A renaissance in voting reform

The version of Ranked Choice Voting in use today was invented in 1870 as a simplification that made tabulation of ranked ballots easier and faster. Previously, when counting ranked ballots, people would look at all the rankings to determine the candidate who was preferred over all others in head to head match-ups. While the old way was (and is) very accurate, it was hard to count.

In comparison the Instant Runoff Method was a breeze. Rather than counting all the rankings, candidates would be eliminated in tournament style rounds, and as candidates were eliminated votes would transfer from one to the other. Ballots would shuffle from one stack to the next, and at the end the candidate with the tallest stack of ballots was the winner.
It wasn't perfect, but the gains in logistical viability were considered worth it by many. By the early 1900s what we now know as Ranked Choice Voting had caught on, and was in use in countries like Australia and Ireland, as well as a few cities in the US. 

 

Modern Voting Theory

The age of computers changed everything. Modern simulations, modeling, and statistics revealed new information on exactly how pervasive known issues and unrepresentative outcomes with Ranked Choice actually were. (Votes can backfire in 15% or more of competitive elections with 3 or more candidates.) For decades RCV and its multi-winner version were repealed in more places than they were adopted, and the Choose-One voting method reigned, despite the fact that every single election science study showed that it was the absolute worst. 

Many reformers, academics, and researchers started advocating for alternative voting methods like Approval and Score Voting, as these methods didn't waste votes and consistently outperformed RCV, especially in competitive races. In particular, studies from the year 2000 had shown that a score voting primary followed by a top-two general election outperformed all other methods tested. But there were drawbacks to that as well, and neither side was willing to compromise with so much at stake. The reform movement was divided, and the result was that the reform movement floundered. On the one side were reformers who liked ranking methods, on the other were preference voting fans. Enter STAR Voting. 

 

Score Voting + Instant Runoff = Score Then Automatic Runoff = STAR!

In 2014 at the Equal Vote Conference in Eugene, Oregon, STAR Voting was invented. A hybrid of the previous proposals, STAR Voting delivers on the goals of reformers on both sides while addressing valid concerns with the older proposals. The 5 star ballot allows voters to show their level of support and allowed the ballots to be tallied using simple addition. The Instant Runoff ensured that even if a voters favorites couldn't win, their vote would still make a difference. It was a breakthrough!

As predicted, studies found that the new method topped the charts for both accurate representative outcomes, and also resilience to strategic voting. Finally, a win-win proposal that everyone could get behind! 

 

Traditional Ranked Choice or Version 2.0?

 

RCV and STAR have a lot in common:  Only one election is needed, primaries are optional. Both are user friendly. Both can be used for single winner, multi-winner, or proportional representation elections ...

… That said, there are significant differences. 

 

#1: Ranking vs Rating:

Technically "Ranked Choice Voting" or RCV is an umbrella term for any number of ranked ballot voting methods, however in practice RCV is Instant Runoff Voting, the method discussed below. STAR Voting is often referred to as RCV 2.0.

 

  • Ranked Choice uses a ranking ballot

    Rank your candidates first choice, second choice, third choice and so on. This allows voters to show their preference order, but it doesn't show how much or little a voter may support their candidates. For example, a 2nd choice candidate may be just as good as the 1st choice, or they may be a lesser evil who is barely better than the worst-case-scenario.

    Full rank ordering can take up a lot of space on a ballot, so in many jurisdictions the number of candidates which can be ranked is limited to 3, 4, or a fixed number. Alternately, some elections may allow voters to write in a number rather than filling in a bubble, but this can result in ballots being discarded due to handwriting and optical scanning issues.

    Ranked Choice Voting doesn't allow voters to give multiple candidates the same ranking, and ballots with these kind of errors are considered "spoiled" and must be thrown out.

    Cognitive load theory has demonstrated that while people like to have choices, when the number of options is higher than 7 or so most people start to feel overwhelmed, confused, and have less clear preferences. Ranked Choice elections with larger numbers of candidates tend to see increased rates of voter error leading to spoiled and exhausted ballots which are not counted in the deciding round of the election. 

  • STAR Voting uses a 5 star ballot

    Score candidates from 0 up to 5 stars. A star ballot allows voters to show their level of support for each candidate, in addition to their preference order.

