The 2016 Polling Failure
What actually went wrong with the polls in 2016, what was misunderstood, and what the industry learned from the experience.
Separating Myth from Reality
The narrative that 'the polls were wrong in 2016' is both true and misleading. National polls predicted Clinton would win the popular vote by about 3 points; she won it by 2 points. That is within historical norms of polling accuracy. The failure was in state-level polls, particularly in Wisconsin, Michigan, and Pennsylvania, which underestimated Trump's support by 4-7 points.
The American Association for Public Opinion Research conducted a post-mortem and identified several factors. The most significant was the failure to weight by education. Non-college-educated white voters, who strongly favored Trump, were underrepresented in poll samples. Pollsters who weighted by education had smaller errors. Late-deciding voters also broke disproportionately for Trump, and polls conducted before the final weekend missed this shift.