Monday, March 21, 2016

Trump to Political Science: Drop Dead, Reboot Part II, sans-Khan

Many analysts are writing off Donald Trump as a sure-loser against Hillary Clinton.  As I continue with the rebooted "Trump to Political Science" series, I think I will take a look at some of the many reasons political scientists are prone to underestimating The Donald.

Let's start with the "Bradley Effect."  The effect comes from Tom Bradley's 1982 gubernatorial campaign in California.  He was ahead in the polls, and then lost.  A similar effect is often claimed with respect to Doug Wilder's Virginia gubernatorial campaign.  So, sometimes polls are wrong, right?  What's the deal?  The deal, according to legend, is that both Bradley and Wilder were African-Americans.  OK, that part isn't a legend.  It's true.  You can look it up on this magical looking-up machine you are using right now!  The question is whether survey respondents claimed they were going to vote for the African-American even though they had no intention of doing so because of race.

The underlying concept in political science is what we sometimes call a socially desirable response.  When a survey question gives respondents one answer that makes them somehow look better, they are more likely to give it even when it isn't true.  So, maybe people were more racist than they wanted to admit, right?

The evidence on either Bradley or Wilder was never conclusive, though.  Then, in 2008 and 2012, the polls nationwide got it pretty much right, despite there being an African American nominee.

How might Trump scramble this?  Well, if Trump's campaign is seen too much as a vehicle for social forces with which people don't want to be associated for reasons including race, then survey respondents may understate Trump's support.

Normally, this kind of thing doesn't affect general elections in particular because there are so many ways to interpret a respondent claiming to support Candidate X.  Will Trump be different?  Possibly.  We might see the biggest Bradley effect ever.  Or maybe reverse-Bradley.  The determinant will be the associations people make with Trump throughout the general election campaign.

That will depend on the media, and Hillary.  You better believe that Hillary will emphasize Trump's racially divisive language to motivate turnout among non-whites.  That may have the effect, though, of creating a reverse-Bradley scenario.  The media will be constrained by the campaign, and Hillary will try to make sure the race issue keeps coming back up.  The more the campaign is about race, the more likely the polls are to understate Trump's support.

Trump will try to change the subject, though.  More on this in a future post, but that brings in the complication.  Trump is a master of getting the press to talk about whatever he wants to talk about.  The question is how these balance out.

The Bradley effect is something of a legend in political science, barely glimpsed out of the corners of our eyes, but always in our minds.  Trump may make it real.  And if he does, our polls will underestimate him.

See?  No Khan.  Fine.  Here's Khan.  No, wait.  That guy isn't even Indian.  You know, like Ricardo Montalban.

No comments:

Post a Comment