Our current voting method, Plurality or First-Past-The-Post, is only fair and accurate if there are two candidates in the race. We can do better! But what’s the best alternative? Here we present the pros and cons of four options in detail. They are not necessarily the best four options, but they are the leading voting methods under consideration for updating our elections.
STAR Voting: "Score - Then - Automatic - Runoff"
Voters us a 5 star ballot and score candidates from 0-5. Candidates left blank receive a 0. You can give multiple candidates the same scores.
STAR Voting takes one ballot and counts it in two different ways. The first round adds up all the scores and determines the two highest scoring candidates. The second round looks at each voter's preference order and determines the candidate who was preferred by more voters. Think of it like quality and quantity- Rather than having to choose, STAR maximizes both.
STAR Voting is a hybrid of Score and Instant Runoff Voting.
- The 5 star ballot is highly expressive, showing both level of support and preference order. The highly expressive ballot collects more data, and then fully utilizes it to maximize election accuracy.
- STAR voting is the most accurate voting method according to Voter Satisfaction Efficiency (VSE) studies by the Center For Election Science. (1.)
- “[STAR Voting] has a Voter Satisfaction Efficiency of 91% all the way up to 98%… [STAR] is undeniably a top-shelf election method, and arguably the best out of all the ones I tested.” -Jameson Quinn, Harvard Statistics.
- STAR guarantees an equally weighted vote, which is the legal definition of one person one vote. This means that the voting method does not favor any type of voter or candidate or give anyone an unfair advantage.
- STAR Voting usually elects the Condorcet winner if one exists. If VSE and Condorcet disagree on the best winner, the more expressive ballot data from the 5 star ballot showing degree of support (as well as preference) order offers compelling reasons why.
- The same Voter Satisfaction Efficiency studies showed that STAR does not incentivize strategic voting. Voters will get their best outcomes by voting their conscience.
- The best strategy with STAR is to show your preference order and level of support. As a result, STAR Voting encourages positive, issue oriented campaigning by incentivizing candidates to earn support from their competitors constituents.
- STAR allows voters to give the same score to candidates if they don’t prefer one over the other. This means that a voter can rate each candidate one by one, rather than having to figure out how they will vote on each candidate before they can start. Compared with Ranked Choice Voting this decreases voter burden.
- It’s fairly simple to count and understand results. Results show total scores for each candidate and also the percentage of voters that preferred the winner over the other finalist, much like how we vote now.
- STAR Voting is precinct summable, which is essential for high level election security. Precinct summability means that ballots so not need to be centrally tabulated and that each locality can tabulate their own ballots without losing chain of custody. This is because with STAR Voting each round is counted using basic addition and all data on every ballot is used, unlike RCV.
- STAR Voting is compatible with Risk Limiting Audits.
- STAR Voting could be counted on most voting machines with a simple software upgrade.
- The 5 star ballot can be used for single-winner, multi-winner, or proportional representation elections.
- STAR Voting is new and hasn’t been used in a government election yet.
- Ideally voters should have an informed opinion on all candidates they prefer to their least favorite.
- Some thought is required when deciding what scores to give candidates who aren’t your favorite or least favorite.
- Strategic voting is possible, but in practice it's not exploitable. In the very unlikely case that a voter would benefit from tactical minimization or maximization strategy, there is no way to know which would be best. In order to benefit from strategic voting a voter would need impossibly accurate polling data in a close 3 way tie scenario.
- STAR Voting doesn’t always pass the controversial “Later-no-harm” Criteria, but we believe this is actually a good thing. Later-No-Harm states that a voter will never hurt their favorite by showing support for others, but this is at direct odds with overall representative outcomes. If there is a good compromise candidate that would make voters happier overall it is good for a system to encourage voters to show that support.
- Critics worry that people could “bullet vote” giving only 5 stars or 0 stars. For some voters this is their legitimately honest vote, but when voters have a more nuanced opinion they naturally want to show it. The runoff adds a huge incentive to show your preferences between each candidate so that no matter who the finalists are your vote will make a difference and help prevent your worst case scenario. If all voters Bullet Voted STAR Voting devolves to Approval Voting, which is still a pretty good system.
Instant Runoff Voting (IRV) aka Ranked Choice Voting (RCV):
Voters rank candidates on the ballot in order preference: 1st, 2nd, 3rd and sometimes more choices. If a candidate has a majority of first choice votes, that candidate wins. Otherwise, the candidate with the fewest first choice votes is eliminated. If your first choice is eliminated, your vote goes to your next choice if they haven’t been eliminated yet, and the process repeats in rounds until one candidate has a majority.
Note: RCV technically refers to all voting systems that use ranking, but in the USA RCV is often used to refer to a specific single winner version, IRV, and sometimes it’s multi winner version, STV. We discuss IRV here because it’s the most popular despite the fact that it’s the least accurate version of RCV.
- Fairly simple ranking ballot.
- Much more expressive ballot then Plurality.
- Most voters should be honest about their rankings.
- RCV is the most common alternative voting method used in the world and has 100 years of data behind it.
