The 2016 election polls were inaccurate, predicting a win for Email-Server-Hillary, whereas Mussolini-Arpaio-Trump prevailed and now is the Oval Office One. But pollsters, because they’re addicted to asking, are seldom able to admit their mistakes. Instead, they try to deceive by putting a positive gloss on their failures. In the newsletter below, “AAPOR Releases Evaluation of 2016 Election Polls”, which I received from a university survey research center, I indicate in bold the deception and in brackets [ ], bold, and italics my comment on it.
AAPOR Releases Evaluation of 2016 Election Polls
On May 4, the American Association for Public Opinion Research (AAPOR) released its much anticipated report concerning the accuracy of 2016 national and state election polls in the U.S. Key conclusions from that report include:
“National polls were generally correct and accurate [generally? “Generally” is not accepted in social SCIENCE. In science we need to know which specific polls, and how many of them, were correct and accurate, and which ones, and how many, were not] by historical standards” [The standard for scientific correctness and accuracy is correspondence with reality, not what pollsters accepted historically, in the past, for correctness and accuracy.]
“State-level polls showed a competitive, uncertain contest [this is a positive gloss on state-level polls that attempts to mute the following comment that these polls under-estimated Mussolini-Trump’s support] but clearly under-estimated Trump’s support in the Upper Midwest” [A non-deceptive statement would read: State-level polls clearly under-estimated Trump’s support in the Upper Midwest.]
There were multiple reasons why the polls under-estimated support for Trump,[Yeah! Non-deceptive statement] including:
“Real late change in voter preference during the final week of the campaign”
Adjustments for over-representation of college graduates was necessary, but many polls failed to do so [Yeah! Non-deceptive statement]
“Some Trump voters who participated in pre-election polls did not reveal themselves as Trump voters until after the election, and they out-numbered late-revealing Clinton supporters” [Yeah! Non-deceptive statement]
“Ballot order effects may have played a role in some state contests, but they do not go far in explaining the polling errors” [Yeah! Non-deceptive statement]
Predictions that Clinton had a very high probability of winning “helped crystalize the erroneous belief that Clinton was a shoo-in for president, with unknown consequences for turnout” [Yeah! Non-deceptive statement]
“A spotty year for election polls [A non-deceptive comment would read: The 2016 presidential election polls failed to predict the winner] is not an indictment of all survey research or even all polling” [Failure IS an indictment! A non-deceptive statement would read: Failure of the 2016 presidential election to predict the winner is another one of the many examples of the unreliability of polling]
For a complete assessment of polling and other forms of survey research, see my book, THE PROBLEM WITH SURVEY RESEARCH.