Polling

Nate Silver: “With the 2018 midterm elections approaching, we’ve updated FiveThirtyEight’s pollster ratings for the first time since the 2016 presidential primaries. Based on how the media portrayed the polls after President Trump’s victory over Hillary Clinton later that year, you might expect pollsters to get a pretty disastrous report card.”

The Associated Press said that “it will begin conducting an elaborate election voter survey designed to replace the traditional in-person exit poll, which has been criticized in recent years for inaccuracy and failing to keep up with changes in how Americans vote,” the AP reports.

“The new AP VoteCast service, developed with NORC at the University of Chicago, uses a combination of online and telephone surveys conducted four days before Election Day and through the close of polls. In all, AP expects to conduct more than 85,000 interviews with voters for this year’s midterm election survey… That’s far more than the roughly 19,400 conducted by the exit poll in 2014… allowing for a deeper and more accurate understanding of the electorate.”

This is great: Civiqs lets you track public opinion on dozens of issues, candidates, and campaigns, so you can see how events affect public opinion in real time.

The dashboard is updated daily with current measures of President Trump’s job approval rating, the 2018 generic House ballot, favorable ratings of the Democratic and Republican parties, attitudes towards gun control, and much more. There are no gaps between polls because polling is done daily. All of the results can be filtered and tabulated by age, race, gender, education, and party identification.

Despite the perceived failures in recent high-profile elections, a  new study in Nature Human Behaviour — conducted across 45 countries over the last 75 years — finds that there is no evidence that election polling is any less accurate now than it was in the past. If anything, polling may be becoming more accurate.

Looking at the crosstabs of the new USA Today/Suffolk poll, there’s a white Republican woman from the Midwest who says she’s never heard of Donald Trump.

A new Public Opinion Quarterly article finds that late movement to Donald Trump and a lack of education weighting were the best explanations for the state polling error in the 2016 presidential campaign.

Interestingly, there is little evidence that “shy voter” responses were an important contributor to polling error.

Pew Research: “These probabilistic forecasts can give potential voters the impression that one candidate will win more decisively and may even lower the likelihood that they vote… The use of such probabilistic forecasts was a constant in coverage of the 2016 presidential race, with an average of 16 mentions per day in cable news broadcasts, according to the study. And at least in 2016, outlets with more liberal audiences featured more coverage. Forecasters uniformly favored Hillary Clinton to capture the White House, with odds ranging from 70% to 99%.”

“The new study finds that numbers like these can leave people with the impression that the race is far less competitive than when they see polling data presented as the percentage of the vote they are expected to get – something familiar to the public.”

“When the wave comes, it’s always underestimated in the polls. That is the reason that Republicans are ducking for cover.”

— A conservative political strategist, quoted by the Washington Post.