We’d be better off without pollsters
Americans should settle in for a long night Tuesday to watch the U.S. Senate runoffs in Georgia, we’re told, because pollsters say the races will go down to the wire.
FiveThirtyEight polling guru Nate Silver noted Wednesday that while “many prestigious pollsters [are] sitting the Georgia runoffs out,” plenty of others are all in. Polling averages showed Democrat Raphael Warnock 0.5 percentage points ahead of Republican Sen. Kelly Loeffler, while GOP Sen. David Perdue had a 0.4 point advantage over Democrat Jon Ossoff. “We aren’t planning to make probabilistic forecasts in Georgia,” wrote Silver, “but it’s safe to say that a ‘polls-only’ view of the runoffs would put each race at about 50:50.”
Or not. After polls performed so miserably in the presidential race in 2016, especially in several critical swing states, Americans were assured that pollsters had corrected their mistakes. Wrong. In an opinion article for The Washington Post after November’s contest, pollster David Hill admitted what many have long suspected: “People don’t talk to us anymore, and it’s making polling less reliable.” Hill acknowledged that Trump supporters were particularly difficult to reach, and he concluded that “we no longer have truly random samples that support claims that poll results accurately represent opinions of the electorate.”
Based on previous undervaluing of Republican support – as evidenced by inaccurate state polling of November’s presidential and congressional races – my best guess is that Loeffler and Perdue probably each have comfortable leads. But right or wrong, when the results are known, the only benefit of polling will be bragging rights for anyone who was close, and egg on the face of everyone who wasn’t.
Our focus on polling and other horse-race aspects of campaigns contributes to our tendency to conflate elections with sporting events rather than a consideration of competing ideas. Irresponsibly equating campaigns with boxing matches or pennant races – where one participant comes “off the mat” or another “makes up ground” in the closing days – also bleeds into election night coverage. An example of the harmful effect of treating elections like sport is found in the question asked accusingly today by the president and his supporters based on early vote counts: “How could President Trump have been so far ahead in so many swing states but end up losing?”
Of course, neither Trump nor President-elect Joe Biden was ahead or behind in any state after the polls closed. One of them had already won. Discerning the winner was achieved by counting all the votes, but reporting early results as though campaigns were still active – “Biden has a lot of ground to make up” – contributed to suspicions of foul play. In reality, what seemed like a wild swing in one candidate’s direction simply represented known political patterns as votes were tabulated from different regions and voting methods.
Granted, not reporting early totals would obliterate an entire industry of experts on television standing in front of interactive maps explaining what votes are left to be counted and which candidate they’re likely to favor. Can’t we live without that? And while we’re killing off outmoded practices, how about bidding farewell to news media “decision desks” and waiting until the actual vote counts determine the winners?
But confusing elections with sports starts with our unending and unhealthy obsession with who’s ahead. In 1948, Time magazine profiled groundbreaking pollster George Gallup, who made his mark in politics by accurately predicting, against another prognosticator, that Franklin D. Roosevelt would prevail in the 1936 presidential contest. Time noted that after that election, Gallup knew he had a challenge in “convincing newspaper publishers that there was still news for pollsters to report.”
Across the media landscape, pollsters have succeeded on that front beyond Gallup’s wildest dreams. But predictions supposedly based on scientific methods set up expectations that, when proved wrong, lead to accusations of election shenanigans. Imagine instead a world without the daily tracking polls, or random focus groups, or the constant reminders of who’s ahead or behind. Such a dramatic shift would be jarring, but beneficial if the result were more time discussing ideas than critiquing strategy.
But we’re still far from that world. On Tuesday, coverage of Senate races in Georgia will begin on the cable stations long before the polls close. First, they’ll tell us exit poll results on the most important issues for voters – ironic considering how little coverage issues receive – followed by early returns shortly after the voting ends, followed by experts pointing to maps telling us where the votes have yet to be reported and whom they probably will favor, followed by “breaking news alerts” and play-by-play calls of candidates “making up ground” as they “come from behind.”
Eventually we’ll find out who won, which is the most important thing, leading to more analysis of how the results matched up to the polls, which isn’t important at all.