Should opinion polls be banned during elections?

Ontario’s Chief Electoral Officer thinks they shouldn’t be published beyond a certain date.

In a recently released post-mortem report on the 2022 Ontario provincial election, the province's Chief Electoral Officer suggested that the historically low turnout of the last election could be partly blamed on election polls that showed Premier Ford's Progressive Conservative government cruising to another majority government.

He then went on to recommend that there should be a publication ban on election polls two weeks before election day.

As one can imagine, this recommendation has created quite a buzz, particularly among pollsters themselves. 

As someone who has participated in many election campaigns, I've seen elections shift after polls have been released. But whether or not this is a bad thing and whether or not polls should be banned is another matter entirely. With many high-stakes elections on the books across the country in the next several years, this is an important topic.

So to give you a more objective take on this matter, noted Canadian polling analyst and author of The Writ Substack, Eric Grenier, joins me today to debate the critical points of this issue, including whether or not a ban on publishing polling results is a good idea.

1.) Do election polls influence the results of elections?

MRG: Yes, but the question is to what extent. 

As Abacus Data's David Coletto presents on his Substack, I've seen situations where polls have likely influenced the outcome of elections. For example, the bandwagon effect Coletto mentions can be particularly acute during a change election; an election where the governing party loses power. This 2015 analysis you wrote, Eric, just days before election day, suggested the Liberals would form a minority government. The media was saturated with polls showing the Liberals would win, and in the end, they ended up with a majority government. 

Did the polls influence the outcome of that election in a bandwagon effect? Maybe. 

But attributing the change in government to the publication of polls alone seems a stretch. For example, if polls had suddenly shown the federal Conservatives holding onto power during the same time, I'm not convinced that alone would have caused more people to cast ballots for us. Instead, as a candidate who campaigned in many parts of the country during that election, the polls confirmed what I heard on the doors rather than causing the effect itself. 

The better way to describe the effect of polls is that they give voters validation to proceed with behavior they've probably already committed themselves to based on the overall tone of a campaign and its top issues.

EG: I think that's true, but there's also the potential for polls to reveal to the general public what you on the ground might be hearing at the doors. The 2011 federal election is a case in point. Without polls, would we have taken seriously any claims that the NDP was breaking through in Quebec? I remember seeing the first poll putting the NDP ahead in the province; I didn't believe it. And I certainly would have chalked it up to spin if it was an NDP candidate telling me about how good the reception had been lately!

Revealing what was going on had an impact. It allowed other voters to see what was happening around them and incentivized those tired of the Bloc and the Liberals to give Jack Layton's party another look. That could be a bandwagon effect, but it's also that the polls gave the NDP in Quebec standing and legitimacy as an option that they didn't have before.

But even when voters are confronted with polls, that doesn't mean they'll believe them. In an election you probably remember very well, Michelle, all the votes in the 2015 Alberta campaign showed the NDP was on track to form a government. Those same polls, talking to the same people, found that Albertans believed the PCs would win. In part, that's because the media wasn't taking those polls seriously and had a role in setting the narrative. I remember some pundits speculating the day before the election, when the NDP was ahead by 20 points, that the PCs would, at worst, be reduced to a minority. Instead, they finished in third place!

So, we agree that polls have an influence. I'd reframe it, though. I suggest that it's knowledge of public opinion that has an impact. Arguing for banning the publication of polls is like saying that voters should be kept in the dark on some things for their own good.

2.) Is a ban on publishing polls before election days the right answer?

MRG: No.

Logistically, what the Ontario Election Commissioner is proposing would be virtually unenforceable. This new policy would be much the same as the ban on the early reporting of federal election results that was lifted in 2014 (by then Minister of Democratic Reform and now leader of the federal Conservatives, Pierre Poilievre). That ban, in place for decades, prevented broadcasters from publishing the results in eastern Canada before the polls closed in other parts of the country.

With the advent of social media, keeping a lid on election results was virtually impossible. Poilievre did the right thing by overturning it.

EG: Enforcing it is one problem (whether it would stand up to a Supreme Court challenge, as a previous ban on polls couldn't, is another), but there's also the problem of what happens when pollsters in other jurisdictions publish polls.

France has a ban on polls published in the late stages of its campaigns, and what happens is that media outlets and pollsters in Belgium and Switzerland publish numbers instead. Would a pollster based in, say, BC be prohibited from posting numbers about an Ontario campaign? What about a partisan pollster in the United States? They could publish polls to try to influence the result, and we'd have no local pollsters (who have more reputational risk if their polls prove to be wrong) putting out surveys to help validate or refute those numbers. So it would be a mess.

MRG: That's a great point, and it leads into my second - and more important - reason why a polling publication ban should be rejected. It infantilizes the voter by assuming they aren't critical thinkers that can't discern whether or not the methodology behind a poll has been slanted to take it in a partisan direction. It also assumes that political commentators or politicians won’t communicate with the public on this issue.

