Dissecting obsession with polling amid the Ontario election

How our collective addiction to polls explains Ontario’s lacklustre election

In 1996, NYU-based researchers Vicki G. Morwitz and Carol Pluzinski set out to study a very simple question. To what extent, they wondered, was the rising presence of polling in media and election coverage coming to influence the very public whose opinions it was supposed to be measuring? More than 20 years later, the question has only grown more relevant.

Writing in the 1990s, Morwitz and Pluzinski barely had to grapple with the 24-hour news cycle or the internet, both of which today enable the relentless release of polls to a public and media who crave them. Nor did they really have to deal with an electoral landscape where the logic of polling was integrated into daily discourse, not only in the press but also among political activists, partisans and even many ordinary voters. For that reason, the passage of time has only magnified the force of their analysis.

In Canada, as in many countries, the political and media ecosystem has gradually become so saturated with polls it’s now difficult to imagine what elections would even be like without their incessant presence.

On TV panels, in print media and all over Twitter, polling increasingly provides the ambient backdrop of our democratic culture: its ubiquity taken for granted, its dispassionate objectivity seldom questioned. Once a periodic novelty during campaigns and the private instrument of backroom strategists, polls have since come to colour virtually everything about how elections are covered, framed, and discussed from top to bottom — Ontario’s recent campaign being a particularly glaring, and instructive, case in point.

It would be reductive to blame either the province’s record low turnout — which at only 43 per cent saw the incumbent PC government re-elected by just 18 per cent of eligible voters — or the campaign’s general lack of vitality on any single cause. But I believe that the omnipresence of polls bears a very real share of the responsibility.

I started my career as a student of political science, later becoming a staffer and briefly, for a few weeks eight years ago, a candidate. Today I work as a journalist writing mainly about the politics of Canada and the United States. Having written about and participated in campaigns from various angles I’ve come to think that ubiquitous poll discourse, both during and between elections, has increasingly made our politics feel artificial and removed.

As Morwitz and Pluzinski observed more than two decades ago, the basic premise of election polling is that it straightforwardly reflects popular opinion. Thus, if Party A has X per cent support and Party B has Y on the eve of an election, it’s reasonable to expect that those numbers will be borne out on the day itself, within a certain margin of error.

For what it’s worth, while pollsters didn’t predict last week’s election with perfect accuracy, many indeed came fairly close. But to what extent do the polls themselves contribute to producing the very results they anticipate? In the academic literature, the potential effects of polling on voter behaviour may differ in different contexts but they often have to do with shaping or reinforcing perceptions about an election’s outcome.

If, in other words, people continuously see a certain narrative in the media or elsewhere, it can become essentially self-fulfilling. For Tim Abray, a political science researcher at Queen’s University whose research focuses on voter behaviour and voter psychology, this helps explain both Doug Ford’s victory in Ontario and the election’s historically low turnout.

“Election coverage has become a lazy affair,” Abray wrote on Twitter. “Public opinion firms looking for clout publish polls. Constantly. Media, strapped for time, resources and expertise, cover them. Daily. Result: a constant barrage of numbers and not much else.”

“If (polls) are consistent (they were)”, he concluded, “it convinces people that the result is baked in. Why vote when you know it’s futile?”

All of this speaks to a reality that strategists and operatives have in fact long understood: namely, that the received wisdom about how a campaign is going can ultimately affect its outcome at least as much as voters’ actual political or ideological preferences. For precisely this reason, elections are increasingly fought around abstract momentum narratives, and politicians and parties now devote considerable energy and resources to shaping and contesting them. In Ontario, this meant the usual flood of polls from media outlets was suffused by a relentless meta-discourse about who was in second place — a discourse driven heavily by the NDP and Liberal campaigns themselves.

As the election entered its final week, opposition operatives leaked internal polls and partisans from both parties posted the results across social media with the hope of solidifying the idea that their team was ahead. Even in individual ridings, the traditionally less impersonal domain of local politics has given way to the same polling mania.

