Semi-Proportional Electoral Methods
Proportional representation (List PR, STV PR) enables blocs of voters to elect their fair share of representatives. 20% of the voters elect 20% of the seats, 60% of the voters elect 60% of the seats, and so on. By contrast, in a majority-take-all system, a simple majority of voters has the power to elect all the seats.
Semi-proportional voting methods fall somewhere in between. They are methods used to fill multiple seats in which a voting bloc can achieve some representation through voting strength (share of the electorate), but where full proportionality, where each voting bloc is represented in proportion to its size, is not guaranteed. The lack of full proportionality is typically due to the fact that, under semi-proportional methods, a voting bloc must accurately predict their voting strength and then strategically coordinate their votes in order to achieve full proportionality.
Semi-proportional methods (such as the limited vote, single non-transferable vote (SNTV), and cumulative voting) are sometimes proposed to encourage minority representation, but share the serious flaw that they achieve proportionality only when groups of voters can predict their share of the vote and cooperate to implement voting strategies. To the extent that any of these systems are simpler to implement than true PR, that advantage is more than offset by the need for complicated voting strategies.
Overview of Semi-Proportional Methods
Consider an ordinary five-seat plurality election, in which each voter is instructed to “vote for up to five candidates”, with the top five vote-getters elected. This is a majority-take-all system, in which a simple majority of voters can elect all five seats, excluding the rest of the electorate from any representation at all. This is especially obvious in explicitly partisan elections, where a party with a small majority can control all the seats on a council.
Limited Voting and SNTV
Limited voting works like a “normal” plurality election, except that each voter has fewer than five votes. SNTV (single non-transferable vote) is limited voting with exactly one vote per voter. SNTV has the most potential for proportionality, and is the version we’ll examine further.
Under SNTV, it’s easy to see that if 20% of the voters in a five-seat election all vote for the same candidate, then that candidate must be elected (it’s not possible for any other five candidates to each get more than 20% of the votes cast). Similarly, a voting bloc consisting of 40% of the voters have the power to elect two of the five seats, and so on.
The problem with SNTV is that a voting bloc must strategically coordinate their votes in order to be sure of electing a proportional share of the candidates. If the 40% bloc of voters all vote for the same candidate, then they will elect only one of the two candidates that would be their proportional share. Alternatively, if they split their votes among too many candidates, they risk electing no candidate at all. And even if they divide their votes between two candidates, but don’t do it equally, they risk electing only one instead of two candidates.
Getting this right requires that the voting bloc know in advance what their voting strength is (what percentage of the electorate they make up) as well as coordinating the allocation of their votes across the right number of candidates.
(SNTV does have one practical advantage, shared with limited voting: it requires no change to voting and counting procedures, except for the limitation that a voter is allowed to vote for fewer candidates than usual—only one, in the case of SNTV.)
Cumulative voting gives each voter as many votes as there are candidates, just like “normal” plurality voting, but the voters can divide their five votes among fewer than five candidates, if they so choose. A voter can cast one vote each for five candidates, or five votes for one candidate, or any combination in between.
The strategy problem with cumulative voting is essentially the same as for SNTV. A bloc of voters constituting 20% of the electorate (in our five-seat example) must cast all five of their votes for the same candidate in order to be assured of success, and so on.
In order to be sure of electing a proportional share of seats under one of the semi-proportional systems, a bloc must know what that share is. If a bloc of voters believe that they have 40% of the vote, and divide their votes between two candidates, but they actually have only 20% of the votes, they’re likely to elect nobody at all. Conversely, if they actually have 60% of the votes, they’ll elect only two instead of their rightful three candidates.
If the opposition is fragmented, it may be possible to win three seats (for example) with significantly less than 60% of the vote, further complicating the voting-strength considerations.
These are generally difficult, if not impossible, decisions to make in advance of an election, especially a local election with no polling.
Once the strategic decisions are made, the strategy must be implemented on election day. But this too is difficult. How do we identify our voters? How do we ensure that each one will vote in the “right” way to maximize our success? Even in a small city-council election we might be talking about coordinating tens of thousands of voters, and in larger cities or counties hundreds of thousands of voters. How do we identify them? Will they consent to being coordinated, or will they vote their own mind?
Who Are “We”?
The strategy discussion above implicitly assumes that some “we” are doing the strategizing and coordination. But it’s not obvious who “we” might be. In a partisan election, “we” might be the central party committee, but that’s not an answer for non-partisan elections (which make up all local elections in many states, such as California). And even for partisan elections, what about independent and potential crossover voters? How do “we” identify and coordinate them?
Another problem with “we” is that the need for strategy and coordination necessarily takes power away from the electorate as a whole and concentrates it in whatever centralized body is concocting and coordinating the strategy. “We” effectively act as kingmakers. (It’s possible that this accounts, in part, for the popularity of semi-proportional methods among some activist groups, though, in fairness, perhaps they simply need to learn more about PR.)
When there is no coordinated strategy, semi-proportional voters are faced with a strategy decision in the voting booth. Consider SNTV as an example. If I’m convinced that my favorite candidate is a sure winner, then it’s to my advantage to vote for my second-favorite candidate, and hope to elect both. Or, if I suspect that my favorite candidate is too weak to win a seat, perhaps I should vote for a stronger candidate, and not “waste” my vote.
These individual strategies, like centrally coordinated strategies, can backfire. If too may voters ignore the favorite, the favorite might not get enough votes for election. Alternatively, a candidate might be mistakenly perceived as not having enough support for election, and be defeated only because of that mistaken perception, and not because of insufficient real support.
STV PR Solves the Problems
STV PR addresses the strategy problems that semi-proportional methods raise. STV (single transferable vote) is described in more detail here. From the voter’s viewpoint, it’s easy: rank as many candidates as you like in order of preference, keeping in mind that your preferred candidates cannot be harmed by ranking additional candidates below them.
Using STV PR, suppose that a bloc of voters rank four candidates, Alex, Billie, Carla and Don, at the top of their ballots (not necessarily all in that order). Now, if the bloc of voters turn out to be 20% of the voters, STV PR guarantees to elect one of the four candidates (depending on which one the bloc voters prefer overall); if they’re 40% of the voters, then two will be elected; if 60%, then three. All without strategy and coordination—simply because enough voters preferred that group of candidates.
The need for a voting-booth strategy disappears as well. If a candidate has more first-choice votes than required for election, then the surplus votes are passed on to lower-choice candidates. If a candidate is defeated, those votes contribute to subsequent choices. There is, for all practical purposes, no good reason to rank candidates in a different order than my actual preferences.
This is the basic reason that we should choose fully proportional methods like STV PR to semi-proportional methods like limited voting, SNTV, and cumulative voting: PR achieves proportional results without complicated strategizing—simply by asking voters to express their preferences.
We note in closing that proportional results can also be achieved with List PR, but that List PR is seldom if ever used in non-partisan elections. List PR does, of course, require a certain degree of centralized coordination in order to draw up lists of candidates—coordination that is not required under STV PR.