Monday, September 19, 2016

Another episode of "Donald Trump is a social science experiment": ground game and polling

I have been writing periodically about how Donald Trump's weirdness as a candidate allows us, as social scientists, to see what happens when we change something that is normally a constant.  In methodological terms, this is actually a problem because Donald Trump is different from other candidates in so many ways that we can't tell which kind of weirdness is really important, but it is worth considering.

Today, as we consider Donald Trump's polling surge, let's consider what the polls can't pick up.  Normally, candidates each have a "ground game."  I hate the term, but I can't make it go away because everyone else uses it.  It refers to the candidates' get-out-the-vote operations.  Each campaign has offices scattered around swing states devoted to mobilizing supporters on election day.  Why?  Voting is a hassle.  Not a big one, but a hassle nonetheless.  Employees and volunteers help reduce the hassle through organization, and nag people into voting despite the hassle.  Normal campaigns worry a lot about their ground games.

Trump, as in so many ways, is different.  His campaign, as many have noted, has no ground game.  That's a campaign asymmetry that polling can't pick up.

Will it matter?  I don't know.  But maybe we should factor that into our interpretation of the polls.

If ground games matter and normal polling methods assume a symmetry that isn't there, then current polls assume a non-existent but important Trump operation, and overestimate how many people will show up for him on election day.  It's not that they get his support wrong in the population-- it's that they get the election day effects wrong.  Maybe.  If the ground game stuff matters.

Then again, we've never seen a candidate operate this way, so we don't know what happens, and any lefty, Clinton partisan types who are panicking about the polls should be careful not to read this post and go down the rabbit hole of "unskewed polls," as the Republicans did in 2012.  Back in 2012, Republicans just convinced themselves that all of the polls had to be wrong, because, well, they just couldn't be right.  Yeah, they were right in 2012.

As usual this year, lots of caveats.

No comments:

Post a Comment