    The 5 star rating is ubiquitous and familiar to people all over the world. In the private sector it's the go-to for measuring a nuanced opinion. The 0-5 scale takes up a fixed amount of space on a ballot regardless of the number of candidates, which is important for ballot formatting and election officiation. 

 

#2: Two Rounds vs Many Rounds of Tabulation:

Both RCV and STAR use built-in runoff rounds to determine the winner, and voters only need to vote once. STAR voting has two rounds of tabulation with a scoring round and an automatic runoff, RCV may have as many runoffs as there are candidates, minus one. 

  • In RCV, otherwise known as Instant Runoff Voting, the final contestants are determined by successively eliminating the candidate with the fewest first-choice votes amongst the voters. With each round of elimination, the votes of those with the eliminated candidate in top position transfer to the next choice on their ballot, if possible, this continues until one candidate is top-ranked on a majority of ballots that remain. 

    When ballots are hand counted, RCV tabulation makes sense, but without acting out the process it's easy to miss some of the implications of this kind of process. Voters whose 1st choice is eliminated early will always have their 2nd choice counted, but voters whose down ballot rankings were eliminated before their 1st choice will have nobody left for their vote to transfer to. These are known as 'exhausted' ballots. In competitive elections voters who prefer strong underdog candidates more likely to have their ballots exhausted, puts them at a disadvantage. This can result in counter-intuitive and sideways election results. 

    When voting machines are used to tabulate elections, this type of ballot sorting requires more complex code, and the multi-round process means that ballots must all be physically or digitally compiled in one place for counting to proceed beyond the first round. Tabulation of RCV elections can't begin until all ballots are in at the end of election day and election results can't be broken down at the precinct level because its unknown which rankings will ultimately be counted and which will not. This presents issues for election security and risk-limiting audits as well. 



  • In STAR, other wise known as Score Then Automatic Runoff, there are always two rounds of tabulation. A scoring round, and a runoff. The two highest-scoring candidates are finalists, and the finalist preferred by the majority wins.

    In the STAR Voting runoff, if a ballot gave both finalists 5 stars, that is counted as a "vote of no preference" between those two candidates, but it's important to note that this is not a wasted vote. It's a vote which was counted, and the scores given made a difference helping those candidates beat out the competition. This is not the same as a spoiled or exhausted ballot, which is not counted but which may have actually made a difference if it had been considered in the deciding round of the election. 

    STAR voting is counted using addition. In the scoring round the stars are totaled. In the runoff round the votes for each finalist are totaled. All ballot data is counted and all data is used. This means that STAR Voting doesn't require new voting machines in most cases, and counting can begin even if all the ballots aren't in yet. 

It's important to note that in some elections a candidate supported by a majority may not exist. In other cases there could be multiple candidates who are supported by a majority of the voters. Preferences can even by cyclical, like in a rock-paper-scissors three way tie. The important thing is that the voting method finds the candidate with the most support overall, the candidate who best represents the will of the people. 



#3: Majority preferred winners vs a majority of remaining ballots: 

In order to guarantee majority preferred winners, voting methods need to narrow it down to the top 2 finalists. RCV and STAR Voting use different methods for determining these finalists. 

  • RCV elects a winner who was preferred on a majority of remaining ballots in the final round of tabulation, but not all ballots are counted in the final round. 
  • STAR counts all ballot data. Of voters who had a preference, STAR Voting elects the finalist preferred by the majority. 

 

#4: Equity and Fighting Implicit Bias in the Vote:

Voting methods are not created equal. Studies show that the current Choose-One Plurality system only elects the correct winner in the around 75% or 80% of elections. Ranked Choice does better with around 85%-90%. When you consider that many elections are not competitive and that any system will get it right if there are only 2 candidates in the race, that's actually pretty bad odds. STAR Voting tops the charts with 98% accuracy, even in competitive fields with many candidates. 

 

 

And when elections fail they tend to favor some types of voters and candidates over others.

  • RCV's vote splitting issues mean that it maintains the same polarizing bias as our current Choose-One system, where candidates who are similar are more likely to split the vote, while those who stand apart from the rest of the pack get an advantage. Vote splitting also leads to an electability bias, where voters worried about throwing their vote away are hesitant to support candidates unless they know the candidates has the money, media coverage, and institutional backing needed to be a safe bet.  

