BOSTON (CBS) – Many pollsters predicted a big Joe Biden win, and some foresaw a Democratic takeover of the Senate. But for the second presidential election in a row, those forecasts were wrong.
The pollsters we spoke with today pointed out that most of the presidential state polls were fairly accurate, within their margin of error. But for a variety of reasons – structural problems like failing to accurately estimate the turnout of certain pro-Trump voter groups – some were way off.
Look at the gaps between the Realclearpolitics.com average of polls and the unofficial returns in three supposed battleground states: in Texas, Trump only up by 1% in the final polling, wins by 6%; in Ohio, Trump again up by only 1% according to the polls, but wins by 8%; and in Iowa, Trump by 2% in the polls, also winning by 8%.
Prof. Joshua Dyck, head of the UMass Lowell Center for Public Opinion, told WBZ-TV the main problem isn’t the polls themselves, but the way they’re misused to make predictions that just don’t hold up. “As a public institution that’s trying to do polling to provide knowledge and information to the public – not just to characterize everything as a horse race, ‘who’s ahead, who’s behind’ – that’s, for me, the purpose of why we do this. And the moment it just becomes a data point in a projection model that spits out a probability of who’s ultimately going to win, that can actually have some negative implications on the democratic process,” he said.
Dyck is referencing the work of prognosticators like Nate Silver of 538.com and Nate Cohn of the New York Times, who aggregate and analyze polling data and create probabilities of victory that can be highly misleading. It’s a form of tabloidization of data to create pot-stirring clickbait: Wow, look at the needle, a 95% chance of victory for (insert candidate here).
Legitimate pollsters who missed the mark will redouble efforts to fix their modeling and weighting. But in the end, it’s up to the news media and the citizens to do a better job of understanding what they’re looking at and taking it with a pound of salt.