It doesn’t matter. But in what way?

There are a lot of things in politics that just don’t matter much. That is, they don’t have any independent effect on the political outcome — which candidate wins, what policy passes, etc. This is often true even about things that seem like they should matter a lot, especially in the world of campaigns and elections. Primary debates don’t seem to matter much. Who you choose for your running mate doesn’t seem to matter much. Hell, there’s an entire literature in political science that says campaigns themselves don’t really matter all that much.

One thing that I think gets overlooked, however, is the way in which something doesn’t matter. I know that sounds silly, but hear me out. Some things literally don’t matter; they have zero independent effect on the outcome. On the other hand, some things don’t matter because they don’t generate a comparative advantage in any direction. The former category is filled with things that both theoretically and empirically seem meaningless or utterly unrelated to outcomes. For instance, best I can tell, candidate penmanship has zero effect on election outcomes (God help me if there’s some political science literature out there that says otherwise). The latter category  is filled with things that theoretically and empirically might actually matter a whole lot. Take, for instance, money in presidential general elections. In a vacuum, it could be important. But because both sides can raise a ton of money, there’s a declining marginal utility of money in an election, and no one is really that much better at spending money than anyone else, the end result is that no one can get a comparative advantage. And thus it doesn’t really matter.

I see two important dimensions in the world of things that don’t matter because of comparative advantage problems: differentiation and measurability. Let’s say you are running a campaign and have to make a choice. The choice matters less when there is little difference between the options, and also when you can measure that difference precisely, such that there’s little variance in your estimation of the difference. Let me use a sports analogy to get at this point: football kickers in the NFL.  From the point of view of an individual team, kickers don’t really matter much. No one drafts kickers with their top picks, and no one trades good players for a kicker. The kickers aren’t even paid that much. But there’s no doubt that the kicking game in football matters — it directly affects the outcome of lots and lots of games.

The problem is that no one can really get a comparative advantage with kickers, for two reasons: there’s no variation in ability, and the ability is easy to measure. The top 35 kickers in the world are pretty much equally as good as one another, and the skill they possess is so easily quantifiable that it’s almost impossible to make a bad judgment in acquisition. Yes, the top few are a little better than the pack and the bottom few a little worse, but overall there’s just not much difference, and we can measure and know that precisely. Now, the situation changes dramatically when you switch to high school football (or even college); there is enormous variation in kicker ability there, and thus a comparative advantage can be gained.

Now, is any of this actually important in politics? I’m not sure, but my hunch is that it might be. When we say that something doesn’t matter much, often what we are implicitly saying is that it doesn’t matter much so long as the candidate/campaign is sufficiently skilled in decision-making, because they will make decisions that, while perhaps sub-optimal, are at least good enough to effectively neutralize the possibility of a comparative disadvantage. It’s like when we say a House Member is unbeatable; he’s not actually unbeatable, he’s only unbeatable if he continues to do the things he’s doing with his votes, constituent service, and the like (Mayhew 1974).

And this directly points to the idea of mistakes/miscalculations as well as lack of skill. Two examples: the choice of Palin as running mate by McCain in ’08, and the way the Cain campaign has handled the abortion flap in the past few days. In the abstract, these are things that don’t matter: vice-presidential nominee selection and minor gaffes in soft media interviews. But they did matter, because the campaigns handled both things so poorly that a comparative advantage was derived by any opposing candidate. In the case of McCain, it strikes me as a miscalculation borne of poor measurement; they simply didn’t research Palin enough to realize she wasn’t at the minimum threshold which makes a VP nominee not matter (see political science studies here, here, and here for analysis of the unusually-large-for-a-VP effect Palin had.) In the case of Cain, it just looks like an unskilled choice; there are a lot of ways to handle such a gaffe and yet they somehow managed to choose a way that did not allow a quick escape.

These are rare situations at the national campaign level. So two things emerge: first, we should expect a lot more things to matter in elections as the election becomes more localized and amateur. Second is that things that don’t matter tend not to matter because campaigns tend to be good at avoiding comparative disadvantages in situations where no comparative advantage exists. Which, in turn, I think suggests that campaigns are rightfully risk-averse; managing the downside of things is often the only way to get even a potential advantage — simply minimize your mistakes and hope your opponent makes more. Even among things that don’t matter.

References

David Mayhew, The Electoral Connection, Yale University Press, 1974.

Share

8 thoughts on “It doesn’t matter. But in what way?

  1. Clay

    Your two examples (McCain with Palin, Cain with abortion) are problematic. If they mattered then you are saying that McCain and Cain get a lower share of the vote than they would have if they had not made the mistake. Where are you getting the data to back that statement? Doesn’t political science refute many claims that seem intuitive? How is intuition relevant here?

    Reply
    1. Matt Post author

      Fair point.

      But I think the data does show that McCain lost votes due to Palin — I don’t have it handy, but I can say for sure that it affected my voting decision. I was considering McCain prior to the selection. But if you don’t like my examples, how about Admiral Stockdale? Did he not affect the vote share for Perot? How about a theoretical VP selection worse that Stockdale: a 35 year old high-school dropout with a criminal record? Obviously, that would never happen. But that’s kind of my point: these things don’t matter because they are easy mistakes to avoid.

      Reply
      1. Clay

        I guess I would respond by going further down the rabbit hole. Even if polling data showed that McCain lost support after the Palin nod, it could only support correlation and not causation. Sorry to belabor it, but I’m slow to be convinced.

        Reply
        1. Matt Post author

          Below are two political science papers that address the issue (I’ve also updated the blog post). They both find a large negative effect of the Palin pick. One actually finds a 2% overall decline in McCain voting due to it. Absolutely massive for a VP! I stand by my claim that Palin hurt McCain.

          As for Cain, there’s no paper yet, so yeah that’s intuition.

          As for proof of causation, in general I’m a skeptic too. But there’s a limit to the usefulness of skepticism. You can be the kind of person who rejects all non-experimental studies for lack of causal proof and also be the kind of person who rejects all experiments due to issues of external validity. But that’s a hard way to go about learning anything. Sometimes when all the imperfect evidence says the sky is blue, you just have to accept that it is.

          Matt

          The papers on Palin:

          http://convention2.allacademic.com/one/mpsa/mpsa04/index.php?cmd=Download+Document&key=unpublished_manuscript&file_index=8&pop_up=true&no_click_key=true&attachment_style=attachment&PHPSESSID=f8ead4ee03772d66b360feb037a1731d

          http://www.stanford.edu/~elis/Elis%20et%20al%202010%20Dynamics%20of%20Candidate%20Evaluations.pdf

          Also see John Sides comments on the first paper here:

          http://themonkeycage.org/blog/2009/04/05/a_tantalizing_graph_featuring/

          Reply
  2. Pingback: What to Read Before You Unwonk for Thanksgiving | Bear Market Investments

  3. Pingback: On Warm Buckets of Piss | Matt Glassman

Leave a Reply to Clay Cancel reply

Your email address will not be published. Required fields are marked *