Momentum and legitimacy

by Henry Farrell on April 23, 2008

Brian Knight and Nathan Shiff have a “paper”: on momentum and voter choice.

This paper provides an investigation of the role of momentum and social learning in sequential voting systems. In the econometric model, voters are uncertain over candidate quality, and voters in late states attempt to infer the information held by those in early states from voting returns. Candidates experience momentum effects when their performance in early states exceeds expectations. The empirical application focuses on the responses of daily polling data to the release of voting returns in the 2004 presidential primary. We find that Kerry benefited from surprising wins in early states and took votes away from Dean, who held a strong lead prior to the beginning of the primary season. The voting weights implied by the estimated model demonstrate that early voters have up to 20 times the influence of late voters in the selection of candidates, demonstrating a significant departure from the ideal of “one person, one vote.” We then address several alternative, non-learning explanations for our results. Finally, we run simulations under different electoral structures and find that a simultaneous election would have been more competitive due to the absence of herding and that alternative sequential structures would have yielded different outcomes.

I’ve not even a scintilla of the technical expertise that would be required to assess the claims of the paper. And they could certainly have chosen a better election year to make it in (later votes in the primary process clearly count for quite a bit more than usual this time around). But the basic underlying argument – that peoples’ primary votes in Iowa will usually count for some multiple of the influence that people’s votes in, say, Pennsylvania count for, seems to me to almost certainly be true. So is this something that people should be concerned with on basic grounds of equity etc? Does this provide enough grounds that people should push for reform (either through having all primaries on one day, or perhaps semi-randomizing the allocation of slots in the calendar if that isn’t feasible)?

Obviously, there are similar inequities in the apportionment of US Senate seats by population – but that is built into the system by design, and can’t be gotten rid of without constitutional change. Calendaring is in the remit of the parties and the states themselves. My memory is that a couple of states benefitting from the current set-up have sought to make their threats more credible through amendments to their domestic constitutions, but I am skeptical that these commitments would in fact be credible if every one else converged on a single date or changed system. This is, indeed, one of those cases where we would be better off if the simplest one-shot game theory prediction came true (i.e. the outcome in which every state party converges on the equilibrium of the earliest possible date). So would this be a bad idea?



Sam 04.23.08 at 9:03 pm

Is it, in fact, necessarily true that this time around later votes in the primary season count for more than usual? The results of the primaries following the Super Tuesday results haven’t really had a meaningful impact on the delegate breakdown for either candidates. After Super Tuesday, the subsequent primaries merely confirmed the outcome of those earlier primaries. It looks to me like it’s been a long time since Clinton actually had a meaningful opportunity to win the nomination.


Ben Alpers 04.23.08 at 10:15 pm

I happen to think that a single day, national primary would be the best way for a party to decide its presidential nomination. With a few caveats….

1) The rules would need to be uniform across the country.

2) The method of winning would need to be something other than plurality voting. Otherwise, you’ll constantly end up with nominees winning on the basis of thirty-some-odd percent of the vote. Alternatives would include IRV, range voting or Condourcet.

3) This would probably be impossible to achieve so long as primaries are treated as semi-official state functions, thus requiring coordination among state legislatures and governors and even across party lines. If parties arranged for their own national primary elections, coordinating the kind of vote that I have in mind would be substantially easier.

4) The primary election would directly select the presidential candidate. Delegates would be eliminated from the system of presidential candidate selection. Needless to say, superdelegates would play no part in the process.

(Can we all agree at this point that the superdelegates, who were instituted in 1982 to create more decisive conclusions for Democratic primary battles, have proven to have almost precisely the opposite effect this year?)


Jacob Rus 04.23.08 at 10:17 pm

The other question is what effect campaign strategies have on voter preferences, and, further, what the primary is really meant to test.

Clearly moving to a single date would significantly diminish candidates’ ability to campaign effectively in every state, which would tilt the scale towards candidates who could raise large amounts of money very early in the race, and candidates with name recognition, many connections, and/or pre-existing activist networks. It would make every voter have more equal influence, but might it equally reduce the influence of voters as a bloc? Likewise, are momentum effects really so bad?

I think looking at primary campaigns as ideally being one-person-one-vote tools to select the “voter-preferred” candidate, based on preferences at a single point in time, is overly simplistic. The primary process, it seems to me, is designed just as much to measure campaign building and fundraising, to decide on a party message and platform, to vet candidates, etc., so that the eventual nominee both adequately represents the party, and has the best chance of winning in the fall.

