Thursday, May 5, 2016

Political scientists can't tell you why Trump won, and neither can anyone else

For the last couple of days, I have been basking in my own ego for seeing Trump's victory before the rest of the crowd (even though I had to pull my head out of my own ass to do it), but now comes the difficult question:  why did Trump win?

Political science cannot tell you.  Science is a method, not a subject.  We observe, hypothesize, test our hypotheses, and update our models accordingly.  Whether the subject is chemistry, physics, biology or politics, that process is the same.

Diatribe time.  Want to get under my skin?  Talk about "hard" science versus "soft" science.  The "hard" label refers implicitly to either experimentation or mathematical sophistication.  There are plenty of experiments in political science, but more importantly, if "hard" science requires experimentation, then Einstein wasn't a scientist.  Mathematical rigor?  Those of us trained in statistical modeling and game theory use more difficult math than many in the physical sciences (the proper term).  "Physical" science, not "hard" science.  "Social" science, not "soft" science.  End of diatribe.

Regardless, science is about finding patterns.  Trump's victory was fascinating precisely because it broke so many patterns.  The last time a major party nominated someone without prior political experience, it was the guy who beat Hitler.  Trump's policy positions are an incoherent mess, with a history of statements placing him to the left of Bernie Sanders.  He lies, not just more brazenly, but more transparently than Tommy Flanagan.  He incites his followers to violence, and promises to pay their legal bills.  He brags about the size of his genitalia during a presidential debate.

This... is really weird.  Political science is about finding patterns.  This breaks from the pattern.  So let's talk about the philosophy of science.  One of the most important books for you to read is Thomas Kuhn's The Structure of Scientific Revolutions.  If you haven't read it, you aren't an educated person.  Go read it.  Kuhn distinguishes between "normal science" and the process of a "paradigm shift."  Normal science occurs when we solve small puzzles within an existing model (or, "paradigm," if you are a pretentious ass).  However, sometimes something happens that is unexplainable within the context of an existing model:  an anomaly.  The scientific problem is that we don't know in advance which weird observations are puzzles, solvable within the context of existing models, and which are anomalies that require a paradigm shift, meaning the adoption of a new model that can explain more stuff.

Trump is a problem.  Is he a puzzle?  Is he an anomaly?  We don't know.  All we can do in political science is find patterns, and it looks like Trump breaks from the pattern in some "yuge" ways.

What can we do?  Once the academic surveys come out, we will be able to tell which types of voters were more likely to support Trump, and which types were less likely to support him.  The obvious variables to consider are as follows:  opinion on immigration policy, attitudes towards immigrants, perceived threat of terrorism, attitudes towards muslims, beliefs on economic redistribution, religiosity, trust or distrust in Republican "leaders," etc.  I can keep going, but the point is that once the surveys come out, we will develop some statistical models that can tell us, given the considered variables, the probability that any one voter supports Trump.

But that's not the same thing as explaining why Trump won.  That brings us to the basic nature of "causation."  If A, then B.  If not A, then not B.  These are different statements.  The former means that A is a "sufficient" condition for B to occur.  The second means that A is a "necessary" condition for B to occur.  When both statements are true, then A is a "necessary and sufficient" condition for B.

How do we know when we have a necessary and/or sufficient condition?  Since we can't go back in time and experimentally manipulate the past, all we can do is observe what happens when A is true, and when A is not true.  But if we only have one case in which B is true, then we just don't have enough variation to study.

Political science, and science more broadly, is ill-equipped to explain individual events.  Sure, we can tell stories, and stories are fun!  But I can always construct different narratives for the same sequence of events.  Narratives are a bullshit method of assessing causation.  We need multiple observations, or we aren't engaged in science.

We don't have that.  We may not for a while, if ever.  Why did Trump win?  I have no idea, and neither does anyone else.  Anyone who claims to have an answer is full of shit.

No comments:

Post a Comment