Saturday, June 3, 2017

Should Clinton "take the blame" for losing in 2016?, Part II

Picking up from yesterday's post, I left off with the observation that if elections turn on campaign platforms, then they are strategic contests between candidates, and if they turn on "valence" characteristics, like competence and honesty, then they are not strategic contests.

Warning:  I'm going to do some math.  It will be simple math, though, and I'll explain everything.

In normative terms, are we comfortable with a diminished role for valence characteristics?  Economic theory allows us to construct what we call "utility functions," in which we plug in a bunch of values for the stuff we like, and the equation spits out a number to tell us how happy a person is, in numerical terms.  Observe:

(1)  U(policy) = - | policy - i |

Simple equation.  The parameter, i, represents an actor's "ideal point" for policy, and the equation takes the absolute value of the distance between that actor's ideal point and any given policy, and then multiplies it by -1 so that bigger distances are bad, regardless of direction.  Simple.  A policy is 10 units away?  -10 "utils."  A policy is 100 units away?  -100 utils.  Yes, the units are called, "utils."  Fucking economists...

So, how do we combine that with, say, a candidate's score on a dimension representing honesty?  Let's say every candidate gets a score along an honesty dimension.  Let's say the policy dimension goes from -1 to +1, just like the NOMINATE scale we use to measure congressional voting scores.  For the sake of consistency, let's say honesty also goes from -1 to +1.  Here's a utility function:

(2)  U(policy, honesty) = - | policy - i | + honesty

That's fine, right?  They're additive.  If a candidate is more honest, that's a bonus.  More dishonest, that's a negative.  The range of scores is -3 (2 units away and a lying sack of shit) to +1 (totally honest, and right at the voter's ideal point).  But, I could write the equation differently.

(3)  U(policy, honesty) = - | policy - i | + 10000000000*honesty

See how that changes things?  Policy no longer matters that much.  It is now all about honesty, just because of that multiplier.  Is Equation 3 more right or wrong than Equation 2?  Nope.  As far as economic theory is concerned, both are just dandy.  Microeconomic theory begins with utility functions as assumptions.  But, there are normative issues built into them.  That last utility function is a utility function that values honesty more than policy, given the scales, and that isn't intrinsically wrong.

What is more important for democracy?  Convergence to the median voter, or honesty among elected officials?  Um, I haven't the slightest fucking clue.  Like I said, that's a value judgment.

But, it is a value judgment critical to the question of whether or not we are comfortable with the notion of campaigns being strategic contests, because if we are not comfortable with campaigns ignoring large gaps in valence characteristics, then we should not be comfortable with campaigns turning on strategy, which brings me back to the beginning of yesterday's post.  Trump, and his apologists want to deflect all attention about 2016 shenanigans away from Russia, Comey, etc., and towards any supposed strategic mistakes Clinton made.  But, if we are not comfortable with elections that ignore gaps in valence traits, that's a problem because, as I have commented frequently, particularly during the "Assessing democracy..." series, there isn't a "valence" trait on which Trump could potentially score higher than Clinton.

Of course, I have only addressed the basic questions of spatial theory here because I'm a spatial theorist, but really, Trump and his apologists haven't been talking about spatial theory.  They have been saying that Clinton lost by, for example, not campaigning in Wisconsin.  That's a separate component of strategy anyway, opening up a separate set of both normative and empirical questions.  That is an argument that a campaign is a chess game between candidates in which voters aren't even actors at all, which is a really interesting argument.  I guess I have more to write...


  1. When people say that "Clinton didn't go to Wisconsin!", two different people likely mean two very things.

    Person 1 is thinking (a la your Risk example next post) that its mechanistic and campaign-based.
    Person 2 is using "go to Wisconsin" as a proxy for "adopt policy positions such that the median Wisconsinite would vote for her." Though, not directly, as you would then physically GO to Wisconsin to highlight those elements of your platform, yadda, yadda, yadda.

    Anyway, the larger point is that the argument is, in a sense, about the platform, but multidimensionally.

    I don't buy it, of course. I'm mostly in the "2 terms plus Comey" camp with you, though I also think we had mostly countervailing forces of pussy-grabbing/douchiness vs a quarter century of being made into the boogeywoman. I think Biden could have beaten Trump, and that Rubio would have won the popular vote against Clinton, but that Biden v Rubio looks pretty similar (we probably hold WI).

    1. I don't think I have ever heard a context in which the "didn't go to Wisconsin" claim was clearly a proxy for policy. Every time I have encountered it, it was pretty clearly about physically going to Wisconsin.

      Anyway, I don't think Clinton was specifically weak, but if she hadn't been on the ballot, you take away the Comey incident, and if Comey cost Clinton the election, then Comey cost the Democrats the election, so replace Clinton with a generic Democrat, and yes, a generic Democrat beats a very weak Republican (Trump), but a generic Democrat versus a generic Republican reverts to Abramowitz. So, I sort of think you are right on the ticket changes, but only really because of Comey.