PR-USA: We still need it

Thanks to a shout-out at Twitter by Michael Latner, I went back and re-read a few very old posts (from 2005 and 2006) that I did in the category, PR-USA.

Although all were written with respect to politics of the moment, here on Election Day, 2018, the urgency of significant electoral reform remains. For instance, take the forecast for the House. Using their “classic” forecast, we see that “Democrats are favored to win a majority of seats if they win the popular vote by at least 5.6 points”.

That’s right. Democrats could win the popular vote by more than FIVE percentage points and we could still have a Republican House seat majority. That would be a scandal of representation. No electoral system should be considered justified on democratic (or republican–note small initial letters) grounds if it is within the realm of realistic probability that a reversal of the voting plurality could occur even with a five-point edge for one party. (Their forecast gives Republicans about a 14% of retaining their seat majority; if they do so, it will almost certainly be without a plurality of the vote.)

It hardly matters whether the root of such an outcome would be gerrymandering (partisan-biased district-boundary drawing) or simply the geographic distribution of votes (i.e. Democrats running up huge margins in their safest seats while Republicans eke out many more close wins). Both causes are inherent to use of the single-seat plurality (or sometimes majority) electoral system.

Of course, it is easier, in principle, to fix the gerrymandering cause. And there are several such measures, along with other electoral-reform measures, on ballots around the country today. As I said in a post in 2005 opposing (with some reluctance) a measure in my state that was billed as terminating gerrymandering, these do not solve the fundamental problem, even though they would help.

In addition to almost totally ensuring that the party with the most votes also has the most seats, proportional representation would limit polarization, open up alternative dimensions of issue competition, and institutionalize a voice for the sort of anti-establishment sentiment that now only bursts forward in spasms of “radical middle” or “populist” voting.

Henry Droop made many of these points a century ago. I made variants of them a dozen or so years ago. And they remain relevant today. Literally today.

20 thoughts on “PR-USA: We still need it

  1. Single Transferable Vote or MMP? I think it would be wise for a US state to experiment with a PR system before it could be tried at the Federal level. I would like to see STV system for the House, and the AV for the Senate; a la Tasmania.

    • “Single Transferable Vote or MMP?”


      As in, either one is fine with me. As is pretty much any type of PR system.

      My guess is that STV would be a much easier sell in the US, but we have a long way to go before we get to the discussion of types of PR (outside of the narrow circles already doing that, such as on this blog and among activists).

      AV for the Senate makes a lot of sense, given that we are stuck with single-winner contests there, anyway. Unless, of course, we undertook major constitutional restructuring, which is not likely. And if you have AV for the Senate, it certainly makes sense not to have something too different for the House. Seems like a good case for STV, then!

      • I think a US state could experiment with an MMP system for the Assembly and the Senate uses an STV system, but mixing a bullet ranked, and number ranked electoral systems would cause a lot of confusion unless the MMP system was a 1 vote MMP system using a preferential vote for the SMD and the First Preference was a party vote using the best loser system after the distribution of preferences, but I wonder under such a system could parties game the system to win more seats, but decoy lists are unlikely as it is 1 vote system.

        Other alternatives may be just using an MMP system using FPTP for SMD, and a Party List component.

      • I was so hoping the governorship of Maine would go to preferences, but it looks like the Democrat has an absolute majority.

      • Ranked choice voting isn’t being used for the gubernatorial election in Maine, since the state constitution requires a plurality vote for that office: it was used for the primaries and for federal offices only.

      • Rob,

        Seeing how U.S. parties work, they’d find a way to use decoy lists or worse pretty quickly. And people would believe that as they vote locally for an “Independent Republicrat” while giving their list vote to the party.

  2. STV would be the best electoral system for the US as it is Candidate based, not party based. I don’t think we would ever be like Mexico and draw districts across state lines to ensure regional and/or national proportionality when using an MMP system that would require a constitutional amendment, it is best to avoid those as much as possible. Could a DMP system be used for the Senate assuming the Constitution was amended abolishing the staggered election nature of it? That should be avoided as well, still an interesting blue sky thought.

  3. I always try to avoid debating the merits of “STV” as though it were one system. In Canada, Winnipeg elected 10 MLAs by STV from 1920 to 1945, quite nicely proportional. (It took anywhere from 15 to 37 counts to elect them, but with no TV to watch, they just kept counting.) Tasmania got better results with STV-7 than when they shrunk it to STV-5. Northern Ireland with STV-6 was better than the Republic of Ireland, where most Irishmen will tell you 3-seaters are s**t, and even the Dáil itself concluded that districts should generally have at least 4 TDs. But in Europe, Turkey’s 10% threshold is generally considered undemocratic, and of course I agree. So 13 states have enough Representatives to get results with STV-10 (or MMP for that matter), but the others are a bit small. Still, 24 states, electing 85% of the House, have at least 6 Representatives. However, has the USA become so wedded to a 2-party system that it just cannot grasp what PR is really for? So it would seem; American electoral reformers seem to want districts no larger than 5.

