Based on the results of the Michigan Democratic primary, you might remember that Harry Enten at FiveThirtyEight wrote this:
Sanders’s win in Michigan was one of the greatest upsets in modern political history.
The reason it was considered such a huge upset is because polls showed Clinton with a 25 point lead going in to the primary, but Sanders won 50/48. Until we have some understanding of how pollsters got that one so wrong, it is hard to take any of them very seriously.
My contention is that a big part of the explanation on why the polls were so wrong in Michigan is that Independents made up 27% of voters and they supported Sanders overwhelmingly (71%). Clinton actually won in Michigan among registered Democrats (58%).
Rumors have floated (as they do in most primaries) that – because the Michigan primary is open – some Republicans identified themselves as Independents to vote against Clinton. But I suspect that even if that was true for a few people, it would amount to minimal impact.
What is much more likely is that Sanders’ supporters – who are overwhelmingly young – are Independents who haven’t yet affiliated with a particular party and in many cases, might have been voting in a primary for the first time. Pollsters probably missed these voters when setting up their “likely voter” screens.
One way to test whether or not my theory is accurate will be to watch the results in Florida – because it is a closed primary. As of today, Clinton has a big lead in the polls there (RCP average is running at more than a 25 point lead). If Sanders pulls off another upset in Florida, my theory is completely toast. But if the polls there are accurate, perhaps there’s something to it.
On the other hand, both the Illinois and Missouri primaries are open. In Illinois, Clinton has a 2 point lead in the polling averages, but there haven’t been enough polls conducted in Missouri to garner much of a prediction (although FiveThirtyEight gives Clinton a 54% chance of winning there – a virtual tie at this point).
None of these states is a perfect replica of Michigan, so they will be incapable of addressing exactly what went so wrong with the polls in that state. But it will be interesting to watch how the current polls fare on a day with both open and closed primaries. That might tell us something about what (if anything) the pollsters missed in Michigan.
