Viewpoints can change in four or five years, which is a potential limitation of the study. However, there’s a lot of value in knowing a large sample’s preferences and then observing what they do. When one sub-sample votes A and the other votes B, you can then go back and observe where those sub-samples differ. That difference probably played a role in their respective voting decisions.

Though not impossible, it’s unlikely the portion of the sample that ended up voting Sanders — or any other large sub-sample — dramatically changed its viewpoints from 2012 to 2016 on the various issues included in the study.