It is unclear to me that these suggested changes would really have such positive outcomes. Giving Iowa the first shot every time has clear disadvantages as well, of course.


mikesdak 04.23.08 at 11:11 pm

I think the way to do this would be to have three groups of primaries. The first group would allocate 25% of the delegates. This should be big enough to shake out at least some of the field. The next group would be a month later and would allocate 30%, which should whittle the field to the real contenders. The last group another month later would allocate the remaining 45%. This should be sufficient to make them important enough to eliminate the ridiculous rush to be early. The whole process would be over in two months (not counting however far in advance candidates start campaigning). I realize that this would require a great deal of cat herding and/or strong-arming by the national party to implement, but a guy can dream.


D Jagannathan 04.23.08 at 11:15 pm

@jacob rus

Rotating regional primaries seem to be a reasonable compromise. The theory this year was to include some variety in the early voting states (+NV +SC) to reinforce the idea that staggered primaries are there to allow candidates to interact with lots of different kinds of voters in different states. But the damn thing still takes too long in any event.

So split the country into, say, the 5 postal regions or the 10 federal regions established in the 1974 Office of Management and Budget Circular A-105, set up a queue, then swing through them every 20 or 40 years. It’ll save on travel expenses, too, letting lesser known candidates stay in neighboring states, and who knows, it might kill the bloody ethanol subsidies someday.


Matt Weiner 04.24.08 at 12:21 am

The results of the primaries following the Super Tuesday results haven’t really had a meaningful impact on the delegate breakdown for either candidates.

I was under the impression that this wasn’t true — Obama built up his delegate lead by winning every primary/caucus for the next month by large margins. The March 4th and later results haven’t been able to shift that, but if the contests between Super Tuesday and March 4th had gone differently we’d be looking at a very different race.


salient downs 04.24.08 at 12:36 am

I’d prefer a primary system in which the smallest states by population vote earliest, and the largest states vote last. This would allow smaller states to influence the “momentum” factor and thereby be significant, and would allow larger states to retain their significance as sources of many delegates. In that way, a vote in any state carries weight.

The way I see it, the media will treat any primary as a horse race; might as well set up a system that maximizes the positive impact of that approach (by maximizing the extent to which voters feel empowered or feel as though they’re part of “making the narrative”).


Glen Tomkins 04.24.08 at 2:22 am

“Momentum” seems to have lost its momentum

“Momentum” sure seems to have been the dominant dynamic for the last several presidential nomination cycles before this one. Without having read the article in question, I can’t say that the authors might not have made the mistake of inferring from that observed pattern some mechanism for this momentum effect that they, wrongly, convinced themselves was an inevitable and necessary feature of reality. This momentum certainly didn’t always dominate the nomination process, as contested conventions used to be the norm, not the exception. I’ve certainly never convinced myself that I understood the dynamic driving this phenomenon. And now the phenomenon of momentum seems to have failed, as inexplicably as it started to dominate 60 years ago. None of this argues for momentum as a necessary and inevitable feature of reality. I would hope that the quoted figures of early votes having approx 20X the power of late votes was intended to be an empiric “rule”, a description, really, of what had been observed to happen in the last few cycles, right up to this one.

Nothing fails like failure. The failure of momentum this year will almost certainly further weaken it next cycle. Do you think that Edwards would have withdrawn when he did had he not been counting on momentum to give the nomination, quickly, to either Obama or Clinton? If he had known then what he knows now, he would surely have stayed in the race, and his backers would have bankrolled him in that effort, because an Edwards who hadn’t withdrawn would now be either the kingmaker, or even the nominee, had a convention split three ways — in which not even the superdelegates could give it ot any of the three — had such a convention fallen into a deadlocked inability to nominate either of the two front-runners and thus turned to the third place candidate as a compromise nominee. Well, next cycle, candidates in Edward’s position will have this knowledge of what happened this year, and will have a much higher threshhold to abandon even a candidacy that has no hope of winning a majority of the delegates on the first ballot.

I wouldn’t be worried about reforms designed to end momentum. Nature seems to have done that for us already. If you think momentum is a good thing, and want to revive it, then maybe you should be looking for reforms that might do that.


nick s 04.24.08 at 5:46 am

Like others, I’ll say that the victories for Obama in February have made the actual voting element of the late primaries essentially irrelevant. The state votes now simply extend the primary season further, and what matters for the Clinton campaign is that this provides opportunities for a gaffe, a skeleton, a ‘dead girl / live boy’ moment on the other side.