A ban also puts the government in a position where they have the authority to say what information an individual can or can't see during an election. On principle alone, the government shouldn't be given more of these types of powers. And as you said, voters will get this information from somewhere. So I'd rather that they have multiple sources of information to consider, as opposed to, let's say, one Twitter poll run by an anonymous account.

EG: Two key points there — one, that we shouldn’t be deciding what priorities voters should have when making up their minds. Want to vote strategically? That’s your choice, and in our electoral system, it is rational. Polls can help guide you to make a strategic decision, and depriving voters of that information isn’t helpful.

But the biggest issue with a proposal like banning polls is what would fill that vacuum instead. If people dislike poll-focused horse race journalism, they’ll really hate the gut-based, rumour-fuelled, confidential sources-informed horse race journalism that will take its place. Worse will be the leaked polls that would filter their way out into social media and mainstream media, which can’t be verified or refuted by surveys typically published by reputable pollsters and media outlets. Voters will be misled and confused.

3.) If not a ban on publishing polling, what's the answer?

EG: Look, the way polling is conducted and reported on during campaigns isn’t perfect. Journalists should take more care to put polls into context and use them to help inform their reporting, rather than be the sole focus of their coverage. Recognition that polls come with a margin of error (both a statistical one and a real-world one) probably needs to be emphasized a lot more. Pollsters themselves need to make sure they aren’t sensationalizing their results to get attention. Those who approach their job as researchers first, pundits second, are on the right track.

But the premise of Elections Ontario’s proposal is not centred around the quality of polling or the reporting on polls, but instead that knowledge of the outcome of the 2022 provincial election discouraged turnout, and this proposed ban is supposed to be one solution. 

So, I have a question for you, Michelle.

In your riding and in neighbouring ridings in Calgary, turnout in the last federal election was in the 60s — on par with the national average. But a lot of those races were hardly competitive at all. So, what motivated voters in your riding and ridings like yours to go out and cast a ballot when most probably knew that the result would be a victory for the Conservative candidate?

MRG: No candidate should take any vote for granted, and strong campaigns matter. Turnout is usually higher when both the central campaign and local candidates put effort into directly engaging with people and convincing their supporters to get out and vote.

Additionally, many voters in my riding dislike the current federal government, to put it mildly. So past high turnout in my riding has indicated a strong campaign, but also a desire to see change at the macro level. 

The 2022 Ontario provincial election might prove this point but from a different angle. Anecdotally, I know many people were not historically Progressive Conservative voters but didn't see viable options in the provincial NDP or the Liberals, and were satisfied enough with Ontario Premier Doug Ford that they chose not to cast a ballot against the government. Arguably, the results also suggest the PC’s ran a much stronger campaign than the NDP or the Liberals did. The bottom line of the 2022 Ontario provincial election was that for voters, it was not a change election.

And this is the point that was missed in the report. To me, the 2022 provincial election result proves that Ford managed to build a sustainable voter coalition that reaches beyond people who have historically voted Progressive Conservative. The result also shows he governed in a way that established a relatively enduring level of trust with those same people - who also delivered him his first provincial election in 2018. At the same time, the same election result also suggests that voters weren't enamored with the provincial NDP and Liberals enough to get out and support them.

EG: I couldn’t agree more. The main reason I believe there was low turnout in the last Ontario election was, well, that it was a dull campaign with unexciting options. It’s up to parties and politicians to get people enthusiastic about voting for them. And, as someone who has lived in Ontario for most of his life, I can tell you that provincial politics just doesn’t rate highly enough in people’s priorities. Federal, maybe even municipal, can grab people’s attention, but to get Ontarians excited about casting a ballot in a provincial campaign is a challenge. It’s not the fault of the polls. Ontario’s election was not the first one in history where the polls forecast the outcome weeks in advance — most elections are pretty predictable. But most elections don’t have sub-50% turnout.

MRG: Now, these types of dynamics constantly change, which is why we have regular elections. But instead of banning the publication of polling results, leaning into other options to encourage democratic participation is probably the smarter route to take. The report contained several other recommendations which could be studied, but at the end of the day, deciding not to vote is a choice in and of itself. 

For me, that's the real question that every agent in the political system should ask themselves - why would a voter stay at home? Are they satisfied enough with the current government that they feel that they don't need to participate? Do they feel like their vote doesn't matter? Do they know when and where to vote? Do they have the resources to get out and vote (e.g., transportation, childcare, etc.)? Once the reasons behind those questions are answered, we can devise ways to overcome those barriers - which many campaigns and electoral agencies are already working on. 

Bottom line - I believe voters need more information, not less, and a ban on the publication of polling results wouldn't make our democracy any healthier.

*******************

Post-script….

MRG: I often read Eric’s Substack, The Writ, to get analysis of Canadian poll results. Today he has a thoroughly interesting read on the upcoming PEI election, which you can read here.