It’s thus become commonplace for campaigns to repackage ridinglevel polls from major firms for partisan purposes and even advertise their own surveys, often of decidedly dubious quality (the Liberal campaign in my home riding of University-Rosedale, for example, promoted an unsourced poll on Facebook with skewed axes that for some reason featured federal party logos). Sometimes, voters will now even bring up polls at the door.

This process is in turn amplified by what has become an entire cottage industry of projection and strategic voting websites such as StrategicVoting.ca and VoteWell that are supposedly designed to help voters maximize the impact of their ballot by casting it for a candidate more likely to win. An obvious problem with this is that ridinglevel polls are more likely to be inaccurate. Mainstreet Research’s poll for Thunder Bay-Atikokan, to take one random example, overstated Liberal support by 10 per cent while understating the PC vote by the same margin. The Conservative candidate ended up winning. In many cases, however, projections aren’t actually drawn from local polling data at all. As Abray has pointed out, many are based on historical voting patterns or province-wide numbers, and thus extrapolate quite messily from broader trends. Suffice it to say, such projections are a shaky and unreliable foundation for deciding anyone’s vote. Either way, the issue isn’t merely one of accuracy.

Thanks to the prevalence of polling discourse, even riding level campaigning now often revolves around meta-narrative. Ordinary voters, meanwhile, are encouraged to think more like pundits: concerned with horse-race numbers, how leaders are seen to be “performing,” and how others are expected to vote. Taken together, our collective obsession with polls can thus foster a very strange and counterproductive feedback loop: daily opinion surveys discussed throughout the media encourage political parties to try and spin the ensuing narrative to voters who, in turn, engage in a self-defeating effort to game out the electoral map riding by riding.

Because of the various incentives involved, this cycle tends to be selfperpetuating. Media outlets hungry for content pump out horserace stories because they’re easier to assemble and often attract a larger audience than detailed issuebased material.

Partisans, operatives, and canvassers, determined to show momentum, spend their time arguing about how their chosen team is perceived by voters instead of debating the issues or promoting their election platforms.

As for voters, who are incessantly told what the hivemind of public opinion thinks before they’ve really had the chance to formulate an opinion, some end up building horse-race calculations into their own thinking.

The result, as we’ve just seen in Ontario, is often sclerotic elections that feel more like abstract spectacles than meaningful exercises in democratic engagement. And while there are many possible explanations for last week’s poor turnout, it’s easy to understand why so many responded to the blitz of poll and horse-race chatter by simply switching off or staying home.

As far as solutions go, there’s certainly no silver bullet. Thanks to particularly egregious errors in polling, such as projections of Britain’s Brexit vote or the 2016 U.S. presidential election, there have been more critical discussions of polling, but most have hitherto been concerned with improving accuracy. Ultimately, though, more substantive and engaging elections will require us to break our collective addiction to it. That begins with the simple recognition that, far from merely reflecting public opinion, polls have also come to actively influence it as well. Alongside the long overdue reform of our electoral system, stricter regulation of push polls in campaign law could help put an end to tiresome metadebates about where parties stand and encourage greater discussion of the issues. Though it might be subject to court challenges, the actual blackout of opinion polls during a specified period is already on the books in many countries (it’s as long as ten days in places like Lebanon and Ecuador) and is arguably just as valid as the various restrictions on political advertising which are currently in place here.

The media certainly has an important role to play as well. TV panels, daily reporting, and debate coverage might take the harder (though ultimately more constructive) route of de-emphasizing numberdriven narratives and recentering discussion around public policy. The alternative, as Ontario’s uninspiring campaign rather depressingly demonstrates, is a political ecosystem so saturated with polls it’s often closer to sports or infotainment than it is to democracy.

LUKE SAVAGE IS A WRITER BASED IN TORONTO WHOSE WORK HAS APPEARED IN THE WASHINGTON POST, THE ATLANTIC AND THE GUARDIAN. HE WAS A CANDIDATE FOR THE NDP IN THE 2014 PROVINCIAL ELECTION. HIS FIRST BOOK, THE DEAD CENTER, WILL BE PUBLISHED LATER THIS YEAR BY OR BOOKS.