The Topline: When bad things happen to good polls
Pollsters live in a constant state of anxiety that some external event will happen while their poll is in the field that will upset public opinion and render their data obsolete. The classic example from recent history is when Superstorm Sandy hit the Mid-Atlantic just before the 2012 election. That was a double whammy: The event and the government response could have changed public opinion on the race, and many voters in one of the most densely populated areas of the country were displaced or otherwise unable to be asked about it. Those times seem quaint and far off now. The news cycle is now a news cyclone that spins 24 hours a day, seemingly every day. This week HuffPost polling editor and pun-dit extraordinaire Ariel Edwards-Levy put to Tweet what many pollsters are feeling this year:
News about the Russia probe, Trump tweets, hurricanes, healthcare, Charlottesville, terrorist attacks, and now taxes overlap and run over each other. This makes the prospect of a stable read of public opinion a pleasant and very distant memory for pollsters. Three or four banner headlines stories could hit while a single poll is in the field, each potentially shifting voter sentiment on key issues.
Sometimes the problem is simply that the questions the poll asked are obsolete by the time the poll is finished fielding. It was a challenge to poll on the Republican health care proposal earlier this year because the policy details changed constantly. Still, data about an older version of a policy may still be relevant, in so far as it gives a sense of how the public is viewing the topic generally.
But what can you do if an external event is so big that it actually changes who answers your poll while it’s in the field? Studies have shown that some shifts in election polling are due to differences in who is answering the survey (i.e. “non-response bias”) rather than anyone changing their mind. But it’s hard for the pollster to tell, in real time, which is which. One study found this effect after Barack Obama’s poor showing in his first debate against Mitt Romney in 2012. Dejected Democrats did not answer surveys for a few days afterwards, and Romney saw a bump in support, even though few minds were changed.
It’s an open question whether the news Monday of indictments and guilty pleas in the Russia probe would have a similar effect in, say, polling of the hotly contested governor’s race in Virginia. But for pollsters trying to get a steady read on public opinion in the midst of a maelstrom of news, it’s one more thing to worry about.
The Horse Race is hitting its stride! Last week WGBH’s Mike Deehan joined us to break down the Boston mayoral debate. This week, Kathryn Burton comes on to dissect the latest political ads in #mapoli. And next week, Horse Race Host/MPG President Steve Koczela will join WBUR for election night -- don’t miss it!
Save the date! November 28th after work, The Horse Race will be live from Ned Devine’s at Faneuil Hall in Boston. Details to come, but trust us -- if you read this far into a nerdy polling newsletter, this is an event for you.
MPG was in the audience for a pollster panel on the American Voter at the Harvard Kennedy School. Click here to watch the full event, which features Trump campaign pollster Tony Fabrizio and The Pollsters podcast host Margie Omero.
Donald Trump hit a new low job approval rating in the Gallup tracking poll early this week. But then he rebounded, even after the Monday news from the Mueller investigation. The lesson here is don’t read too much into the day to day gyrations of a single poll. And remember the margin of error.
The indictments and guilty plea in the Russia investigation have made waves in Washington, but just ripples elsewhere. HuffPost finds 60 percent of Americans say they’ve heard little or nothing in the news recently about the Trump/Russia relationship, and fewer than 3 in 10 say they’ve followed news of the indictments closely.
A CBS News poll finds most Americans support a tax cut for the middle class; majorities support increases for large corporations and wealthy individuals.
Pew’s new political typography report finds that divisions within political parties are just as important a factor in American politics as divisions between Democrats and Republicans.
The polls in the Virginia governor’s race have been all over the map. Politico’s poll watcher Steve Shepard think the difference may be whether polls are calling Virginia phone numbers at random or drawing from a list of registered voters.
The Cato Institute’s 2017 Free Speech and Tolerance Survey finds 71 percent of Americans say political correctness has silenced important discussions society needs to have. Two-thirds say colleges and universities are not doing enough to teach students the value of free speech, while three-quarters think protests against and cancellations of speakers indicate a pattern of how college students handle offensive ideals.
A Washington Post poll finds 36 percent of Americans are not proud of the current state of our democracy. This is not exclusively among Trump critics; 25 percent of Trump supporters hold a similar opinion.
Democrat Phil Murphy leads the race for New Jersey Governor 57/37 percent. Women support Murphy 65/29 percent, while men are divided between Murphy and Republican Kim Guadagno.
According to a new NBC News poll, 49 percent of men say the outpouring after the allegations about Harvey Weinstein have made them think about their own behavior around women.
A new Suffolk University/Boston Globe poll finds one in four Bostonians say they have experienced sexual harassment. This came a few days before Yvonne Abraham’s column in The Boston Globe on sexual harassment in the Massachusetts Statehouse.
A new Gallup poll finds majorities in all three political parties approve of marijuana use, making it one of the least polarized political issues.
Gallup also finds 61 percent of registered American voters say gun control is an important factor when choosing a candidate, up from 54 percent in 2015.
- - - - - - - - - - NERD ALERT TEARLINE - - - - - - - - - - -
Halloween is over. See which candy your kids like the best with this Very Important Poll from FiveThirtyEight. The moral of the story: Americans unite around chocolate and peanut butter. And stop giving out Chiclets and Good & Plenty, you savage monsters. If you stare long enough, the FiveThirtyEight logo kind of looks like a candy corn, so there’s that.
Nerd Alert bonus: booo!