Princeton Election Consortium

A first draft of electoral history. Since 2004

Nomination rules and November nerdery at the NYT

May 25th, 2016, 9:57am by Sam Wang


In the New York Times, an article about the Republican Party considering changes to its nomination rules process to eliminate “chaos,” which I think is a code word for “Trump.” However, my quick scan does not reveal any proposed changes that would solve their Trump problem. As I wrote last July in The New Republic, some kind of instant-runoff or ranked-preference voting might have gotten them out of their current jam. Probably that’s too fancy for them to consider.

Also, I see The Upshot is drilling into the Erikson and Wlezien data that I used for the post below about November predictions. They only use data from 1980, but their answer appears to be similar to mine – except that I added a predictive calculation based on the data. Again, that’s just my quick scan of it.

Tags: 2016 Election · President

31 Comments so far ↓

  • Amitabh Lath

    I just read the Times article again and I’m still unclear about what the problem is that they are trying to fix.

    Trump declared, shot to the top of the polls, stayed there through fall and winter, came in first or second in most primaries and caucuses, won support from a broad swath of their primary voters, including the deep South, East Coast, Midwest, Southwest.

    A plurality preferred Trump to any other candidate, and a majority tolerated him as their preferred ones dropped out. Isn’t that the guy that’s supposed to win in any fair system? So what’s the argument that the “wrong” candidate won the nomination?

    • Ed Wittens Cat

      Trump is the “wrong cadidate” for the GOP establishment who recognize full well that non-hispanic caucs made up 78% of the population in 2000, and make up a bare 69% today.
      Going forward the GOP will need to attract 40% of hispanics to shore up their vote share.

    • Lorem

      I think the idea (whether right or not) is that:

      a) Donald does not seem to stand for many of the things that the Republican party generally stands for, so it is questionable for him to represent it. (This pertains to the remarks that he won because of independents, etc.)
      b) That could be more palatable if he were electable in the general and brought support to downwind races. Despite recent hype, however, this seems dubious.

    • Amitabh Lath

      I have not seen any proof that independents won Trump the nomination. Republican party members caucused for him, and voted in closed and open contests. Yes, there is interesting intra-party friction between certain national leaders and the rank and file that selected Trump. But how is that going to be solved by changing voting rules?

      I suggested simulations because I have a strong suspicion that no matter how you had tweaked the rules of the contests, Trump would have come out on top.

    • Lorem

      I think it might depend on what you mean by “on top”. In particular, if you just switched everything to true proportional allocation of delegates, he’d probably end up with 30-40% of the delegates (depending on how long others stayed in, etc.), right?

      In such a scenario, it would be much more plausible to come into the convention making the argument that “65% of voters want someone who’s no Trump!” and a deal could possibly be reached to select someone else.

      Alternatively, it also seems like early on, the number of anyone-but-Trump voters may have been greater than the number of Trump voters, so under one of the alternate systems, he may have lost more of the early states. Beyond that point, it could be hard to model, since we would have to make assumptions about to what extent him winning early on helped him win later.

      Of course, the article seems to consider neither of these interventions, so I’m not sure what they’re on about either!

    • Ed Wittens Cat

      Lorem, there is a vast difference btwn what the GOP *claims* to stand for and the common substrate and wishes of the base. Like I said, the GOP elites realize that the GOP needs hispanic votes going forward– how many of the rank and file would agree to that? By 2044 hispanics will be the largest demographic in the US (Pew). Of course the base understands this too, at a far more visceral level.
      The emergent substrate in this election (on the GOP side) is white fright fascism. Trump is the first reality tv style candidate– a natural evolution (or devolution if u wish) of the American political landscape.
      as for b)
      given the example of Martinez, Trump seems unwilling to gracefully allow other elected repubs to distance themselves from him.

  • Periwinkle

    I like any system that at least tries to cope with more than two candidates without spoilers. Instant Runoff, Bucklin, Approval Voting, Range Voting… I have preferences, but I’ll compromise and accept any of them over First Past the Post.

    I do not live in the US, but looking at the primary processes from outside, it looks like either party could gain a massive advantage from using one of those voting methods to select candidates. The nominee would have a better chance of reflecting the preferences of the party, and have a good chance of better appealling to the electorate at large.

    Ideally, if the parties started doing this, they could skip the primary process altogether. With robust multi-candidate voting rules in place, why not just run the general election with all candidates from all parties?

    • alurin

      With robust multi-candidate voting rules in place, why not just run the general election with all candidates from all parties?

      That would be optimal. Unfortunately, while the methods of selecting party nominees are largely governed by the parties themselves (and state legislatures), moving to a multi-candidate system for choosing the President would require a Constitutional amendment. Currently, the President is not chosen by the popular vote winner (even in a first-past the post fashion), but by an Electoral College. Rules for electing the Electors are governed by state legislators and thus vary from state to state. A multi-candidate voting system would bypass this College.

