October 01, 2003

135 candidates, Duverger's Law, and Rationality of McClintock's and Bustamante's candidacies

Because of an early decision of the Secretary of State, it has been very easy for recall challengers to qualify for the ballot. All one needs is 65 signatures and $3,500. The Secretary's decision was based on a dubious interpretation of the Elections Code, but no matter---the California Supreme Court refused to get involved in the question, and eventually we ended up with 135 candidates on the ballot.

Has this large number of candidates mattered? As a matter of substance, I think the answer is no. Political scientists have long noted that first-past-the post voting rules tend to lead to the emergence of two parties. This idea is named "Duverger's Law," after a French political scientist Maurice Duverger, though the idea predates Duverger. A similar dynamic is occurring here. At least on Part 2 of the ballot, the dynamic has been for people to coalesce around just a few candidates with a chance of winning. This explains why some candidates, most recently Arianna Huffington, have effectively dropped out of the race. (I say "effectively" because these candidates' names have remained on the ballot.)

The dynamic is a bit thrown off here by the presence of Part 1 of the ballot, asking whether Davis should be recalled. If we had strong parties intent on maximizing the chances of holding or capturing the governor's office, we might predict two things that should have happened (or should happen soon) but did not happen. (1) McClintock should withdraw, so that Republicans have a stronger chance of winning in Part 2 and (2) Bustamante should withdraw, so that Democrats have a stronger chance of winning Part 1.

What explains McClintock's and Bustamante's continued presence in the race? It might be that McClintock is simply ideologically committed to staying in. Or it might be something else. If Schwarzenegger in the polls looks like he is pulling away from Bustamante, more Democratic voters and independent voters who prefer Davis to Schwarzenegger might change their vote on Part 1. So keeping McClintock in the race might be a rational Republican strategy.

As for Bustamante, the latest Los Angeles Times poll shows an 8 point advantage for Schwarzenegger over Bustamante. Maybe internal models of turnout (we don't really know who is going to turn out to vote) show that lead considerably closer. Maybe Democrats are going to use payments to get the vote out in reliably Democratic areas as they have in the past. Or maybe Bustamante simply does not want to do Davis any favors, and the Democratic party in California is not strong enough to force him to do so.

Having 135 candidates on the ballot has not seemed to affect the ability to hold reasonably informative debates either. Where the 135 candidates matter is on the ballot itself, especially in conjunction with California's rules on randomization and rotation. Candidate names are listed in random (not alphabetical) order, and randomized across assembly districts. So a candidate cannot even campaign with "Vote # 88." The randomization is done because of a perceived ballot order advantage of being first on the ballot. I'm critical of some of the social science on this effect, and working with others on looking at this question. It will make balloting slower on election day, which may raise other problems as well.

Posted by Rick Hasen at October 1, 2003 10:11 AM