- Instant Runoff Voting is the single winner version and Single Transferable Vote is the multi-winner Proportional Representation version. Both use the same ballot.
- IRV is vulnerable to the spoiler effect aka. vote splitting, where adding an extra candidate can cause the candidate with the most support to lose.
- RCV is the least accurate voting method covered in this article according to simulated testing by the Center For Election Science.
- RCV doesn’t take all rankings from all ballots into account and so is not the most accurate way of counting ranked ballots. If your first choice candidate is eliminated in later rounds your second, third, or fourth choices may never be counted. (Ranked Pairs, Schultz, and Bucklin Voting are much more accurate ways to count ranked ballots.)
- Some voters will have all of their rankings counted while others will only have some of their rankings taken into account. Some voters have their first choice eliminated and their next choice isn't counted.
- It’s not necessarily safe to vote for your favorite, there are cases where voting for your first choice is a bad strategy that can backfire when there are more than two viable candidates. (IRV fails Favorite Betrayal Criterion.) In this scenario you might want to rank your preferred front-runner 1st and your favorite 2nd. This is particularly true if your favorite is a strong underdog who you don't think can win.
- RCV favors voters who prefer very strong or very weak candidates but puts moderately strong candidates and their voters at a disadvantage.
- Because you can’t give tied rankings, voters have to figure out which order to put all the candidates in before they can assign ranking. This can be tricky if you aren’t sure which you prefer or if there are similar candidates.
- Hard for elections officials to process and understand the results because of the confusing process. Because some 2nd and 3rd choice votes are worth the same as a 1st choice vote, while others are worth nothing there’s no way to compare how many votes each candidate got at the end without re-running the whole election using another ranked choice algorithm.
- Votes must all be processed in one central location and can’t be tabulated by precinct. This makes it more vulnerable to election fraud and is a major logistical challenge.
- If there is a non-representative result voters may never know about it due to to complexity of results. Full data has not been published for most IRV elections that have been held. Once data is published it would be up to elections officials and data analysts to crunch numbers and determine if the election picked the winner with the most support.
- “Recent work by Robert Norman, a mathematician at Dartmouth, suggests that IRV’s …[tabulation] issues would create non-representative outcomes in one in five close contests among three candidates and that with larger numbers of candidates, it would happen even more often. The 2009 Mayoral IRV election in Burlington, Vermont was one such sideways election, and the results led to the repeal of IRV in Burlington the next year.” (1.)
- IRV has been repealed in a number of US cities where it’s been tried. Reasons for these repeals range from the method’s complex vote transfers, to high cost voting machine upgrades required to tally the IRV ballots, to unpopularity of the candidates elected.
- IRV puts viable 3rd parties at a strong disadvantage because that is the scenario most likely to trigger the spoiler effect or to encourage favorite betrayal strategic voting.
Check a box for as many candidates as you approve. The candidate with the most approval votes wins.
- Simple and doable using existing ballots and infrastructure.
- More accurate than Plurality Voting or Ranked Choice.
- Vote totals show a much more accurate representation of support for candidates who were unable to win.
- You can't show your preference order, which means that you can't show that you prefer your favorite over another candidate who you approve and who may have a better chance of winning. Because of this Approval strongly favors the candidates that are perceived as most electable, giving the media out-sized influence on who can win.
- Doesn’t let you chose your favorite over a lesser-evil candidate so it doesn’t pass the test for honest voting.
- Favors centrist candidates and strongly discourages 3rd party candidates.
- Strategic Voting required for best results.
- Doesn’t allow voters to express how they actually feel about the candidates.
- Because voters must be strategic there is no way to know how well the results matched the actual will of the people.
- Approval Voting can be used for single-winner, multi-winner, or proportional representation elections.
Voters give each candidate a score, the candidate with the highest total wins.
- Simple to understand and explain.
- Allows voters to express detailed opinions of each candidate.
- Simple to implement and use for elections officials and easy to understand the results.
- Not vulnerable to favorite betrayal strategy where voters feel they have to vote for a lesser evil candidate. It’s always best to give your favorite a max score.
- Score Voting performs well in simulations and testing and is quite accurate.
- Uses the most expressive kind of ballot which lets us accurately gauge how representative the results are.
- The score ballot can be used for single-winner, multi-winner, or proportional representation elections.
- Could hypothetically be vulnerable to strategic voting tactics. Voters from the dominant parties could “bullet vote” and get an advantage by giving their favorite a max score and everyone else a zero, even if they really do have a more nuanced opinion.
- Score Voting produces the best, most representative results if everyone shows their honest, nuanced opinion, but people can gain an individual advantage with tactical voting.
- If voters are tactical, Score Voting devolves to Approval Voting, which is still a pretty good system.
- Voters from minor parties are incentivized to do “bullet voting" where they give their favorite a max score and also give their preferred front-runner a max score as well with zeros for all others, even if they really do have a more nuanced opinion.
1.) Voter Satisfaction Efficiency (VSE) studies by Jameson Quin of the Center for Election Science.
2.) “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