Free-for-All 2008: Primary Games


With more and more states holding early primaries, pundits are baffled as to how the primaries will play out. Does the overload on February 5th give early states more power or less? Will any candidate have the time needed to dispatch some of his or her opponents before the 5th? And if not, will the frontloading paradoxically lead to a protracted battle between two candidates in one (or both) of the parties.

While I can no better predict the individual outcomes, I am aiming to apply some basic methods of estimation, which would help us understand what really motivates primary voters. Do their own preferences formed over the Long Campaign matter most? Or are they swayed by a candidate's recent successes and their own desire to pick a winner?

Thus, I posit a model of primary vote share determination where candidate i's vote share in state s (vi,s) is a linear function of pre-primary polling data (pi,s), results of previous states' (1,2...s-1) primaries weighted by size and date (f(vi,1,...,vi,s-1)), and delegates already locked up by the candidatate (deli).
vi,s = alpha1*pi,s + alpha2*f(vi,1,...,vi,s-1) + alpha3*deli
With the data from previous elections available, a model such as this one could be estimated to yield the coefficients (alpha1, alpha2, and alpha3); then, with polling data and election dates from this cycle, the entire primary season could be simulated. (If any readers have access to that type of data, I would be very grateful for help on this project).

The latest media chatter is almost constant from October. The big gainers are Hillary and Huckabee. The Republican side continues to be a dogfight between the Big Three, with Fred Thompson slipping and out of contention in Iowa and New Hampshire.

Rank Candidate ChatterRank Change
R.1 Rudy Giuliani 5,4600
R.2 Sen. John McCain 3,469+1
R.3 Gov. Mitt Romney 3,402-1
R.4 Fred Thompson 2,9190
R.5 Rep. Ron Paul 1,6110
R.6 Gov. Mike Huckabee 1,283+1
R.7 Newt Gingrich 98+1
R.8 Sen. Sam Brownback 285+1
R.9 Rep. Tom Tancredo 276+3
R.10 Rep. Duncan Hunter 212+1
....................................................................................................
D.1 Sen. Hillary Clinton 9,4390
D.2 Sen. Barack Obama 5,7800
D.3 Sen. John Edwards 5,5150
D.4 Sen. Christopher Dodd 1,145+5
D.5 Sen. Joseph Biden 940+2
D.6 Gov. Bill Richardson 9370
D.7 Al Gore 763-3
D.8 Rep. Dennis Kucinich 548+2
D.9 Mike Gravel 231+2

Notes: The Chatter Rankings are created by searching each candidate's name plus "2008" in the Google News database. Tested but not qualifying is Alan Keyes (28). Purged this month are almost all non-contenders; the only ones who remain are distinct vice-presidential possibilities. The purged are Condi Rice (who said she will not be VP), Chuck Hagel, Howard Dean, and John Kerry. Al Gore remains.

See recent graphs of the Chatter Rankings plus Chatter Rankings from October, September, August, July, June, May, April, March, February, January, December 2006, November, October, September, August, July, June, May, April, March, February, December 2005, August, July, June, and May.

The monthly prediction...

Nov '07: Clinton & Richardson over Giuliani & Thompson
Oct '07: Clinton & Richardson over Giuliani & Thompson
Sep '07: Giuliani & Thompson over Clinton & Richardson
Aug '07: Giuliani & Thompson over Clinton & Warner
Jul '07: Giuliani & Thompson over Clinton & Warner
Jun '07: Clinton & Warner over McCain & Romney
May '07: Clinton & Warner over McCain & Romney
Apr '07: Clinton & Warner over McCain & Giuliani
Mar '07: Clinton & Obama over McCain & Giuliani
Feb '07: Clinton & Obama over McCain & Giuliani
Jan '07: Clinton & Obama over McCain & Giuliani
Dec '06: Clinton & Obama over McCain & Giuliani
Nov '06: McCain & Giuliani over Clinton & Warner
Oct '06: McCain & Giuliani over Clinton & Warner
Sep '06: McCain & Giuliani over Clinton & Warner
Aug '06: McCain & Giuliani over Clinton & Warner
Jul '06: Clinton & Warner over Allen & Romney
Jun '06: Clinton & Warner over Allen & Romney
May '06: Clinton & Warner over Allen & Romney
Apr '06: Clinton & Warner over Allen & Romney
Mar '06: Clinton & Warner over Allen & Rice
Feb '06: Clinton & Warner over Allen & Rice
Posted by Chops at November 21, 2007 1:30 PM
Comments
Comment #238995

Seems fairly straight forward to me, what the Primaries and polls mean.

