Primaries, Democracies, and Electability

Politics is about winning, so if you aren’t winning by the current rules, the standard move is to argue against the rules. One very bad argument (both politically and substantively) bandied about in primary season is that results in states which the party has little chance of winning in the general are somehow second class and should be discounted (perhaps to zero). This is usually a post-facto claim, but we could see it as a proposal to rearrange the primary schedule so that battleground states come first, then safe states, and only then no-hope states. One could even allocate more delegates or super delegates to the contested states.

There is even a fairly reasonable justification for such a schedule. Primaries have three basic purposes:

  1. To determine the “electability” of a candidate.
  2. To sort out issue stances and priorities within the party.
  3. To build up experience and infrastructure for the general campaign.

1 is an informational goal, while 2 is a political goal, and 3 an experiential or training goal. Primaries help us know something, agree on something, and do something.

The best rationale for a hopeless state last schedule is 1. The preferences of voters in hopeless states, one may argue, is a bad predictor of how candidates will do in winnable states. Relatedly, it might be a waste of effort to pursue 3 in such states as that infrastructure build up isn’t useful in the general.

I think this is a bad argument. The problem is twofold: The offered rationale aren’t correct and 2 raises a compelling reason against.

First, it’s very unclear that performance in a primary is a good predictor of general election performance either nationally or even in that state. (At some point, I’ll try to dig out an existing or construct a new quantitative argument. The hardest part is getting all the data, but it should be easy to determine correlations between primary performance and general performance going back quite a ways.) Conceptually, it’s easy to see how they come apart (voter cohorts are quite different, campaigns are different (and in contrast to different opponents), etc. etc.), so any argument here needs some good evidence about predictability. And, of course, we need that hopeless state voters behave systematically differently than the overall voter pool. This is more true due to bad sampling than to “winningfulness”. Finally, the logical outcome here is to make the nomination subject to polling at the time of the convention. That’s a better predictor than votes from months before.

If you are a 50 state strategy type, 3 becomes a plus not a minus. Even if you think that in the general, focusing resources is critical, we’re still a 50 state country and we need some sort of party infrastructure everywhere. We need to content the local races and issues that we can. If the general starves such states, then at least the primaries can provide some boost.

Second, our horribly undemocratic electoral college makes large swathes of the country essentially voiceless in the general. As Dilan wrote: “The primaries are the only chance for Mississippi Democrats (or New York Republicans) to have a voice.”

This is the heart of the problem. Voters in those states deserve some chance to influence the presidency. The president nominally answers to allthe people and, indeed, is the only official (along with the vice-presdient) in the federal government for whom everyone in the country has a vote on. Further undermining people’s voice here is something devoutly to be avoided.

If hopeless states were clearly and repeatedly forcing selection of bad candidates that swing or safe states would have rejected, then we do have to consider this argument again. Systematically losing helps no one in the party. But we would need strong evidence to overcome the democratic considerations against that schedule.