Sunday, May 15, 2016

Political Science to Trump Denialists: Drop Dead

If you head on over to RealClearPolitics, you'll see that Clinton's lead in the polls is, on average, just shy of 6 points at the moment, and there is even an outlier poll with Trump holding a 2 point lead over HRC.  What's going on here?

I started this blog with the "Trump to Political Science: Drop Dead" series, detailing the ways that Trump's victory in the Republican nomination contest has forced political scientists to eat crow.  This time, it is political science's turn to gloat.

If there is one point I beat into my students' heads, over and over again, it is the centrality of party identification in all things electoral.  Any poll that tells you that "independent" voters are a significant portion of the electorate is done by people who don't read or understand political science.  Most voters who claim to be independent are lying.  They are closet partisans, whose opinions and behavioral patterns are clearly partisan.

In fact, in any given year, between 85% and 90% of all votes cast are just partisans voting with their party.  Republican voters, for the most part, will vote for the Republican on the ballot.

But wait, you'll say, Trump isn't a Republican!  And you'd be right.  Trump has no loyalty to anyone or anything other than Trump.  But that means he's also not a Democrat.

The political scientist whose work is most important here is Alan Abramowitz, who has coined the term, "negative partisanship."  It isn't that Republicans really like the Republican Party.  They just hate Democrats more.  HRC is a Democrat.  Most Republicans will vote for Trump just because he's not a Democrat.

And that means there is very little room for deviation from 50-50 in a presidential election.  And how should we understand that process?

With more from Alan Abramowitz.  Long ago, he built a model of presidential election forecasting called the "time for a change" model.  It did a remarkable job forecasting outcomes using just three variables:  GDP growth, the president's approval rating, and an indicator variable for whether or not one party has won two elections in a row.  After two victories, the other party tends to get a turn.

During the 2008 contest, Abramowitz gave a presentation at the American Political Science Association meeting in which he made fun of people who made ad hoc adjustments to their models, and scoffed at those who thought about changing their predictions because suddenly one nominee was an African-American, etc.  Abramowitz stuck with his old model, and it worked.

Then, he made a change, so to speak.  And it turned out to be at least somewhat justifiable, if a bit hypocritical after his 2008 rant.  He observed that partisan polarization limited the range of potential outcomes in a presidential election.  There is a floor for any candidate, no matter how weird they may seem.  And the floor is pretty high.

Add to that the fact that the Democrats have won two in a row, and a tepidly growing economy, and we really shouldn't be shocked that the election is relatively close.

For 2016, one party has nominated Tony Clifton.  Strangely enough, political science still works.  Our models of general elections were always better than our models for primaries anyway.

This will be a closer election than non-political scientists predict.  And Donald Trump might win.

No comments:

Post a Comment