Voting methods comparison
The three important questions
Do you favor majoritarianism or consensus, and how strongly?
Assume an election has two very polarizing candidates who are loved/hated by opposing halves of the organization's members and a few compromise candidates who are qualified but who people don't feel strongly about. Should the winner be one of the polarizing candidates or one of the compromise candidates? Electing one of the polarizing candidates would give the organization an officer who is viewed as the worst candidate by almost half of the membership. This might make it hard for them to do their job and leave a lot of people feeling bitter. Electing one of the compromise candidates would give the organization an officer who wasn't many people's first choice and would be ignoring the more popular polarizing candidates.
How much do you trust your members not to vote strategically?
Some voting methods, including Condorcet and Borda Count, can produce crazy outcomes when there is a high amount of strategic, rather than honest, voting. If many members are likely to be dishonest on their ballots with the goal of helping their first choice win, then Borda Count and Condorcet should be avoided. To understand this problem, see this scenario.
How willing are you to tolerate many rounds of voting?
Different voting methods lend themselves to longer or shorter elections. With some methods, like Condorcet, only one ballot is ever needed. With other methods, like exhaustive ballot, many rounds of voting can be needed.
Comparison table
Understanding this table
Polarize/compromise
Assume an election has two very polarizing candidates who are loved/hated by opposing halves of the organization's members and a few compromise candidates who are qualified but who people don't feel strongly about. Will the winner be one of the polarizing candidates or one of the compromise candidates?
Condorcet loser can win
This column answers the question: can the candidate who would lose to every single other candidate in a head-to-head matchup somehow win the election?
Favorite betrayal
This column answers the question: Can voting for your true favorite candidate as your first choice help some other candidate who you dislike to win? If so, members are incentivized to vote dishonestly about their first choices; they are incentivized to take candidates' perceived viability into account when choosing their first choice instead of voting their true preferences. This effect can also be summarized as "Every vote that's not for X is a vote for Y."
Later harm
This column answers the question: Can your second and further choices hurt your first choice's chance of winning? If so, members are incentivized to vote dishonestly about their second and further choices in an effort to help their first choice candidates.
Reaction to strategic voting
In the presence of high amounts of strategic (dishonest) voting, some voting methods still produce a reasonable outcome while others go haywire and choose a winner who almost nobody wanted. In this column, "good" means that the voting method chooses a reasonable winner in the face of high amounts of strategic voting, whereas "poor" means the voting method can go berserk.
| Voting method | Polarize/compromise | Condorcet loser can win | Favorite Betrayal | Later Harm | Reaction to strategic voting |
|---|---|---|---|---|---|
| Plurality | Polarizing candidate wins | Yes | Yes, greatly | N/A | Good |
| Exhaustive ballot | Polarizing candidate wins | No | Yes, somewhat | No | Good |
| Instant runoff | Polarizing candidate wins | No | Yes, somewhat | No | Good |
| Borda count | Compromise candidate wins | Only with dishonest voters | Yes, greatly | Yes | Poor |
| Condorcet | Compromise candidate wins | Only with dishonest voters | No† | Yes | Poor |
| Approval | Depends | Yes | No | Yes | Good |
| Score | Depends | Yes | No | Yes | Good |
† Technically yes, but not in a significant manner.
The columns "Favorite betrayal" and "Later harm" together measure how much a voting method incentivizes strategic, rather than honest, voting. Whether strategic voting actually occurs also depends on the culture of the organization. The "Reaction to strategic voting" column indicates how well a voting method copes with strategic voting when it exists.
Although this table portrays the "Polarize/compromise" column as binary, it is actually a spectrum and voting methods can lean more or less to one side or the other.