Why You Shouldn't Worry About Polling

August 6th 2016

Aimee Kuvadia

The time leading up to a presidential election can feel like a circus. And public polling — the frequency of it, the ubiquity of it, and the media's obsession with it — may be why.

Case in point: a recent Public Policy Polling survey found 5 percent of Americans would vote for the slaughtered gorilla Harambe in the presidential election if given the option.

Poll Showing Harambe favorability

Harambe's poll numbers raise an important question: should Americans place so much stock in a science that sometimes produces frivolous results?

First, a disclaimer. Not all polls are created equal.

Taking a trip down memory lane to the 2008 and 2012 elections, one can see that August polls did come pretty close to predicting the November election's results.

In August 2012, President Barack Obama garnered around 46 percent of the vote in the polls and Mitt Romney had around 44 percent; in the final election, Obama won with 51 percent of the vote and Romney had about 47 percent.

Likewise, in August 2008, then-Sen. Obama had around 47 percent of the vote in the polls and Sen. John McCain had around 44 percent of the vote. In November, Obama and McCain had around 52 percent and 44 percent, respectively, according to Real Clear Politics.

Blogger Nate Silver made himself into a household name thanks to predictions he made based in part on those public polls. Mashable reported that in 2008, his mathematical model called election results in 49 out of 50 states, but in 2012, it accurately called all 50 states.

But polls can be dead wrong, too.

Take the 2014 midterm elections, in which former Republican House Majority Leader Eric Cantor suffered a historic upset after seasoned pollster John McLaughlin predicted a landslide primary victory in which Cantor would win by 34 percentage points. Instead, he lost by 10, Ben Terris wrote in The Washington Post.

For his work, Terris points out McLaughlin had been paid $70,000. So, of course, his miscalculation should have been a career-breaking move, right? Actually, no. Politicians continue to put their faith in this pollster even today. In fact, Trump hired McLaughlin to focus "exclusively on New York, polling to determine what type of climb Mr. Trump would face in a state that hasn’t voted for a Republican in a presidential race since Ronald Reagan in 1984," The New York Times reports.

Who are these people who participate in polls, anyway?

Another factor contributing to the inaccuracy of polls are the individuals being polled. When people opt out of responding to polls, they are essentially skewing the accuracy of the poll results. Similarly, when people do respond, they are seen as "self-selecting" into a sample — think about it: it takes a special kind of person to willingly sacrifice five or so minutes out of their day to take a poll.

Regardless, samples aren't entirely random because “if the people who participate in the poll are different from those who do not, results can be biased because of these differences,” according to the American Association for Public Opinion Research.

If you're not a politician, why should you care about polling accuracy?

There's no question political polls have at least a slight impact, if not more, on how citizens vote in elections, according to a University of Michigan study that analyzed a 1988 Canadian election.

Below are the four major takeaways from the results of the study:

“Polls affected voters’ perceptions of the various parties’ chances of winning.”

“Polls affected the vote.

"Polls affected strategic voting as some voters became less inclined to support a party whose chances of winning appeared slim.

“Polls did not have a contagion effect, since voters did not come to evaluate the parties and leaders who were doing well in the polls more positively."

What does all this mean?

The polls are not currently in Trump's favor — he's polling at 41.2 percent versus Clinton who's polling at 47.5 percent, according to Real Clear Politics. And while it's tempting to write off his electoral chances entirely, a lot can change between now and November.

And even if Trump doesn't mount a real comeback, that won't necessarily stop the media from making it seem like he is. As former Obama speechwriter Jon Favreau noted in The Ringer this week, all it will take is one poll hinting at a Trump surge to incite a frenzied comeback narrative.

"Any day now, some Quinnipiac poll that shows a tied race in Pennsylvania will force Democrats to lose control of their bladders. A Trump surge in a stray tracking poll will result in a CNN Breaking News Countdown Clock that will tick down the seconds to an emergency panel of 37 pundits," Favreau writes.

When that happens, just remember that, at one point, a dead gorilla was polling at 5 percent, and calm down.