OK, here’s one story about the primary campaign, and I think it’s the most commonly heard one and the most commonly believed one, and it goes like this: because of their first-in-the-nation delegate selection, Iowa and New Hampshire voters have vastly disproportionate influence over which candidates can be nominated. The process is volatile, with voters in later states basing votes largely on the outcomes in earlier states, so the candidates have to manage the momentum of many state elections. The advantage is that the candidate is prepared for the general election by the drawn-out ordeal of the primary season, and also that the “retail” politics of the early, small states allows those disproportionately influential Iowa and New Hampshire voters to learn about the candidates in detail, essentially doing the homework for the rest of the nation. The disadvantage is that the largely white, rural populations of the early states do not reflect the nation as a whole, and that much of the nation is essentially disenfranchised by the winnowing process. Yes? Is that all familiar?
Here’s another story about the candidate selection process, courtesy of William G. Mayer. People largely make up their minds about their preferred candidate before the Iowa caucuses, and after all the primaries, that preferred candidate generally is the nominee. Events that take place once delegate selection has begun, including the Iowa caucuses and the New Hampshire primary, have little to no effect.
Which one do you think is true?
I mean, by empirical observation, it’s Mr. Mayer’s. His predictors are twofold: who leads the last national Gallup poll before the Iowa caucuses among self-identified Party members, and who raised the most money in the year preceding.
For the polls, the leaders were Ronald Reagan and Jimmy Carter, in 1984 it was Walter Mondale and of course the unchallenged incumbent Ronald Reagan, in 1988 it was Gary Hart and George H.W. Bush, in 1992 it was George H.W. Bush and Bill Clinton, in 1996 it was Bob Dole and Bill Clinton, in 2000 it was Al Gore and George W. Bush. His paper, by the way, was published in 2003; we’ll get to that. You can see that in 11 out of the 12 races (two of which were unchallenged incumbents and two of which were challenged incumbents), the leading candidate is the eventual nominee. The only exception is Gary Hart, who outpolled Michael Dukakis.
The list for fundraising is essentially the same, except that Mr. Dukakis out raised Gary Hart, and John Connelly out raised Ronald Reagan in 1980. So in ten out of the twelve races, the two criteria point to the same candidate, and in every case that candidate won the nomination. And both of those factors were in place before any delegates were selected.
That’s a longish explanation, so I’ll leave my new take on it for a separate note, but I’ll add to this one a bit more by talking about 1992. I assume most Gentle Readers remember the campaign, although I suspect that my memory of it is not exactly what happened. Anyway, the Iowa caucus was won by Tom Harkin with a local hero 76%, with another 12% of delagates uncommitted, and the remaining candidates grouped under 5% (with Paul Tsongas at 4%, Bill Clinton at 3%, Bob Kerry and Jerry Brown at 2%). Paul Tsongas then went on to win as a New Englander in New Hampshire, and won some other early states outside the South, but got crushed in the Southern States on what was then Super Tuesday. There were some wild shenanigans with Illinois and New York, where people briefly thought one or another candidate was gaining momentum, but in the end, it was Bill Clinton, who had started the whole process with the most support, who was the candidate. I was living in California at the time; my candidate of choice was Tom Harkin. He didn’t make it to the California ballot; by the time we voted, it was essentially all over. The question is whether it was essentially all over before the voting began.
Tolerabimus quod tolerare debemus,
-Vardibidian.
Bibliography:
“Forecasting Presidential Nominations or, My Model Worked Just Fine, Thank You”, William G Mayer. PS, Political Science & Politics. Washington: Apr 2003. Vol. 36, Iss. 2; pg. 153, 6 pgs
The Press-Politics of the Presidential Primary Process, Andrew C. Cline, 26 May 2003 (updated October 2003).
So who are the nominees this time around?
Well, the less important criterion is fund-raising in (for this cycle) 2007; I believe that the leaders there are Hillary Clinton and Rudy Giuliani. The more important criterion is the Gallup poll, and I don’t actually have the numbers for the Gallup Poll that was released on (I believe) January 3rd. I could probably get them from Lexis-Nexis. My recollection is that Hillary Clinton and Rudy Giuliani retained a lead in the national polls at that time (they had been ahead for pretty nearly the whole previous year), but that their leads over Barack Obama and Mike Huckabee (respectively) had dwindled to within the margin of error.
Note that in a case like this, the Mayer Model could still be the more accurate story, even if it does not accurately predict the nominees. Or it could have modeled what the process was like between 1980 and 2000, but the process has changed in ways that make the model no longer accurate. Or the model could still be largely accurate as a model, but the predictions fail for some particular conditions which we are now discovering. I have a suspicion which it is, but it’s only a suspicion. And besides, discussion of that question rests on a decision whether the Meyer Model was the accurate story between 1980 and 2000.
To that end, I’ll add to the bibliography Iowa, New Hampshire Results Often Shift National Preferences, by Jeffrey M. Jones, which looks at those elections and the Gallup polls before and after the onset of delegate selection.
Thanks,
-V.
I had been wondering about what exactly “momentum” was in this context; that is, why a win in Iowa &/or NH would make a difference. So thx for this entry. I find it interesting that Gallup presents essentially the same data as Mayer but doesn’t reach quite the same conclusion; given the data, Mayer’s model seems to make much more sense than the idea that sometimes an early win helps and sometimes it doesn’t.
In particular, Gallup seems to focus on the fact that an early win usually provides a bump, but fails to explicitly note that that bump is almost never enough to get the nom.
So all this brings us to Kerry vs. Dean, which it took me a while to figure out was the point of your mentioning 2003. My perspective in 2004 was that the Dems were desperate to support someone electable, so the minute Kerry started looking more electable than Dean, they dumped Dean. My impression @ the time was that it had more to do with the anyone-but-Bush idea than with any particular liking for Kerry per se.
In the 2007 national polls, Clinton has been the clear & strong frontrunner all along, despite electability concerns. I’ve been watchng pollster.com to see whether iowa wd change that, but I’m guessing it won’t.