Sat 20 Apr 2019 3:26PM

How does/should STAR voting work with only 1 or 2 candidates?

AW Aaron Wolf Public Seen by 15

If the ballot has full 5 stars for just one or two candidates, it seems excessively complex. But inconsistency is also complex.

I may want to not use the full score range or specifically mark the single candidate with a low score to indicate disapproval etc. but that encourages not-as-good thinking about what the 5 stars mean and how to use them with lots of candidates. We don't want a ballot where most 1- or 2-candidate positions get people in the habit of low-ranking or otherwise being "bad to good" rather than "worst to best" for what they think the stars mean. That risks them continuing the pattern when a multi-candidate seat shows up later on the ballot.

If we want to get away from treating the stars as a good-bad poll but instead as use of 1-person-1-vote, maybe all seats with only 1 or 2 candidates should just be approval voting?


BTernaryTau Sun 21 Apr 2019 1:58AM

I don't think 5 stars is that complex for just one or two candidates. Using multiple ballot types seems like it would be more complicated overall.

I'm not quite sure what your second concern is. Are you saying you're afraid voters will avoid normalizing their ratings to make use of the full 0-5 range?


Aaron Wolf Sun 21 Apr 2019 2:42AM

Yes, there's no reason to normalize with only 1 or 2 candidates


BTernaryTau Sun 21 Apr 2019 3:51AM

I doubt voters would learn that behavior, but in the event they did, the correct response would simply be to drop the automatic runoff and just use plain score voting. I think that would actually end up being a better situation than voters normalizing under STAR.


William WAUGH Sun 21 Apr 2019 1:27PM

Mention of one-candidate elections makes me think again about the benefits of including "None of the Above" (NoTA) as a candidate. Even if it were included with a determination not to enforce a NoTA win (however strong) by leaving the office vacant, and to elect the second-highest-scoring candidate instead, having the option to protest the unsuitability of the candidate field would help the electorate communicate among themselves with precision about the question.


Aaron Wolf Sun 21 Apr 2019 3:08PM

I personally like NoTA as an option.

Besides 1 or 2 candidates, there's also the ballot proposal question, which is yes/no. Score voting (not STAR) on a yes/no question allows intentional under-weighting of your vote when you don't feel strongly. I really like that in principle, but I worry that mixing STAR and Score will confuse people about the normalizing.


Sara Wolf Wed 15 May 2019 4:48PM

We did a recent star.vote on who to nominate from the progressive caucus for a seat in an upcoming plurality election. There were 2 candidates only, so beyond showing preference the scores were essentially irrelevant to the outcome, but the option to show level of support was still interesting. We had a few votes of no preference which used the middle of the scale (4-4 and 3-3) and we also had a few people, (myself included) who didn't use the full scale (I did 3-5). The candidates finished with average scores of 3.4 and 3.6, with the candidate with a higher average winning by 1 vote out of 31 voters.

It's pretty great that STAR Voting automatically normalizes your ballot in the runoff. Even if voters don't vote optimally the impact is mitigated. Moreover voters are able to express a less polarized or decisive opinion without compromising their voting power.


Jonathan Bright Thu 16 May 2019 2:50PM

It's pretty great that STAR Voting automatically normalizes your ballot in the runoff. Even if voters don't vote optimally the impact is mitigated.

I mean, that's theoretically not pretty great, since it could choose a less representative candidate than plain Score, but that argument depends on everyone voting honestly, etc.

Since you're going with STAR for its other practical advantages, I think reducing it to Approval in cases where the scores are ignored anyway would be a good thing. It's functionally identical, makes it more clear what effect your vote has on the outcome, and simplifies ballots if there are many such races.

The only advantage I could see to using STAR in those situations is that you could report the overall rating distributions of each candidate, which affects their "mandate", but that doesn't have much practical effect.


Sara Wolf Fri 17 May 2019 12:27AM

That comment was about elections in general, not just ones with 2 candidates. To be clear.

You're worried about a person with weak preferences over-riding a voter with strong preferences, and that's valid, but there's another phenomena at play that's just as or more relevant- When a voter doesn't like either of the front-runners, but their vote still deserves to count just as much as everyone else's. The fact that STAR normalizes those ballots in the runoff is why STAR does manage to out-VSE Score Voting, despite the concern you linked above.