GENERAL QUESTIONS ON STAR VOTING:
1. Please describe how your recommended method works.
With STAR Voting you give your favorite(s) five stars, your last choice(s) zero, and then score the rest as you like. Voters can give candidates equal scores to indicate no preference. Voters can score as many or as few candidates as they like. Those left blank receive a zero. STAR Voting uses a five star rating style ballot, allowing voters to show their preference order, and also level of support.
STAR stands for Score Then Automatic Runoff and that's exactly how it works.
Voters only have to vote once, and the ballots are counted in a two step process:
- Scoring Round: the two highest scoring candidates are finalists.
- Automatic Runoff: your vote automatically goes to the finalist you scored higher. The finalist preferred by the majority wins.
Both the scoring round and runoff are tallied using addition. The scoring round sums the stars given and measures the overall level of support for each candidate. The runoff counts the number of voters who preferred each finalist.
2. Is there evidence that your method is easy for voters to understand and use?
Yes, millions of people have used 5 star ratings and have filled out standardized testing and survey forms similar to the five star ballot pictured above. We are confident that STAR Voting would be easy for voters to use and are pleased to say that we have never received reports of voters confused about how to cast their ballots or vote. Additionally, Equal Vote has committed to ensuring a high-quality, accessible, voter education campaign everywhere the method is adopted.
When we look at the ease of use for voting methods, it's important to consider the cognitive load required for each ballot type. For example, consider a voter who has 5 candidates they love, 3 they dislike, and 2 they hate. This voter would have a hard time choosing one only, because they have 5 favorites, and they would have a hard time ranking these candidates because their opinion includes equal preferences which the ranked ballot can't support. In STAR Voting, these voters would be able to give 5 stars to their favorites, 1 star to the disliked candidates, and 0 to the hated candidates.
Cognitive Load demonstrates that people generally have nuanced opinions and can effectively hold a working memory of around 5-7 options in their minds, with various opinions on each, and it's easiest to allow people to express these options honestly without having to translate them into a more restrictive format.
Because the 5 star rating matches the level of granularity of a human opinion, because it doesn't have restrictive rules, and because an honest vote is a strong vote, many voters find it easiest to vote with a 5 star ballot, especially in races with more options.
3a. What does a voter need to know about how to mark a paper ballot at home, without assistance?
In order to STAR Vote all voters need to know can be summed up as 5=Best, 0=Worst. Detailed voter instructions can be found at the top of the ballot, and legislation for STAR is explicit that these instructions be printed on the ballot itself:
- Give your favorite(s) five stars.
- Give your last choice(s) zero stars.
- Show preference order and level of support.
- Equal scores indicate no preference.
- Those left blank receive zero stars.
3b. How is the ballot counted if a voter marks two candidates at the same preference level?
Equal scores are counted equally in the scoring round, and if two candidates in the runoff had been scored equally that ballot would be counted as a vote of no-preference between those two candidates.
A vote of no-preference in STAR Voting is not the same as an exhausted ballot in RCV. In RCV an exhausted ballot may have contained information that could have made a difference if the rankings had all been counted, or the voter may have ranked as many candidates as they were allowed to and been forced to leave multiple candidates unranked, even though they had a preference between them. Most jurisdictions which use RCV limit the number of rankings a voter can give to 3, 4, 5 or at the most 10.
In STAR voting, voters are always allowed to score all candidates and all scores cast are counted. Thus if the runoff comes down to two candidates who had been scored equally by a given voter, we know that that voter had explicitly voted that they like these two candidates equally. Some votes of no-preference will have marked both finalists 5 stars, some may have ranked both at an intermediary level, and some voters may have explicitly voted that they did not want to support either under any circumstances.
3c. Is there a way the voter can mark the ballot that partially, or fully, reduces the influence of that ballot? If so, how do you justify that reduced influence?
No, in STAR Voting the runoff is binary so every ballot is ultimately counted as a fully powerful and equal vote.
Some people have suggested that a voter in STAR Voting who doesn't use the full 0-5 scale, or a voter who is generally less enthusiastic about the candidates, would have a less powerful vote than someone who did give their favorites a 5. In the scoring round this is true, which is why the instructions tell people to give their favorite(s) 5 stars and to give their last choice(s) a 0, but ultimately it is up to the voter if they would like to give their full support to anyone or if they choose to cast a protest vote which shows that they disliked all the candidates. Even if a voter does not follow the instructions and does not use the full scale, in the STAR runoff, their ballot will still be counted as a fully powerful vote for whichever finalist they prefer.
