Why do primary/caucus results create "momentum"?

[Nichlemn]

It's well-established that succeeding in a primary is likely to give a candidate a bump in the polls. What is less well-established is why this occurs. This is especially puzzling for success in terms like "performing better than expectations."

There's only one reason I can think of that coincides with "rational" voting - learning from other voters. If Candidate X wins Iowa,  this is information that suggests that many informed voters chose Candidate X, and so if you're an uninformed NH voter who wants to make an informed vote, an easy short-cut is choosing who the most Iowa caucers picked. Still, this appears unconvincing when there appear to be much better ways of getting informed (say, by following the endorsement of a public figure who you trust and tend to agree with).

Every other reason I can think of suggests dumb and/or shallow voters.

If you're rationally learning from other voters, then Candidate X winning by a tiny margin over Candidate Y should mean you're basically indifferent between them, but instead it appears that Candidate X gets a substantially larger bounce. I'm guessing a lot of this bounce is a media created self-fulfilling prophecy.

Maybe a lot of voters just want to vote for the winner. There's nothing wrong with that, but it's not a particularly good sign for democracy if significant numbers of voters are that shallow.

Why should performances relative to expectations matter? Maybe it's a take on the "learning" model, in which overperforming the polls suggests that there's been a late surge of informed voters changing their minds, so a candidate that wins 15% after polling 5% might have a greater share of the informed vote than a candidate that won 20% after polling at 30%. Or maybe it's another take on the "vote for the winner" effect, where polling higher than expectations suggests an upward trend.  But it seems a lot more likely that the polls were just wrong, and to read too much into this is silly. My guess is that most of the reason for this effect is another stupid media created self-fulfilling prophecy.

If I'm right about the self-fulfilling prophecies, that would suggest that changing media attitudes (either in the composition of media that voters consume, or opinions of those that they do consume) could significantly alter Presidential races. Is it possible that if enough media figures coordinated together, they could create (even more) stupid self-fulfilling prophecies? If every commentator said "the candidate who gets 2nd in Iowa will get a bounce into NH because the winner is too mainstream", would that start happening?

[Mr. Morden]

I'm afraid I have to go with the dumb and/or shallow voters hypothesis:

http://www.slate.com/articles/news_and_politics/politics/2004/02/the_kerry_cascade.single.html

[WillK]

The Sheeple effect.

[tpfkaw]

Primary voters know that a protracted primary campaign with the candidate's attacking each other only hurts them in the general, so there's typically a lot of pressure to unify behind the most popular choice.

[Likely Voter]

If Gingrich loses IA, this will put the cascade effect to the test. Conservative voters have made it very very clear they want someone who isn't Romney. They have moved around from one to the other. If Gingrich loses IA it wont be because they suddenly loved Romney as much as they couldn't decide which non Romney to get behind.

So if a couple of the conservative non-Romney's drop out by SC, will the conservatives then get behind one of the remaining non Romneys, or will they throw in the towell and just give in to the Romney cascade?

[Averroës Nix]

Winnowing. Party actors give up on candidates who can't translate money, organization, and campaigning into results, and shift their support accordingly.

[Jacobtm]

It's largely about performance relative to expectations.

If you're asking why that affects voters' psychology, the answer is Social Proof

The media establishes stories so quickly, that a voter who isn't tied to one candidate can be surprised to learn that, just before the NH Primary, Iowa voters have come up with a brand new ''winner'', whom the voter hadn't considered before, because the candidate was previously being dismissed or called a ''loser''.