold which allows minority groups to have representation based roughly on their share of the population. 

Douglas, Andrew. February 2014. The Effect of Fair Representation Voting on 2013 Cambridge Municipal Elections

• A 2016 Fairvote report examines the effect of RCV on women and people of color running for elected office in the California Bay Area. The findings reveal that RCV increases descriptive representation for women, people of color, and women of color. Some reasons for RCV’s positive effects can be related to how often it replaces low, unrepresentative, turnout elections and that it allows for multiple candidates appealing to the same community to run without splitting the vote.

John, S., Smith, H., & Zack, E. July 2016. Ranked Choice Voting and Representation of Underrepresented Groups.

Representation for Differing Political Viewpoints

• Roughly 40% of voters self-identify as Independent according to Gallup polling. In spite of their numerousness, unaffiliated voters have little influence in Congress or in most state legislatures. 

• RCV impacts the prospects for independent and third party voters and candidates. While single-winner RCV might not increase the election of minor party candidates in the US, as candidates must clear a majority threshold in order to win, RCV allows supporters of those candidates to sincerely rank their preferred candidate first without feeling like their votes are wasted and with minimal chance that support will spoil the election outcome.

• In multi-winner RCV, it also becomes possible for Democrat or Republican voters who live in a district with the opposite majority to gain representation. As long as the Democrat or Republican population is equal to or greater than the threshold to win, people can gain representation where they currently feel left out.

• A 2016 FairVote report explores how RCV might reduce legislative polarization by allowing space for moderate, conservative, liberal and other voters to elect candidates in proportion to their overall numbers in the electorate. Evidence for Cambridge, Massachusetts, which uses multi-winner RCV, indicates that candidates and city councilors are not highly polarized there. 

John, S. & Leinz, B. April 2016. Polarization and Multi-winner Ranked Choice Voting in Cambridge, Massachusetts, City Council Elections.

Who Wins RCV Races?

In this section we explore how RCV has worked in practice in the United States. We summarize research on how RCV has worked to remove the spoiler effect and on the tendency of RCV in single-winner seats to elect candidates with majority support. We also assess RCV in practice on some more technical grounds, namely monotonicity and its tendency to elect Condorcet winners.

Majority Winners

A key reason we need RCV is that our current system often elects winners with less than a majority, or less than 50% support.

• Since 1992, 49 senators from 27 states have been elected with less than 50 percent support. 

Harrow, J. & Shi, V. 2019. The magic of majority rule in elections. The Hill.

• Primary elections are also often decided with small pluralities. The chart below shows plurality wins in primary elections for U.S. House of Representatives races in 2018. In some highly-contested races, winners received as little as 22% of the vote -- meaning 78% of people voted against them. 

https://e.infogram.com/f4ff3f81-be7b-46c2-8b40-fa5850fdaeef?src=embed2018 Plurality Winsnoborder:none;allowfullscreen5507460

• There have been 344 single-winner ranked choice elections in the U.S. since 2004. In 241 of these (or 70%), a majority winner was identified in the first round. The remaining 103 races went into the instant runoff before declaring a winner.

• Sometimes, the winner of a single-winner RCV election does not have a majority of total votes cast. A winner is declared when a candidate has a majority of votes which are active in that round of counting, which excludes ballots which have become exhausted. This has happened in 63 out of 344 elections, or 18% of the time. As noted by Burnett and Kogan, ballot exhaustion is sometimes due to jurisdiction limiting the maximum number of rankings to three. You can learn more about the majority criterion with RCV in our FAQ.

RCV and Incumbents

• RCV appears to have a neutral effect on incumbents. A FairVote report on incumbency in the Bay Area of California shows that incumbents are re-elected at the same rate both with and without RCV. 

FairVote California. October 2017. RCV and Incumbency in the Bay Area.

Come-From-Behind Winners

• There have been 15 RCV races in the U.S. which were won by a candidate other than the first-round leader. That’s 4.4% of the 344 single-winner RCV races since 2004.

• Of those 15 races, 14 were won by the candidate who began in second place. One race was won by the candidate who was in third place in the first round.

• The chart below shows all come-from-behind victories in the U.S. since 2004:

Technical Property in Practice: Condorcet Winners

The Condorcet criterion states that the candidate who would win a one-on-one matchup against every other candidate should win the election. If such a candidate exists, they are known as the “Condorcet winner”. Ranked choice voting does not guarantee that the Condorcet winner will win the election, but it does make it more likely than single-choice plurality or two-round runoff elections.