Lastly, note that in STAR voting, even if a voter does absolutely fail to follow instructions, it is virtually impossible to accidentally void or spoil your ballot and the vote cast will still be counted. Equal scores, which would be counted as an overvoted voided ballot in RCV, are allowed in STAR Voting. For other types of overvotes the protocol would be to first determine if voter intent is otherwise indicated, and then count the mark for the highest score given to that candidate, which would cover the scenario if a voter starred in all the bubbles up to the rank they wanted like a 5 star rating.
The fact that spoiled ballots are a non-issue in STAR Voting is a major advantage over RCV, in which studies showed that up to 9% of ballots were overvoted and thus spoiled in one San Francisco election. This same trend was reported by the Maine Wire. "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, not affluent, predominantly white voting precincts.’
For comparison, generally about 1 percent of votes cast in plurality elections are invalid. In Minneapolis’ RCV elections in 2013, the trend held — researchers observed a higher rate of spoiled ballots in low-income wards than in more affluent ones."
4. Are there any specific kinds of elections that your method is, or is not, suitable for? In particular: Non-partisan elections, primary or general elections, one-nominee-per-party general election, Top 2 open primaries, Top 4/5 open primaries?
STAR Voting is highly adaptable and should be used for any of the election types described above. STAR Voting (or other highly accurate voting methods that eliminate vote splitting) becomes increasingly necessary in elections that are especially competitive and/or have large fields of candidates.
LEAGUE CRITERIA QUESTIONS: Strategy and Manipulation
5a. Will your method prevent or inhibit political manipulation
The current Choose-One Plurality system is the most gameable voting method out there and any alternative voting reform would present a significant improvement.
Currently, voters need to be strategic and vote for those who seem "electable" to avoid wasting their votes. Many candidates and parties will leverage the spoiler effect to coerce voters into not voting for their favorite and instead support a lesser-evil candidate in order to prevent their worst-case-scenario from winning. Worse, candidates and parties are able to game the system by intentionally running or funding spoiler candidates to split the vote.
Candidates and parties can also leverage low-turnout primaries to effectively win races before the electorate is even paying attention, and this is fairly easy to do with targeted Get Out The Vote efforts that can lead to unrepresentative winners.
“[RCV is also expected to] cause spoilers in up to 1 in 5 elections or worse when there are more candidates according to expert analysis," particularly in races with multiple viable candidates. This is because RCV only counts one ranking at a time, (just like a Choose-One Plurality election) and because not all rankings are ever counted.
STAR Voting eliminates vote-splitting by allowing voters to show that they prefer all voters in their coalition or on their side over all others, and by counting all those preferences, thus STAR Voting would eliminate issues like these that stem from vote-splitting and the spoiler effect entirely.
STAR Voting also goes the furthest to eliminate the need for a primary in the first place in many cases, and to ensure that winners are representative of as many voters as possible, even in elections with large fields of candidates.
For more information see the page on this at equal.vote/gaming_the_vote.
5b. Will your method prevent or inhibit gerrymandering?
Proportional STAR Voting and Bloc STAR Voting are both good options which would mitigate the impacts of gerrymandering by essentially reducing the number of districts and the number of boundaries between districts. Unfortunately this reduces local representation, and the pros and cons of switching to multi-member districts for a given jurisdiction should be weighed carefully.
That said, Equal Vote recommends explicitly eliminating gerrymandering using enforceable standards such as Efficiency Gap to measure the impacts of political or racial gerrymandering, and to trigger a redistricting process if the results of recent elections show that the districting is consistently giving one faction or another an unfair advantage. When used in combination with other anti-gerrymandering measures, like independent redistricting commissions and compactness , gerrymandering can be addressed just as easily or more easily than changing the entire voting method to somewhat combat its impacts.
6. In what ways does your method promote sincere voting over strategic voting? Is there evidence of voters or campaigns promoting either sincere or strategic voting with your method?