    • Matt McIrvin

      In theory it might be possible to get an alternative system in through an interstate compact similar to the existing one aimed at eventually going to a national popular vote. The compact would specify that as soon as enough states pass it for their total electoral votes to top 270, all electoral votes in these states would be assigned according to the new national system rather than the old one.

      One tricky thing is that, just as with the NPV compact, actually implementing this might require too much cooperation from the states that didn’t approve the compact.

  • Amitabh Lath

    Surely this can be simulated with a simple Monte Carlo. Start with 17 candidates, distribute the voters among Trump and others using exit polls, impose whatever selection method you prefer, and let it rip.

    For voting methods that require voters to list 2nd, 3rd preferences I believe there was polling to that effect.

    For methods that require voters to divide a fixed number of points among the candidates one could use the very enthusiastic to not-at-all-enthusiastic scale that some pollsters determine.

    • MAT

      After Sam ran the article about Condorcet elections, I ran a model of every possible permutation of GOP candidates using the NYT dataset. Trump won every simulation he was a participant in.

  • Lorem

    I’d prefer approval voting, personally, over instant runoff. It does as well or better on most theoretical metrics, and is simpler from an implementation point of view. Plus, it feels less polarizing, since you are almost required to say: “Yes, I’m okay with several of these people”.

    Admittedly, it might be even harder for some people to accept than instant runoff, but one can dream.

    But yeah, I don’t really see what reshuffling early states is going to do. Is there even strong evidence that they have a major influence? (As opposed to just showing who’s ahead anyway.)

    • Bela Lubkin

      Approval voting in a primary might have weird psychological overtones. If none of the candidates you “approved” of in your state’s primary made it to the general election, do you just stay home?

      In an N-way-pick-1 primary like we have now, you can reason that you’re picking the one you like best (out of a field of others that you might still like). Approval voting requires you to implicitly _disapprove_ of some candidates.

      I would expect the big parties to be sufficiently afraid of this potential effect that they wouldn’t dare experiment with approval primaries.

    • Sam Wang

      After examining a few different kinds of voting, and receiving an unexpectedly largeg amount of correspondence, I formed the opinion that Bucklin voting (also called Grand Junction voting) took care of most problems, was transparent, was easy to implement. Rank your choices, add up support for the top N, and let N increase until somebody gets above 50%.

      By the way, a noticeable fraction of my correspondence on voting systems has been rather uncompromising. These people might benefit from an alternative voting system!

    • Nathanael

      Bucklin’s OK, but I support approval voting largely because it can be used in practically all circumstances. It would be of massive benefit in any committee or legislature meeting, for instance, avoiding the “amendment sequencing” problems.

  • BillSct

    I recently read somewhere that, once upon a time, to get the Republican nomination on the first ballot required an 80% super-majority. I haven’t been able to confirm that, but such a rule would have stopped Trump dead in his tracks.

    As for the Democrats and their Super Delegates, Samantha Bee does a good job of explaining the situation: https://www.youtube.com/watch?v=XtuWiHYmr4U

  • Randy Alfred

    GOP allows winner-take-all only after the point at which they thought a clear leader would emerge and they wanted to speed the nomination toward closure. That was based on a two-contender ((or perhaps three) model, and allows a plurality candidate to get all the delegates (in a state or congressional district) amongst a divided field of several other candidates. Simplest fix, without complicating their ballots, woulfd be to have winner-take-all kick in only if one candidate gets a majority, not just a plurality. Trump would be going into the convention with about 35% of the delegates under such a model.

    • emmy

      And that was a change to avoid “chaos”. Maybe they should just accommodate the base instead.

    • Matt McIrvin

      Well, they did avoid chaos. The convention is going to be an all-Trump event. It’s just not the candidate they wanted or expected.

    • Commentor

      Why are contested conventions considered “chaos?” The optics may be bad, but my understanding is that the conventions proceed under a set of rules and are decided by a vote of the delegates, which is hardly chaos. Its not as if they are decided by a no-holds-barred cage match.

    • Matt McIrvin

      Modern conventions have become entirely about optics. It’s treated as a highly stage-managed event aimed at talking up the national ticket and getting the party’s message across. And everyone counts on seeing the popularity bump that the presidential candidate gets from the convention.

      So a convention that looks bad in some way (not just a contested one) is thought of as a major disaster, whether that has a real, lasting effect on the race or not. 1968 and 1992, which weren’t seriously contested conventions but had bad optics, are often mentioned as debacles that brought down their tickets.

