Balkinization

How Many Candidates Should Be Asked to Debate?

Ian Ayres

Is there anyway to setup in advance some neutral principals for deciding how many candidates to invite to a presidential debate? Last night 7 Republicans debated. But instead of thinking about the specifics of whether Ron Paul should be allowed to debated, I'm looking for rules that might not be articulated now to decide which candidates would be invited 4 or more years hence.

It's a hard question to answer because the debates serve multiple goals. They are both a platform for us to get know relatively unknown candidates and a platform for front runners to articulate their differences.

One approach would be to let the voters (or at least poll responders) decide the question.

A poll could ask likely voters how many or even which candidates they would like to see invited. But its not clear that democratic decision making over invitations would serve democracy's best interest. Front runners would instruct their supporters to favor relatively small debates (or possibly massively large ones where noone had time to carve out a distinguishable immage). On the other hand, letting self-elected elites make the tough call is only as good as elites.

Another possibility is to analyze the results of the most recent traditional ("If the election were held today . . .") polls and invite the "effective number of candidates" with the most support to participate.

What the heck is "the effective number of candidates"? It comes from something called the Herfindahl Index. The Herfindahl Index is a truly non-intuitive measure of industry concentration (does the sum of the squared market shares mean anything to you?). But it is turns out the reciprocal of the Herfindahl index has a natural interpretation as the effective number of firms in the industry. If there are 4 firms in the industry with equal market shares, the reciprocal of the Herfindahl will be 4. As concentration increases in the market shares, the effective number of firms declines.

The same "effective number" calculation can be applied to the candidates shares in the polls. In the most recent USA Today/Gallop Poll, the Republican candidates garnered the following shares:

Mike Huckabee 25
Rudy Giuliani 20
John McCain 19
Fred Thompson 12
Mitt Romney 9
Ron Paul 4
Duncan Hunter 1
Alan Keyes 0
Someone else 2
None/Unsure 9

If we recalculate shares without the "someone else/none/unsure," and undergo the unpleasant task of taking the reciprocal of the sum of the square of these shares produces the helpful result that there are currently 4.98 effective candidates currently running for the Republican nomination. Or put differently, we might say that current support for Republican candidates is divided among 5 effective candidates. So this approach might suggest that inviting the top 5 candidates. If voter supporter over time became more concentrated, the number invited to future debates would decrease. (A similar analysis suggests unsurprisingly that there are 3.1 effective candidates for the Democratic nomination.)

Posted 8:37 AM by Ian Ayres [link]

Comments:

We already have a procedure Americans seem to find legitimate: remove the least popular contestant from the show after each round.

# posted by

ADamiani : 10:40 AM

Wasn't Huckabee polling close to nothing several months ago? Using polling data to decide who debates would have probably excluded him . . . even though he's now among the front-runners.

# posted by

dsc25 : 11:54 AM

What about issuing a number of invitations to the debate equal to the effective number of candidates, but giving out those invitations by lottery? Each candidate receives a number of entries in the lottery proportional to her poll support. That way, the debates are still properly sized, relative to the concentration of the race, but the proposal doesn't have quite the same exclusionary effect? Granted, exclusionary effect is the point of the proposal, but still--some opportunity for rising insurgents to shake up the debates would be a good thing. And since the debates would be more limited in size, candidates who win the lottery against the odds would be more able to make coherent points rather than being drowned out by all the other long shots.

# posted by

James Grimmelmann : 12:19 PM

Given the chicken/egg problem of market share and exposure, I'm not sure that folks would view a market share metric as legitimate.

# posted by

Brendon : 2:09 PM

If the debates are privately sponsored, then -- just a crazy thought -- the private sponsor should determine the invitation list.

Sorry if that formula is too complicated compared to your poll-based "effective candidate" system.

On the other hand, the candidates could, in classic collective bargaining style, unite a priori to set their own joint requirements about invitation lists. Just because Fox News invites you, doesn't mean you have to show up -- especially if you've pledged in advance only to attend fully inclusive debates (and with the media, and therefore the public, watching who commits to what and who breaks which promises).

At the post-nomination level, meanwhile, the test is simple: Any candidate with a mathematical possibility of an Electoral College victory (i.e., on the ballot in enough states) should be allowed to participate. Stated differently, the debates should be expanded from "bipartisan" to "nonpartisan."

Call that the Strong Form; the Weak Form is: "any candidate on the ballot in all 50 states should be allowed to participate."

# posted by

KipEsquire : 6:08 PM

It is indeed difficult: How do you craft a pseudo-scientific rationale for excluding from debates candidates you don't want the public hearing from, without risk of it excluding your own candidate on the next go-round?

The debates are fairly expensive to put on, especially given the cost of the air time. Perhaps they should sell the seats to the candidates: Calculate the cost of airing one candidate's worth of replies and rebuttals, add a bit of overhead, and invite every candidate who'll pony up. If you can't afford your share of the cost of the affair, you probably haven't got the funding to run an effective campaign. If you can afford it, they've got the funding to extend the debate time enough to accommodate you being part of it.

I must agree with Kipesquire: At the primary level, the debates are private affairs, the organizers should not be legally obligated to include or exclude anyone. 1st amendment, you know, which some of us think applies even to political speech.

At the general election level, where our political duopoly has created a de facto government run debate system, (Out of outrage over the LOWV daring to invite a third party candidate, IIRC.) he(?) has identified the logical course: Include every candidate on the ballot in enough states to theoretically win in the electoral college. I can not recall a race, ever, where this number was infeasibly large. The mode, I believe, has been 3, perhaps at most 4.

# posted by

Brett : 10:11 AM