    In politics this often results in new or grassroots candidates being told to step back or wait their turn. In a world where the vast majority of elected officials are white men over the age of 60, an electability bias is another name for a glass ceiling. 

    RCV's ballot exhaustion, and ballot spoilage issues mean that some voters will get their votes discarded, and studies are clear that this happens to some types of voters more than others. During municipal RCV elections in Minneapolis, MN, in 2009, "10.5 percent of the votes cast…were spoiled ballots or contained voter errors. And a higher incidence of spoiled ballots and voter error occurred in low-income, high-minority population areas” Another study "examined 98 RCV elections from 2006 to 2019 and found that, on average, 10.8 percent of ballots casted were considered exhausted by the final round."

    "In 2004, a similar percentage (9.7%) of ballots were invalidated in San Francisco’s municipal RCV elections. When political scientists at UC Berkeley reviewed the election results, they detected a higher rate of spoiled ballots in districts with more racial minorities, senior citizens, immigrants, and low-income residents."

    To be fair it's important to note that RCV is better than the current system on this point. Despite the facts above, RCV elections in the Bay area have shown an increase in the number of women and People of Color to win office. But the data does suggest that a preference voting method without wasted votes would go even further. 

  • STAR Voting eliminates vote splitting, has no exhausted ballots, has no wasted votes, and there are very few ways a voter could accidentally spoil a ballot. Studies from Ka-Ping Yee and Dr Warren Smith PhD on bias in voting methods show that unlike most other methods, STAR Voting doesn't bias in favor of extremist or centrist candidates.

    This is the foundation for fair, equal, and equitable elections, leveling the playing field, eliminating the need for primary elections, and thus significantly lowering the buy in and the hurdles to running for office in the first place. 



The future of voting reform?

Voting reform is the keystone. A single cause with the potential to empower us to be effective on every other issue we care about. But change is scary. If we are going to put in the work to adopt a whole new voting method we might as well get it right.


Sources:

1.) “[IRV] can cause spoilers in up to 1 in 5 elections or worse when there are more candidates according to expert analysis.” Frequency of monotonicity failure under Instant Runoff Voting: Estimates based on a spatial model of elections. By Joseph T Ornstein, University of Michigan, Dept. of Political Science and Robert Z. Norman, Dartmouth College, Dept. of Mathematics, 2013. 

2.) "The rate of ballot exhaustion was high in each election, ranging 9.6%–27.1%." Ballot (and voter) “exhaustion” under Instant Runoff Voting: An examination of four ranked-choice elections. By Craig M. Burnett, University of North Carolina, and Vladimir Kogan, Ohio State University, USA. 2015.

3.) "We find that RCV helps reduce the substantial drop in voter participation that commonly occurs between primary and runoff elections. Otherwise RCV does not appear to have a strong impact on voter turnout and ballot completion. In a case study of Minneapolis we find similar levels of socioeconomic and racial disparities in voter participation in plurality and RCV elections." Voter Participation with Ranked Choice Voting in the United States. By David C. Kimball and Joseph Anthony, Department of Political Science University of Missouri‐St. Louis St. Louis, MO. 2016.

4. "Drawing on previous research conducted by the Maine Policy Institute, McCarty examined 98 RCV elections from 2006 to 2019 and found that, on average, 10.8 percent of ballots casted were considered exhausted by the final round." Expert report reveals weaknesses of RCV. By Isabelle Christie. 2020

5. "Concerns about the fairness of IRV led at least four jurisdictions to repeal... Burlington, VT (2006–2009), Cary, NC (2007–2009), Pierce County, WA (2006–2009), Aspen, CO (2009)." and "Consistently, precincts where more African-Americans reside are more likely to collect overvoted, voided ballots. And this often occurs where more Latino, elderly, foreign-born, and less wealthy folks live." Overvoting and the Equality of Voice under Instant-Runoff Voting in San Francisco, California Journal of Politics and Policy. By Francis Neely and Jason A. McDaniel San Francisco State University. 

6.) Voter Satisfaction Efficiency (VSE) studies by Dr Jameson Quinn Phd. At the time this study was released Quinn was Vice Chair for the Center for Election Science. Quinn is now on the board of the Equal Vote Coalition http://electionscience.github.io/vse-sim/vse.html

7.) Election Accuracy. A comprehensive list and overview of studies, sources, and citations comparing voting methods including RCV and STAR. http://starvoting.us/accuracy