They’re cited by candidates in stump speeches. They determine which candidates make the debates. They control where candidates stand on those debates’ stages.
As they gain seemingly ever-increasing importance, are political polls creating a negative feedback loop that is actually making America less democratic?
In a New Yorker article out on Monday, Harvard history professor Jill Lepore suggests that major errors in the ways polls are conducted may actually be providing an incomplete and inaccurate picture of the race — at a time when polls are having an increasingly large effect on the shape of modern politics and political coverage.
Lepore cites numerous top pollsters and experts who point out several flaws in current polling methodologies, including:
- A massive drop in polling participation. Lepore notes the steady decline of responses to polls from around 90% participation during the advent of polling to single digits today.
- Landline polling. A shrinking number of people own and use landline phones, the traditional method of connecting with voters, while reaching voters on their cell phones is often difficult because of legal barriers.
- Unreliable online polling. Pollsters still have not figured out how to effectively find representative samples using online polls, which tend to reach younger voters.
“Pollsters rose to prominence by claiming that measuring public opinion is good for democracy,” Lepore writes. “But what if it’s bad?”
A slate of recent elections in which polling has been wildly incorrect has clearly unnerved pollsters and political analysts, who appear to be increasingly wary of the over-reliance on electoral forecasting.
On Monday, Larry Sabato’s Crystal Ball suggested it could not rely on polls to predict the outcome of the Louisiana governor’s race later this month. Kyle Kondik, the managing editor of the Crystal Ball, cited the inaccurate polls in the gubernatorial race in Kentucky, where governor-elect Matt Bevin (R) outperformed polling expectations by around 12%.
But not everyone agrees that polling’s increased importance in the early stages of the race is necessarily a bad thing.
University of Michigan polling expert Michael Traugott argued pollsters are improving their methods, increasingly relying on statistical modelling rather than simply aggregating the responses from surveys.
“She operates from a common assumption that survey research is a static, out of date field,” Traugott said of Lepore. “Quite the opposite is true; like any scientific endeavour, its methods are under constant review and revision.”
“The accuracy of pre-election polling has actually improved, not worsened,” he added.
But anyway you slice it, polling has already had an enormous effect on the 2016 race.
With 17 Republican candidates originally vying for the presidency, the television networks hosting the presidential debates have all used national primary polls to determine who qualifies for which stage.
This selection method has boxed out candidates who have loomed large in past elections, including former Sen. Rick Santorum (R-Pennsylvania) and former Texas Gov. Rick Perry (R). Perry dropped out of the race in large part due to dismal poll numbers that made it nearly impossible for him to raise money and attract new donors.
Pollsters and campaigns alike have complained about using early polls to determine who qualifies for the debate stage, a platform that many campaigns believe elevates certain candidates in voters minds.
“I think most pollsters agree that using our surveys more than six months before anyone even votes to decide who gets to be in the debates is a terrible process,” Public Policy Polling Director Tom Jensen told Business Insider before the first debate in August.
Sen. Lindsey Graham (R-South Carolina) put the problem more bluntly.
“It sucks,” Graham told Morning Joe in July when asked about the possibility of being shut out of the main debate. (He has not qualified for the main stage in any of the four GOP debates, and he also did not make the cut for Tuesday’s GOP “undercard” debate.)
Even campaigns that have qualified for the main stage at debates have taken issue with the over-reliance on polls by journalists and media outlets.
The campaign of Sen. Rand Paul (R-Kentucky) assured Business Insider earlier this year that current polls were under-representing the senator’s support because many of his potential backers are students and young voters who do not use landline phones.
Other experts, however, contend that while early polling should be taken with a grain of salt, campaigns may not want to take too much comfort in problems with polls.
“I think the landline problem is a red herring, because cell phone surveys can also make mistakes. Both kinds of surveys have trouble getting a sample that represents people who are likely to vote,” Princeton University polling expert and professor Sam Wang said in an email.
“Rand Paul’s problem is not landline surveys. His problem is that very few people support him.”
NOW WATCH: Bernie Sanders hugged a Muslim student onstage and vowed to fight back against Islamophobia
NOW WATCH: Briefing videos
Business Insider Emails & Alerts
Site highlights each day to your inbox.