In STAR Voting the behavior that is incentivized is an honest and expressive vote, where voters (as instructed) give their favorite(s) 5 stars, give their last choice(s) 0 stars, and show preference order and level of support for their candidates.
STAR Voting was invented specifically to address valid concerns with strategic voting and gameability in both Score voting and RCV, and it has been shown to do so using statistical analysis, which confirmed the conclusions that make sense logically as well. Specifically in STAR voting, voters are incentivized to show their preferences and be expressive to ensure that in the runoff their ballots will show which finalist they prefer. This positive incentive gets increasingly strong in the situations where it's the most relevant, in competitive races where it's unclear who the two front runners actually are.
This type of honest and expressive voting incentivized by STAR Voting would have massive implications to empower currently marginalized voters. Currently if you are in the minority and know your candidate can't win, your vote literally has no impact. With STAR voting even if none of your favorites can win, your vote can help prevent your worst case scenario and make a difference. For example, a Democratic voter-of-color in a deep red state would be able to give good scores to their (non-viable) favorites, give 1 star (or as many stars as they deserve) to a family-values conservative, and help prevent a far right candidate running a campaign based on hate from winning.
Strategic tactics such as not voting for your favorite and voting lesser-evil instead are not incentivized in STAR Voting.
Bullet voting (giving 5 to your favorite and 0 to all others) is not incentivized and is likely to backfire in STAR because voters need to show a preference between the finalists to ensure their preference can be counted in the runoff.
The only "strategic voting" which could be effective or incentivized in STAR voting is to exaggerate the difference in scores between the frontrunners, while still showing your honest preference order, but it's questionable if this kind of behavior should even be lumped in with dishonest strategic voting behaviors, as a ballot like that would still translate to an honest ranking or an honest plurality vote and would still result in good ballot data and help elect representative winners.
Studies on strategic voting for various methods indicate that even if voters are strategic in STAR, or if one faction bullet votes and the other doesn't, that results would still be more accurate than the current system or RCV under ideal voter behavior.
To date there is no evidence of campaigns or candidates encouraging voters to vote strategically, though there have been a number of cases of candidates asking us how to vote strategically because they were unable to figure out a workable strategy on their own. We affirm that an honest and expressive vote is the strongest vote.
For more information see the page on this at starvoting.us/accuracy.
7. Does your method allow more people to run without changing the outcome? That is, if an additional candidate joined a race but did not win, would their presence change which of the other candidates ended up winning? (Ex: with FPTP, Nader joining the race may have caused Bush to win instead of Gore).
Yes, as described above, STAR Voting eliminates vote-splitting and the spoiler effect and does as much as possible to ensure that the system performs well even with large fields of candidates. The fact that voters can give full support to multiple candidates also ensures that the system performs as well as possible.
Unfortunately no system can solve this problem entirely because voters can only research so many candidates, but with STAR Voting and vote-by-mail we would have a system that's as robust as possible even in stress test type elections.
8. Can voters using your method safely express support for more than one candidate without harming their favorite? Do you have evidence that voters using your method actually take advantage of ranking or scoring multiple candidates?
It's important to know that it's impossible for any voting method to allow you to safely give your full support to your favorite, and also to allow you to support other candidates with no risk to your favorite. The Later No Harm and Favorite Betrayal Criteria are mutually exclusive. STAR Voting effectively maximizes both as much as possible
Practically speaking, in STAR Voting, voters should always give their favorites 5 stars, and they can and should express support for all the candidates at the level they feel is deserved. Doing so ensures that even if the voter's favorite can't win, that they will get as good an outcome as possible.
Data from STAR Voting elections is clear that on average voters are highly expressive, and that most take advantage of the opportunity to support multiple candidates. This is increasingly true in elections with more candidates, and for voters with more candidates on their side. As expected, races with smaller fields of candidates, or with factions who only run one candidate will have less expressive voting on average, which reflects the fact that many of these voters may only have one candidate they support.
9. How important is getting a high rating for voter satisfaction efficiency (VSE)? What does this measurement mean to you?
Statistical modeling is one of the most important tools at our disposal for comparing and evaluating voting method accuracy, strategic incentives, and the impacts of various voter behaviors.