Democrats: Hillary wins Iowa, Hillary becomes the candidate. Hillary does NOT win Iowa, Obama and Hillary slug it out through Feb. 5, very likely with Hillary steadily losing steam as corporatist status quo candidate and Obama as the Change candidate. Polls show change is what the public is looking for.

Republicans: Not so clear cut. The Republican Party itself is a house divided: Bush supporters vs. the fiscal conservatives, Bush supporters vs. religious right, Bush supporters vs. moderate Republicans. Huckabee and McCain are viewed as Bush supporters. A large constituency. Romney is the moderate, flip flopper, a smaller constituency of solid support. And Guiliani is the Bush supporter on foreign affairs, and a fiscal conservative by his own words, and the UnReligious Right candidate. Interesting though, that Guiliani will carry some Religious Right vote, on his being viewed as the best candidate to defeat Hillary.

Therefore, Guiliani has the largest base of Republican support when push comes to primary ballot box. Therefore, Guiliani must win the majority of the primaries before Feb. 5, if he is to win the nomination. If he doesn’t, it means his campaign organization is weak compared to Romney’s, and Romney gains ground. It then becomes a two way battle throughout.

See, not so complicated if you are me. But, then, I could be entirely and completely wrong! In which case it is far more complicated than I thought, and Chops is right, this is all new math to an old pundit! :-)

Posted by: David R. Remer at November 22, 2007 12:40 AM
Comment #239025

David,

Remember that the goal of the math here isn’t (just) to predict the result, it’s to figure out how the process works. Can we express the idea of “momentum” as a quantifiable result?

I agree with your analysis, largely, but it’s really hard to know whether voters will weigh previous contests more heavily than their own expressed opinions of just a month before.

I know one thing: Super Tuesday is going to have much closer odds than the Super Bowl two days earlier.

Posted by: Chops at November 22, 2007 11:51 AM
Comment #239027

Chops, there is no process to figure out as I see it. We are not, and were never meant to know the outcome of the elections PRIOR to the elections. This is the primary argument against polling by media on and before the election for public consumption. If the voters know the outcome before election day tallies have been counted, or even suspect to know, voters have little incentive, if any, to go vote.

If the math is to achieve a process goal, it should be to keep the outcome of elections a guessing game, with all outcomes possible, by my line of reasoning.

Posted by: David R. Remer at November 22, 2007 12:04 PM
Comment #239028

Chops, on the matter of front-loading and moving dates forward, the only beneficiaries in the long run appear to be the candidate’s and party PR firms, analysts, and advertisers, with one possible exception.

IF, and its a big if, moving the primary calendar forward results in a better informed and more attentive public prior to elections, then the process meets a primary Constitutional objective. The sociological research has not yet been done, however, that I am aware of, that would indicate whether this moving of the calendar forward will result in more tuning out, or tuning in, better informed voters, or less informed voters.

And let me clarify, that ‘better informed’ does not mean better impressioned. American political strategy is laser focused by PR and marketing firms to give voters Impressions of candidates, and to avoid informing voters of concrete policy positions, strategies, and details of implementation.

It is unclear to me whether a longer political campaign season will enhance voter information. I suspect however, that the premium for candidates and parties would be to avoid ‘informing’ voters for the longer campaign season, resulting in an ever greater divide and animosity between candidates delivering impressions, and voters growing requirement for concrete information.

I relish Iowa being a front runner in the Primaries for the obvious reason that Iowans demand face to face contact and questions and answers with candidates, in ways that larger states like California can’t reasonably expect or demand by virtue of their vastly larger populations.

My fear would be that campaigning evolve to the point that the internet becomes the campaign venue allowing candidates to PREpare their presentations rather than have to respond to voters face to face and in a contemporaneous manner, where at least the potential for getting some concrete information to voters questions exists.

Posted by: David R. Remer at November 22, 2007 12:23 PM
Post a comment