Majority Choice Approval

From electowiki

Majority Choice Approval (MCA) is a class of rated voting systems which attempt to find majority support for some candidate. It is closely related to Bucklin Voting, which refers to ranked systems using similar rules. In fact, some people consider MCA a subclass of Bucklin, calling it ER-Bucklin (for Equal-Ratings Bucklin).

How does it work?

Voters rate candidates into a fixed number of rating categories, e.g. "Good," "Neutral," and "Bad."

If one and only one candidate is given the highest rating by an absolute majority of voters, that candidate wins. If not, the second-highest rating is added to each candidate's vote total; again, if there is only one candidate with a majority they win. This process continues until some candidate has a majority.

Unfortunately, if voters cluster in certain categories (e.g. if there are only a handful of ratings, or if ratings are clustered at multiples of 5 or 10), this procedure is likely to end up with multiple candidates reaching a majority at the same rating. Therefore, a tiebreaking procedure is needed. Some possible resolution methods include:

  • MCA-A: Most approved candidate (most votes above lowest possible rating). This is also called "Majority Top//Approval", or MTA.
  • MCA-P: Most preferred candidate (most votes at highest possible rating).
  • MCA-M: Candidate with the highest score at the rating level where an absolute majority first appears. This system is equivalent to traditional Bucklin voting.
  • MCA-S: Range or Score winner. The candidate with the highest average (mean) score is declared winner, where candidates are given 0 points for the lowest rating (not rank), 1 point for the second-lowest, etc.
  • MCA-R: Runoff. Two finalists are chosen by one of the methods above or an equality-allowed Condorcet method over the given ballots. The finalists are then measured against each other using one of the following methods:
    • MCA-IR: Ballots are counted for whichever one of the finalists they rate higher.
    • MCA-AR: Actual runoff: Voters return to the polls to choose one of the finalists. This has the advantage that one candidate is guaranteed to receive the absolute majority of the valid votes in the last round of voting of the system as a whole.

A note on terminology

"Majority Choice Approval" was first used to refer to a specific form, which would be 3-level MCA-AR in the nomenclature above (specifically, 3-MCA-AR-M). Later, a voting system naming poll chose this term as a more-accessible replacement for "ER-Bucklin" in general. This is also sometimes known as "median ratings", because it can be defined as "the highest median rating wins".

Criteria compliances

All MCA variants satisfy the Plurality criterion, the majority criterion for solid coalitions, monotonicity (for MCA-AR, assuming first- and second- round votes are consistent), and Minimal Defense (which implies satisfaction of the Strong Defensive Strategy criterion).

All of the methods are summable for counting at the precinct level. Only MCA-IR actually requires a matrix (or, possibly two counting rounds), and is thus "summable for k=2"; the others require only O(N) tallies, and are thus "summable for k=1".

The participation criterion and its stronger cousin the consistency criterion, as well as the later-no-harm criterion are not satisfied by any MCA variant, although MCA-P only fails participation if the additional vote causes an approval majority.

Other criteria are satisfied by MCA variants with appropriate tiebreakers, including:

  • The Condorcet criterion is satisfied by MCA-IR if the pairwise champion (aka CW) is visible on the ballots and would beat at least one other candidate by an absolute majority. It is satisfied by MCA-AR if at least half the voters at least approve the PC in the first round of voting. These methods also satisfy the Strategy-Free criterion if an SFC-compliant method such as Schulze is used to pick at least one of the finalists. All other MCA versions, however, fail the Condorcet and strategy-free criteria.

Thus, the MCA method which satisfies the most criteria is MCA-AR, using Schulze over the ballots to select one finalist and MCA-P to select the other. Also notable are MCA-M and MCA-P, which, as rated methods (and thus ones which fail Arrow's ranking-based universality criterion), are able to seem to "violate Arrow's Theorem" by simultaneously satisfying monotonicity and independence of irrelevant alternatives (as well as of course sovereignty and non-dictatorship).