So I’d agree with those upthread: Super Tuesday simply confirmed that Clinton wouldn’t be able to win conclusively.

I’d be inclined to have a single-day national presidential primary built around the popular vote and some form of preference system. Do it over a few rounds, with each round a month apart, and a rising threshold. That way, outsider candidates aren’t necessarily wiped out early.

Why? Because the primary process isn’t shackled by the constitutional biases of the general election.

There’d be complaints that this biases the process towards TV ads, big states, the best-funded candidates at the start. Well, running for president requires raising a lot of money. Furthermore, that’s a problem with the campaign finance system, not the electoral system. Admittedly, that structure would have favoured Clinton over Obama, but I’d be prepared to accept that outcome in exchange for a bit of straightforward popular democracy.


Ben Alpers 04.24.08 at 5:53 am

Clearly moving to a single date would significantly diminish candidates’ ability to campaign effectively in every state, which would tilt the scale towards candidates who could raise large amounts of money very early in the race, and candidates with name recognition, many connections, and/or pre-existing activist networks.

These candidates already always (always already ;-) ?) win under our current system. Your description of candidates favored by a national primary perfectly fits Obama and Clinton (and John McCain), as well as John Kerry in 2004, Al Gore in 2000, Bill Clinton in 1992, Mike Dukakis (at least relative to his competition) in 1988, and Fritz Mondale in 1984. The current system does nothing to counteract this tendency (though it seems to convince a lot of people that it somehow doesn’t favor these candidates). The power of money in politics is quite independent of the primary calendar.

Our primary system is like the BCS in college football. It’s existence can adequately be explained historically and in reference to certain special interests that it serves. But advocates for it have to manufacture ex post facto justifications for it that inevitably fail to hold water.

Just as nobody, and I mean nobody, suggests deciding the national college basketball championship using a version of the BCS, nobody (and again I mean literally nobody) suggests that rolling elections are a good electoral system in any other context.

Why not elect senators from California using a series of rolling county competitions? Why not stagger different states’ general elections to choose our president?

These ideas are not even considered because they do not bear consideration. They are ludicrous, as are the major parties’ presidential nomination systems.

You want to democratically choose a party’s presidential nominee? The one, and only, way to go is a single-day national primary with uniform rules.


lemuel pitkin 04.24.08 at 3:49 pm

Ben Alpers has it exactly right.


Matt Weiner 04.26.08 at 3:16 pm

Lemuel, that doesn’t seem accurate about McCain this year or Kerry in 2004; didn’t both of them run out of money early in the campaign season and have to take moderately desperate* measures to keep their campaigns afloat? And though McCain had national name recognition, Kerry didn’t really — Gephardt had more name recognition, and probably more connections and activist networks due to his labor support.

I’m also not sure it applies to Clinton in 1992, who had no name recognition whatsoever, and who I don’t remember having any particular fundraising prowess or connections; though maybe the DLC served that role. (The 1992 field was pretty obscure in its entirety though IIRC, so maybe he fit these criteria better than any other candidate.)

And I think a rolling county system for senatorial primaries in California would be a great idea. Maybe then someone could knock off DiFi.

*And in McCain’s case, moderately illegal, in combination with his later attempt to withdraw from public financing.


Matt Weiner 04.26.08 at 3:17 pm

–I mean, “Ben, that doesn’t seem accurate” — I was crossposting from a thread where Lemuel quoted Ben’s comment.


lemuel pitkin 04.26.08 at 4:45 pm

I think a rolling county system for senatorial primaries in California would be a great idea. Maybe then someone could knock off DiFi.

If you believe elections should be democratic except when undemocratic elections might produce an outcome you like better, well then, you just don’t believe in democracy.


Matt Weiner 04.26.08 at 6:26 pm

Is the reason that Dianne Feinstein has a mortal lock on the Democratic nomination that there isn’t anyone that California Democrats might prefer, or is it that California’s huge scale makes it impossible for anyone to even think about running against an incumbent without raising something like tens of millions of dollars? — well, it’s probably advantages of incumbency mostly, but it’s not clear that the single-day primary really is doing the best job of letting voters express their preferences in massive media markets.

Comments on this entry are closed.