Throughout the 2016 election cycle, there’s been a lot of talk about polls. But despite all of the news surrounding the numbers, there’s not that much conversation about how they’re conducted and how seriously they should be taken.
Here’s what you should know.
Are polls accurate?
In particular, the GOP has seen some news-making inaccurate polling when it comes to House races.
In 2014, the Republican Party advised its politicians to stop commissioning a certain trusted pollster who predicted House Majority Leader Eric Cantor’s easy reelection. Challenger David Brat won the primary race. The pollster behind Cantor’s campaign said a large number of unanticipated crossover voters were to blame.
However, more generally, a single poll just shows a snapshot of an electorate at a specific moment — when a poll is conducted. The news cycle, the political climate, and a candidate’s issue du jour are all factors into what the poll captures, and that shifts frequently.
What’s more accurate is polling averages which show a “bigger picture,” Voice of America notes:
While outlier polls sometimes reflect a new trend, often they’re just an aberration. For a clearer picture, focus on polling aggregators, which bring together data from many different polls to produce an average. FiveThirtyEight and RealClearPolitics each produce well-respected polling averages.
Are polls biased?
They can be.
As FiveThirtyEight’s Clare Malone and Harry Enten note in “We the Voters,” polls are commissioned for different reasons. A news network may put out a national poll to determine the outlook of the election, while, say, a Senate candidate can commission a state poll to determine what voter demographics they should target or which policies they need to pivot on. Additionally, an advocacy group can commission polls to gauge interest in a specific issue.
Pollsters have control over many factors that can influence a poll’s leanings — question order, the number of people polled, the location of the respondents (rural, suburban, urban), their voter status (“likely”, “registered”), and the method in which they’re polled (an online survey, via a landline phone, via cell phone).
Should you take polls seriously?
It depends on which polls you’re taking seriously.
As noted earlier, it’s often the aggregate of polls that can give voters a better idea of how a large swath of voters is leaning, rather than the single polls which show a snapshot of a smaller amount of voters.
Remember Nate Silver, the New York Times pollster credited with predicting the winner in all 50 states during the 2012 election, and who, in 2008, predicted the winner in 49 of 50?
There was no room for anecdotal knowledge in Silver’s predictions. He didn’t make shockingly accurate predictions because of pundit-like “feelings” about voters he met on the street, or the number of people that attended a candidate’s rally. He studied several polls with painstaking detail and inferred a result based on his analysis.
Still, using polling to make predictions isn’t always accurate.
Perhaps the most famous version of this was when the Chicago Daily Tribune used their own political analyst to predict that Thomas Dewey would defeat Harry Truman and get their copy to the presses prior to election results from several states on the east coast.
A few days later, Truman held up the early edition of the paper in front of photographers:
So remember: 2016 is the election where anything could happen, so proceed with caution before having your own Dewey Defeats Truman moment.
“We the Voters: 20 Films for the People” is a social impact and web campaign aiming to motivate young voters around the country, and encourage them to make informed choices. You can learn more about We the Voters here.