  • anonymous

    It would be great if the data journalism crowd would be more inclusive of similar, previously performed, openly available analysis (e.g. this site) in their articles. Sometimes it seems like they go to great lengths to avoid mentioning relevant analysis performed elsewhere. I guess a similar thing occurs in science, where people avoid citing papers published by rivals, but it does not seem to be as blatant there.

  • 538 Refugee

    Part of the reason the primary process is what it is now is that allowing candidates to focus narrowly helps those with limited resources early on. It seems to me instant runoff in ‘bell weather’ states is the way to go. I’m not sure states current ballot equipment would allow that though and the parties would be forced to come up with their own mechanism.

    Personally, I favor the instant run off for all sorts of elections. We live in a republic but you could overlay a veneer of democracy to the whole process that way. That would give us all a much truer representation on where the voting publics true feelings lie.

    • Jerry Schwarz

      I also favor instant runoff for almost all elections. A great resource on that is http://www.fairvote.org/

    • Michael Hahn

      I must respectfully disagree with the run-off idea. We have it in the state where I live, and the end result is usually that a minority ends up deciding the outcome. The second election rarely draws the turnout that the first election does, and so an even narrower slice of the population ends up deciding the election. Hardly democratic or representative!! The largest problem with the American electoral system is its winner-take-all aspect. This encourages gerrymandering and polarization. My own favorite electoral system right now is the one in place in Germany. It gives political voice to the major political streams in the society, and more importantly, the art of compromise is inherent to the system. But it is probably unrealistic to hope that this country might adopt such a system.

    • alurin

      @Michael Hahn: An instant runoff is different from the two-stage primary-runoff system you are criticizing. In an instant runoff system, instead of voting for a single candidate, voters express their preferences for a range of candidates, so that when their favorite candidate is eliminated, their votes go to their second-choice candidate, and so on. The computations may take a while, but the voter only has to vote once.

    • Michael Hahn

      After looking into the instant runoff idea, I still do not like it. It is still winner take all. The losers still have no voice in government and no leverage with the elected official. Also, there is no inherent incentive to compromise. Both need to be accommodated in order for government to be truly representative. In Germany, you cast two ballots, one for direct election of the representative (winner take all); this elects half the seats in parliament. The other half of the seats are distributed on the basis of the second ballot, which is cast for a party of your choice. The ultimate makeup of the parliament is based on the % vote a party wins. A party’s share of seats is then a combination of its directly elected representatives and those off of the party’s ranked list (which you know ahead of time). Small parties rarely get any representatives elected directly; their representatives gain seats through the second ballot. A party must get at least 5% of the vote to gain any seats based the second ballot. The consequence is that all major political streams are present in Parliament AND you must form coalitions to form a functional government (so you must learn to work together and compromise). Both would help our current political impasses.

    • alurin

      First, we are talking about the system for electing the US President, you can’t actually elect a President by that sort of proportional system, since there is only one President. In order to implement your idea, we would have to shift to a parliamentary system.
      Second, I disagree that proportional representation and parliamentary systems produce superior outcomes. In fact, all democratic systems produce a set of winners and losers. In the parliamentary system, they are called “the government” and “the opposition”. Someone ends up in the government, everyone else ends up in the opposition. The problem with coalition governments, is that you don’t actually know whether you are voting for the government or the opposition until after the election, at which point backroom deals between politicians determine who is in the winning coalition and who is on the outside. Also, such systems have a long history of either being held hostage to small parties (see most of the history of Israel), or producing “grand coalitions” of two major parties who cannot reform anything and generate apathy in the electorate (see Austria and some stretches of German history). Sometimes, you don’t get a government for months after the election, discrediting the democratic system (e.g., Belgium, Iraq).
      The US system has actually been pretty good until the last 20 years, when Republicans in Congress have become unwilling to engage in the traditional tradeoffs and compromises of Madisonian democracy. But in my view, that is not inherent to the Presidential system, but rather a product of the hyper-precise gerrymandering of recent years, along with a particularly insidious ideology. It’s nothing a little redistricting reform and a few giant political defeats can’t fix.

  • Commentor

    Closed primaries, proportional allocation of delegates, and super delegates would seem to me the best system to allow the party to keep control of the process. This is pretty close to what the Dem’s do.

    • Mark F.

      Yes, that is what I have been thinking. If Trump loses, I think we’ll see those changes for 2020. (Note: Only some of the Democratic primaries are closed)

    • Commentor

      Its easy to see how the GOP primary might have gone differently under this system. Trump would have won fewer early delegates and super delegates could have chosen another candiate and/or remained unbound. If Cruz and Rubio took their proportional share of delegates from early primaries, they would have stayed in the race longer, and their delegates plus unbound delegates could make it very difficult for Trump to win on the first ballot at the covention.