• We cannot know for sure how often single-choice plurality elections or two-round runoff elections elect Condorcet winners, because we do not know voters’ back-up choices.

• Of the 322 single-winner RCV elections in the United States since 2004 in which we have sufficient ballot data to assess whether the Condorcet winner won the election, 321 elections were won by the Condorcet winner.

• We have identified only one public RCV election in the United States in which the Condorcet candidate lost: Burlington’s 2009 mayoral election. In that election, the three strongest candidates, once all others had been eliminated, had 37%, 34%, and 29% support. The candidate in third place, Andy Montroll was eliminated. However, analysis of ballot image data from the election showed that supporters of the top two candidates had each overwhelmingly ranked Montroll before the other, making him the Condorcet candidate. Whether it would have been better for Montroll to win the election despite attracting so little core support may be disputed, but it is certain that he would have also lost election under a two-round runoff or a single-choice plurality system, reinforcing that RCV elects Condorcet winners more often than either of those systems. 

• In the rare situation in which RCV would not elect the Condorcet winner, that necessarily means that the Condorcet winner attracted too little core support to come in either first or second place in the final round. 

Technical Property in Practice: Monotonicity

Monotonicity in ranked voting means that ranking a candidate lower can never help them, and ranking a candidate higher can never hurt them. Any voting method in which votes are counted in rounds has the possibility of a non-monotonic outcome, including RCV and two-round runoff elections.

We have not identified any RCV election in which any group of voters has attempted to exploit the possibility of non-monotonicity for strategic purposes. Doing so successfully would both require a highly unusual set of circumstances, and a detailed and accurate understanding of how the electorate will rank the candidates. Because this is prohibitively difficult, the issue of monotonicity under RCV is largely academic - it has never had any impact on any RCV campaign and is unlikely to have any impact in the future. Learn more about monotonicity in our FAQ.

• Out of 344 RCV elections having taken place in the United States as of April, 2020, we have identified one contest with a possibly non-monotonic outcome: the 2009 mayoral race in Burlington, VT. 

Whether this election constitutes a non-monotonic outcome depends on how strictly the criterion is defined. There was no candidate that could have won merely by voters ranking him lower. However, if between 367 and 589 Wright voters had instead ranked Kiss first (without changing any other ballots), Wright would have been eliminated instead of Montroll, and Montroll would then beat Kiss. In other words, there is a group of voters that could have caused Kiss to lose by ranking him higher. However, this would be Wright voters helping to elect Montroll by ranking Wright lower. No group of voters could have elected their own preferred candidate by ranking that candidate lower.

Ranked Choice Voting and Civil Campaigning

RCV encourages more civil discourse between candidates because candidates campaign not only for first-choice support, but also the second-choice support of other candidates. Consequently, candidates need to appeal to a broader range of voters in order to win. Candidates have less incentive to make negative statements about their opponents because they risk alienating that opponent's supporters. 

• The Eagleton Poll surveyed voters in 2013 and 2014 about their perceptions of the tone of campaigning. The study found likely voters in RCV cities were more satisfied with the conduct of candidate campaigns and perceived less candidate criticism and negative campaigning compared to voters in non-RCV cities. 
  

John, Sarah and Tolbert, Caroline. April 2015. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area.

The survey methodology document for the Eagleton Poll is also available.

• Fine-grained analysis by socio-economic and demographic groups is possible for the California 2014 portion of the Eagleton Poll results. Virtually every demographic group studied – including low-income respondents, college graduates, Latinos, African-Americans, women, Independents and unmarried people – reported less negativity in RCV cities than in plurality cities.
  

• A 2013 analysis of sentiment in press coverage showed that articles in RCV cities were 85% more positive than negative, compared to only 77% of articles in non-RCV control cities. 

• Fairvote published the following reports in our Ranked Choice Voting Civility Research Report series: 

1. Douglas, Andrew. April 2014. Ranked Choice Voting and Civility: New Evidence from American Cities. 

2. John, Sarah.  February 2015. Ranked Choice Voting in Practice: Candidate Civility in Bay Area Elections, November 2014. 

3. John, Sarah. February 2015. Ranked Choice Voting in Practice: Content Analysis of Campaign Tone in Newspapers and Twitter Feeds in 2013 RCV Elections.

4. John, Sarah and Tolbert, Caroline. April 2015. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area.

RCV Compared to Two-Round Runoff Elections

Ranked choice voting (RCV) and two-round runoff races are both multi-round systems that aim to promote majority support. Under these systems, a candidate reaches a majority once they have more than 50 percent of votes in the final round.

