# Voting system to elect multiple winners from a large field of candidates

I have a large number of candidates (65) for an award which will be granted to a handful of winners (6). There will be a relatively small number of voters (9). We are looking for a voting method which can fairly select 6 winners from this large field? We are open to multiple rounds of voting. How many votes should be allowed in the first round to begin to whittle the pool but also not feel too restrictive? Of potential relevance; this is for a Non-profit organization.

• Do the candidates also get to vote (for themselves)? Dec 8, 2018 at 17:40
• and how many voters? Dec 8, 2018 at 17:54
• You should understand Arrow's Impossibility Theorem. You can also search on this SE for other questions about it. See also politics.stackexchange.com/questions/34638/… Dec 8, 2018 at 18:13
• Can the 9 of you meet (in person or online) and form a consensus, or could you explain why this is not likely to be possible. Dec 8, 2018 at 19:19
• There seems to be only a minor connection between this question and politics. Is there maybe another StackExchange more suitable? Dec 10, 2018 at 9:32

You could use some variant of score voting. In score voting systems, each voter gives a personal score to each option. The scores of all voters for each option are summed up (some variants discard the best and worst vote for each option in order to reduce the power of outliers). The options with the highest total scores are elected. This system works quite well if the voters are honest and rate every option objectively. However, if a voter wants to maximize their influence, then they would always vote either 0 or maximum score. And if everyone follows that strategy, it turns into approval voting.

Another option would be to use a Ranked voting system where the voters list all the options sorted from best to worst. Ranked systems like Instant Runoff or Single Transferable Vote do not work so well when you have more options than voters, but what usually works quite well in this situation is Borda Count. In this system you assign a point value to each rank. Then you sum up the points of every elector for each option. Or in other words: Score voting, but you must award every possible score exactly once.

If all these systems sound too complicated and you want a system which is simple and easy to understand, then you can use Cumulative Voting. In cumulative voting, each voter has multiple votes. They can distribute their votes to different options in any way they like. The options with the most votes win. Although when you have a situation like this with many options and few voters, you might want to put a limit on how many votes a single voter can give to one option. Otherwise a single voter can almost guarantee the election of an option nobody else wants by pooling all their votes on it. Or in other words: Score voting, but the total amount of points you award must not exceed X.

However, no matter which counting system you choose, make sure that all voters are aware of how you are going to count before you start voting. Otherwise you might get accused of retroactively picking the counting method which makes your preference win.

I'd say to use Approval voting in the first round to whittle down the candidates to a smaller set of most-approved (maybe 12, if there are 6 winners?), then use Score voting in the final round after evaluating each finalist carefully.

This is an award, so you want to choose the 6 candidates who are the best overall, right? Cardinal/rated systems are perfect for this. (You wouldn't want a proportional representation system, for instance.)

You also wouldn't want a ranked-choice system, where you have to rank all 65 candidates relative to each other. This would be very difficult and time-consuming for each voter to do, and many of those relative rankings would be meaningless. Psychology studies have shown that ratings produce better results than rankings anyway:

Many value researchers have assumed that rankings of values are more valid than ratings of values because rankings force participants to differentiate more incisively between similarly regarded values ... Results indicated that ratings tended to evidence greater validity than rankings within moderate and low-differentiating participants. In addition, the validity of ratings was greater than rankings overall.

Rating every candidate on a 0-10 scale could also be somewhat time-consuming, though, which is why I suggest using Approval in the first round. You just Approve or Disapprove of each candidate, and the most-approved candidates move on to a more detailed runoff.

You could also use Combined Approval in the first round for a little more accuracy but not too much more complication (everyone votes Approve, Disapprove, or Neutral on each candidate).

The more complicated rated systems like STAR, Majority Judgment, etc are meant to discourage strategic exaggeration of votes. If the voters are not voting for themselves, and aren't heavily invested in who wins, this kind of strategy shouldn't be a problem, so Score voting should work fine.

How many votes should be allowed in the first round to begin to whittle the pool but also not feel too restrictive?

Anything that limits the number of votes one can give out will lead to spoiler effects, vote-splitting, etc. The best voting systems allow you to evaluate each candidate independently of each other, so these problems don't exist.