Thursday, January 5, 2017

Assessing democracy in the aftermath of Trump's victory, Part XV: Policy, valence and voting in 2016

The point of most of this series so far has been as follows:  we have two primary "valence" characteristics, which are the things that voters are just want, for whatever reason, in a candidate.  Competence, and honesty.  These traits comprise the "valence" dimension, along which voters evaluate candidates, along with the left-right dimension.  For both traits, Clinton had an advantage over Trump, once we strip away the bullshit rhetoric of the campaign.  Then, there's the left-right dimension.  While Clinton was by no means a centrist, Trump's lack of policy commitment means that as President, he will basically just rubber-stamp whatever Paul Ryan does, and since Ryan is more of an extremist than Clinton, for at least two years, Trump will act as more of an extremist than Clinton would.  Expecting that to balance out in the final two years is probably too much of a gamble, so we must treat Trump as more of an extremist than Clinton.

Put everything together, and in rational choice terms, Clinton should have had an advantage in the left-right dimension and the valence dimension.  And yet she lost.

Obvious observations here include the common misperception that she was more dishonest than Trump, but I'll tackle a different problem today since I've already tackled that one.

If I wanted to, could I rig a mathematical model in which it is rational for the electorate to pick Trump given their valence traits and policy positions?

Of course I could!  That's what I do!

How would I do it?  Like this.  Notice that I've been talking about policy in left-right terms, as though all policy can be reduced to one dimension.  Can it?  Nope.  Notice, for example, that Bernie Sanders and Donald Trump sound a lot alike on trade.  Why?  To quote Forrest Gump, "stupid is as stupid does."  Anyway, trade policy doesn't line up as a left-right issue.  There are free-trade lefties and protectionist righties.  There are free-trade righties and protectionist lefties.  Not everything is left-right.

If you let me add enough dimensions to the model, I can make anything look rational.

Is it rational to vote for an incompetent, unqualified pathological liar who will rubber-stamp bills from someone who disagrees with you on every other policy because you heard TPP is bad, and you don't even know what TPP stands for?  Um, that's a separate question, but an important one.

If I rig a model to make it look like voting for Trump was rational even though he is an incompetent liar who will govern as an extremist, compared to a more competent, less dishonest person who would have governed in a less extreme manner, I'm not explaining what happened.  I'm just doing something mathematical that requires washing my hands afterwards.

What did happen?  Democrats had won two terms in a row.  A third term is really hard.  It requires not just a growing economy, but a booming economy.  The economy was growing, but not booming.  The economy wasn't good enough.  Alan Abramowitz's "Time for a Change" model predicted a Republican victory, and Republicans won.  We need to figure out what it means that Abramowitz canceled out all of this other important stuff.  More to come on that!


  1. What if you just add a third valence trait?

    Does the candidate have too much vagina?

    1. Grab her by it. When you're President, they let you do that, you know...

      Seriously, though, there is plenty of research on women in House and Senate races, and their vote shares are statistically identical to comparable male candidates, controlling for experience, ideology, etc. Lawless and various coauthors have run the numbers on that. There's that Fox & Lawless paper that says women are less likely to run than comparably credentialed men, but once they enter the race, vote shares are the same.

    2. But also seriously, does the candidate "tell it like it is" or whatever?

      What I mean is: do we need some kind of "valence" trait to explain this? As you note in your later post, Abramowitz is totally atheoretical (on this).
      The closest you come in a "theory" is the basic thermostatic stuff. I really have trouble with ESM's Macro Polity....but is that where we are?

    3. "Telling it like it is," as opposed to honesty as a valence trait? Now you are posing "truthiness" as a valence trait. Well, uh... Uh... I suppose that's... a falsifiable hy...poth.. esis. Is THAT where we are?

      On ESM, that's theoretical. Maybe too theoretical. It is still basically a pendulum-type model, but perhaps overly-elaborate. If it really is just DDRR and turtles all the way down, then ESM are basically just playing around with attenuation of the effects of Abramowitz.

    4. Here's the thing: people voted for him.

      Not just in the general; we can explain that away with partisanship.

      In the primary.

      Issues? Maybe. But that just points to a second dimension of ideology. Personal characteristics? That brings us back to valence.

    5. Why does issue voting in the primary point to a second dimension? The simpler, unidimensional model is the issue public model: immigration. His platform was straight-up conservative on THAT, at least...