    • There are criteria for an electoral system besides being able to deliver precisely proportional results, and I think describing results as ‘better’ or ‘worse’ on this criteria alone isn’t fair. For example, would ten-member districts impose too high a cost on voters who wished to familiarise themselves with all the candidates, a particular problem with STV? Would it fuel excessive fragmentation in the party system? Would it require confusing ballots (remember, many US states have optical-scan ballots requiring an extra row and column for each ranking)? Ireland’s relatively small DM doesn’t appear to have stopped a multi-party system developing, and I’m unconvinced that the proportionality gain associated with ten-member districts outweighs the negatives.

      • When the vote for the STV-21 NSW legislative council, I confess there may have been elections when I was not completely sure how to separate my twentieth and twenty-first preference.

      • Bavaria has seven districts which, other than the district including Munich, average 20 representatives per district (10 local, 10 regional). With open lists, a large ballot, but a simple one: mark two Xs. Excessive fragmentation? Seven parties, lots of choice for voters, so that even conservative Bavaria no longer has a one-party government. Hard to get that result with STV-5? A strike against STV-5. When Northern Ireland designed their Assembly in 1998 they had watched STV at work next door since 1920, and had even used it themselves for their first two elections in the 1920s. They knew 5-seaters would not be enough to reflect their political diversity, so they chose 6-seaters. In 2016 Northern Ireland finally elected two Greens and two hard-left; taking lessons from Tasmania they shrank to 5-seaters, stopping the Green growth and costing the hard-left one seat. Another strike against STV-5.

      • You’re both right that dealing with a high district magnitude is pretty easy under list PR, or a system that effectively acts like list PR (the NSW LegCo), but I was under the impression we were talking about a method of STV where voters actually had some impact on the candidates elected and in which candidates would attempt personal appeals to the electorate. “Excessive fragmentation” is, of course, another less positive way of saying for “lots of choice for voters”, but I think there’s a trade-off between the two: Carey and Hix (2011) points out that relatively low DM can achieve reasonable proportionality without actually encouraging fragmentation or requiring over-large coalitions. Northern Ireland’s legislature is an odd example, because it’s not designed to deliver majority governments, and six-member districts under STV are counter-majoritarian (four quotas=57% of the vote).

    • Yes. In other words, district magnitude matters, in addition to (actually more than) formula.

      As for a threshold as high as 10%, although I would never advocate it, I would not call it undemocratic if it were not, in Turkey, combined with an otherwise simple districted system. That is, parties there can and do win the most votes in a district and get no seats, for failure to pass the nationwide threshold. (Some later got around it with SNTV-like behavior, owing to a provision allowing for independent candidates. This system has been discussed extensively on this blog.)

      A purely districted system that has a de-facto threshold of 10% or higher, due to its district magnitude, will often see a party with under 10% of the votes nationwide win seats due to sufficient concentration in one or a few districts.

      As for the US, I’d be extremely happy with STV and M=3 (for states with at least that many seats), even though I’d prefer magnitudes around 5-7.

      For various reasons, I think MMP is basically out of the question in the US. I’d like to be wrong about that.

      • You say “A purely districted system that has a de-facto threshold of 10% or higher, due to its district magnitude, will often see a party with under 10% of the votes nationwide win seats due to sufficient concentration in one or a few districts.” However, Canadian reformers may not like this, since a major grounds for criticizing “winner-take-all” here is that it gives a bonus to parties with geographic strongholds, at the cost of parties with broad national support. Everyone mentions the Bloc Québécois winning 54 seats in 1993 with 13.5% of the vote, while the Progressive Conservatives won only 2 seats with 16.0% of the vote. But it is a common problem, most recently seen in New Brunswick where the Liberals with 37.8% spread across the province got fewer seats than the PCs whose 31.9% was concentrated in their strongholds. So strongholds are the problem, not the solution.

        Metro Detroit has seven congressional districts. The rest of Michigan has another seven. Canadian STV fans are mostly found in BC, but the BC Citizens Assembly called for districts up to seven MLAs, and the current proposal for Rural-Urban calls for urban regions to use STV, with district magnitude between two and seven MLAs, although since this is for urban areas “most STV districts should tend towards the higher range.” So full-blown STV fans are not afraid of seven-seaters.

  4. The NY Times is getting in on the game. I almost have to wonder if someone from around here write these. In essence, the editorials suggest expanding the House according to the Cube Root Rule and then suggest STV (calling it Ranked Choice Voting) in 3-5 seat districts.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.