• Under two-round runoff elections, voters are asked to return to the polls and vote a second time. RCV accomplishes the same goal in a single election. Because two-round runoffs require two elections, they are more expensive. 

• Two-round runoff elections can also result in a different group of voters participating in the final round than the first one: generally, fewer voters overall, and a less representative group of voters. Under RCV, the same group of voters can participate in every round. Some voters, however, may only rank some of the candidates. If each candidate ranked is eliminated during the count, such a ballot becomes inactive. Even when taking into account the drop-off in voters between rounds in an RCV election, RCV still outperforms two-round runoff elections both in final round turnout and representativeness of the final round.

• The table below compares RCV or "instant runoff" races with delayed runoff races. It analyzes election results from 84 RCV races over the past two decades from 10 cities (San Francisco, Oakland, San Leandro, and Berkeley, California; Minneapolis and St. Paul, Minnesota; Santa Fe, New Mexico; Takoma Park, Maryland; Telluride, Colorado and Portland, Maine) and the State of Maine’s 2018 Democratic gubernatorial primary and congressional primary and midterm elections. These are compared with the 221 congressional primary runoffs stretching back to 1994, the 22 statewide runoff elections held in 2018, and the 14 runoff races which took place in San Francisco from 2000-2003, before the city started using RCV.

https://e.infogram.com/c078945c-c1bc-4fb9-ad22-89e9785ead2e?src=embedRanked Choice Voting vs Runoffsnoborder:none;allowfullscreen8007580

As intended, both runoffs and RCV resulted in every winner having a majority of active votes in the final round. However, with RCV, the winner’s median share of the vote in the final round was 48.8 percent of the first round vote, as compared to delayed runoff winners’ median of 37.2 percent in congressional primary runoffs, 34.3 percent in the San Francisco runoffs, and 36.4 percent  in statewide runoffs.

• The difference between RCV races and runoff races is even more evident when measured to 40 percent of the first round vote, the traditional measure of a “substantial plurality.” In the final round, more than 96 percent of RCV winners received 40 percent or more of the first round vote, while only 38 percent of the congressional primary runoff winners, 36 percent of San Francisco’s pre-RCV runoff winners, and 27 percent of statewide runoff winners achieved 40 percent of the first round vote. More than half of San Francisco’s 14 runoff winners earned fewer votes in the runoff than in the first round - a loss of support that cannot happen with RCV.

• Not only does turnout tend to decrease by a much greater amount in runoffs than in RCV elections, but the proportion of people of color voting also decreases to a greater degree. While average overall turnout decreases for all demographic groups in runoff elections, the decrease is greater for people of color, changing the demographic composition of voters from the first election to the runoff. 

infogram_0__/Q4k5GKR5jveebKqa7AusDemographics: Ranked Choice Voting vs Runoffshttps://e.infogram.com/js/dist/embed.js?piRtext/javascript

Disclaimer: 

We determined the demographics and voter turnout statistics for each district using voter data from L2. For estimated Citizen Voting Age Population (CVAP) L2 uses weighted census data. For voter turnout, L2 uses surname classifications. While this method is somewhat blunt, it does provide a workable estimate for voter turnout by ethnic group in prior elections. L2 groups voters into ethnicities and groups ethnicities into broad ethnic categories:

• European (including those classified as Portuguese);
• Asian (including East and South Asian but not Pacific Islanders);
• Hispanic (not including Portuguese);
• African American (likely);
• Other including Pacific Islanders and certain categories of Middle Easterners. 

Additional methodology note for L2 VoterMapping: L2 has separate data by party for primaries, but not for runoffs. For example, data is available for everyone who participated in a Republican primary, but runoff data does not distinguish who participated in which party’s runoff. This is a problem because many of the runoff states have open primaries (meaning that just restricting voters by party in the runoff, as L2 allows you to do, doesn’t offer an accurate picture). To get around this issue, both primary and runoff data is filtered for the party in question. For example, if 1,000 Republicans participate in the Democratic primary, those Republicans will be excluded in the runoff (regardless of whether they turned out for the runoff or not) because of the filter for the Democratic party. In order to not skew the data, those 1,000 Republicans must also be excluded from the primary data. This is especially necessary when looking at changes in turnout by demographics because the different parties have different demographic compositions, so including different party's voters in the primary but not in the runoff would have a large impact on demographic data. Most elections don’t see many people voting in other parties’ primaries, so the exclusion should not have a large impact on data.