"Voter Satisfaction Efficiency" is one of the most sophisticated and realistic models done to date. The VSE model was developed and run by Dr. Jameson Quinn, an internationally regarded expert in electoral science and Harvard PhD in Statistics. For these reasons we consider VSE to be an important metric to consider, but we do advocate taking a close look from multiple perspectives. Other models comparing voting methods to date have corroborated the findings of VSE and further strengthened the validity of the comparative conclusions that can be drawn from them. A few other metrics we recommend looking at are the original Social Utility Efficiency studies, Bayesian Regret, Yee Diagrams and models based on them, and Condorcet efficiency. More on how these metrics together form the foundation of the study of voting methods can be found at http://starvoting.us/accuracy.
10. How important is it that the election winner is supported by more than half the voters? Can your method fail to elect the majority winner? If so, why is this failure acceptable?
Finding winners who have as strong of support as possible and who have the support of as many people as possible is the central goal of an election. STAR Voting does both. The scoring round finds the two candidates who have the strongest and broadest support possible, (consensus winners,) and the runoff ensures that the winner is the finalist who was preferred by the majority of voters who had a preference between these two, (majoritarian winners.)
It's important to note that a simple majority only exists if there are two candidates only, and if voters can only support one or the other. In elections with multiple candidates there may be three or more polarized factions and a true majority may not exist. In voting methods where voters can support multiple candidates there are often elections where there are multiple candidates who are supported by a majority of voters. For example, a majority of my family loves Thai food. A majority also loves pizza. The point being it's important to not only look for a majority supported candidate but also the candidate who is the strongest candidate who represents as many people as possible as well as possible.
It's also worth noting that even a voting method like RCV which technically passes the Majority Criterion fails to elect majority supported winners a majority of the time, and that in RCV the definition of majority is a majority of "remaining ballots" only. Not a majority of all ballots cast.
For these reasons we believe that STAR Voting is better at ensuring majority supported and majority preferred winners than RCV.
RE: Election Types
11. Does your method pair well with a method that elects policy-making bodies -- such as legislatures, councils, commissions, and boards -- in a way that proportionally represents all the voters? How does this method work? What evidence do you have that it achieves proportional results?
Yes! The five star ballot can be used for single-winner, multi-winner, or proportional results.
How is STAR-PR counted?: Winners in Proportional STAR Voting are elected in rounds. Each round elects the candidate with the highest total score and then designates a quota worth of voters from that candidate's strongest supporters as represented. The next round tallies only the ballots from all voters who are not yet fully represented and the highest scoring candidate is elected to the next seat. This process continues until all seats are filled. Learn more about Proportional STAR voting here.
We can confirm that Proportional STAR Voting achieves proportional results because it passes the droop or hare quota proportionality criteria and ensures that a faction of voters which votes as a block will be able to elect a candidate as long as they have a quotas worth of voters in their faction.
12. What else would you like us to know?
The Equal Vote Coalition's mission is to fight for true equality in the vote itself, and we would be negligent if we didn't encourage you to take the equality of the voting method under consideration in addition to the questions above.
The U.S. Supreme Court has declared that equality of voting - one person, one vote - means that the weight and worth of the citizens' votes as nearly as is practicable must be the same. The astute reader may have noticed that the Supreme Court gave themselves an out with the "as nearly as is practicable" clause, and at the time of this ruling no voting method in use had ever delivered on this ideal, but that has since changed.
Traditional "Choose-One" and RCV (specifically IRV and STV) elections both suffer from vote-splitting and so they can only ensure that all voters have an equally weighted vote when there are two candidates in the race. These methods thus fail the Equality Criterion, and we believe this is a low bar that should be upheld, both for moral reasons and also for the political viability of the reforms going forward.
Passing the equality criterion and eliminating vote-splitting caused by the voting method itself is absolutely possible with any type of ballot, and there is no defensible reason to not do so. All that is required is that voters be allowed to support any of their candidates at whatever level they want, and to count all the ballot data.
For those who want a simple traditional looking ballot there's Approval Voting, for those who want a ranked ballot there are a number of Condorcet methods which ensure an Equal Vote with the correct details, and with a 5 star ballot there is STAR Voting and others options as well.
The equal vote is the key to eliminating vote-splitting, and is the key to a more equitable and representative democracy.
By Sara Wolk