An example

Tennessee's four cities are spread throughout the state
Tennessee's four cities are spread throughout the state

Imagine that Tennessee is having an election on the location of its capital. The population of Tennessee is concentrated around its four major cities, which are spread throughout the state. For this example, suppose that the entire electorate lives in these four cities, and that everyone wants to live as near the capital as possible.

The candidates for the capital are:

  • Memphis, the state's largest city, with 42% of the voters, but located far from the other cities
  • Nashville, with 26% of the voters, near the center of Tennessee
  • Knoxville, with 17% of the voters
  • Chattanooga, with 15% of the voters

The preferences of the voters would be divided like this:

42% of voters
(close to Memphis)
26% of voters
(close to Nashville)
15% of voters
(close to Chattanooga)
17% of voters
(close to Knoxville)
  1. Memphis
  2. Nashville
  3. Chattanooga
  4. Knoxville
  1. Nashville
  2. Chattanooga
  3. Knoxville
  4. Memphis
  1. Chattanooga
  2. Knoxville
  3. Nashville
  4. Memphis
  1. Knoxville
  2. Chattanooga
  3. Nashville
  4. Memphis

Assume half of voters in each city rate one city preferred, two cities approved, and one city unapproved; and half rate one preferred, one approved, and two unapproved.

City Preferred Approved
Memphis 42 42
Nashville 26 84
Chatanooga 15 79
Knoxville 17 45

There is no preferred majority winner. Therefore approved votes are added. This moves Nashville and Chattanooga above 50%, so a winner can be determined. All the given resolution methods would pick Nashville, which is also the pairwise champion in this example.

Various strategy attempts are possible in this scenario, but all would likely fail. If the eastern and western halves of the state both strategically refused to approve each other, in an attempt by the eastern half to pick Chattanooga, Nashville would still win. If Memphis, Nashville, and Chattanooga all bullet-voted in the hopes of winning, most Knoxville voters would probably extend approval as far as Nashville to prevent a win by Memphis, and/or at least a few Memphis voters (>8% overall, out of 42%) would approve Nashville to stop Chattanooga from winning. Either one of these secondary groups would be enough to ensure a Nashville win in any of the MCA variants. It would take a conjunction of four separate conditions for Chattanooga to plausibly win: it could happen only if Knoxville voters also preferred Chattanooga, and Nashville voters (un-strategically) approved Chattanooga, and no Memphis voters preferred Nashville, and the MCA-P variant were used.

General strategy and notes

If the electorate knows which two candidates are frontrunners, and the pairwise champion is indeed among those two, the stable strategy is for everyone to approve or prefer exactly one of those two, and fill out the rest of the ballot honestly and as expressively as possible. (In the example above, that would mean an east/west split, with Nashville winning 68-32 approval over Chattanooga.) If everyone follows this strategy, the pairwise champion will win with the only absolute majority. And if even half of voters follow this strategy, multiple majorities are highly unlikely.

However, this two-frontrunner strategy does not mean that MCA is subject to Duverger's law. A pairwise champion who is not one of the perceived frontrunners still has a good chance of winning, especially if they have some strong supporters. This fact, in turn, will affect what "frontrunner" means; an extremist candidate with a strong but sharply-limited base of support - the kind of candidate who, using simple runoff voting, makes it into a runoff with a strong showing of 35% but then gets creamed with only 37% in the runoff - will never be perceived as a frontrunner in the first place.

Thus, overall, many elections should be resolved without need for a resolution method, and so all MCA methods should give broadly similar results. However, if resolution is needed, a lack of majorities is, overall, more likely than multiple majorities. Since the design intent is to minimize these situations, the resolution method chosen should be one which tends to encourage extending approvals; that is, one which comes "close" to fulfilling the Later-no-harm criterion, so that extending approval is unlikely to harm one's favorite candidate. From simple to complex, such methods are: MCA-S, MCA-IR, and MCA-AR.

Notes

MCA ballots can be modified to do Smith//Approval with the use of an approval threshold. Limiting the number of allowed rankings in Smith//Approval makes it closer in design to Approval voting than to most Condorcet methods.