The 2016 presidential election was not the first to incorrectly predict the candidate that won the election. Hillary Clinton (2016) and Alf Landon (1936) were predicted to win presidential elections. The 1936 polling error, when survey research was in its infancy, was mainly attributed to non-response bias. The American Association for Public Opinion Research (AAPOR) has proposed three hypotheses that they do not believe are problematic for future survey research for political polling and one that is for the 2016 polling error. These are

(1) undecided voters broke for Mr. Trump in the final days of the race, or in the voting booth

(2) turnout among Mr. Trumps supporters was somewhat higher than expected

(3) state polls, in particular, understated Mr. Trumps support in the decisive Rust Belt region, in part because those surveys did not adjust for the educational composition of the electorate.

The AAPOR believes that these errors are fixable errors. However, some argue that the problem is non-response error again as less civically engaged voters did not respond to the survey research. To what extent do you agree that the polling errors from the 2016 presidential election are fixable as they were in 1936?