Polls offer a snapshot view… sometimes

Opinion

Hey there, time traveller!
This article was published 15/09/2021 (1453 days ago), so information in it may no longer be current.

CANADIAN voters have been polled since the early 1940s. The first published polls on Canadian voting intentions appeared in newspapers across the country in 1942 during the national referendum on wartime conscription.

Having opened its doors in Toronto in 1941 with 27 newspapers signed up as clients, the Canadian Institute for Public Opinion (CIPO) was a northern branch for George Gallup’s American Institute for Public Opinion (AIPO). Gallup’s final poll on that referendum was released two days ahead of the April 27 vote, with a four-point overestimation of the 64 per cent who voted for conscription.

Considering this was the very early days for Canadian polling, it was a good measure of Canadian voting intentions.

Up until the 1970s, the Gallup organization was the only firm to provide federal election polls to the media, with regular releases during each campaign. This soon changed. In the 1984 federal election, which brought Brian Mulroney and his Progressive Conservatives to power in a massive landslide, six firms conducted a total of 10 national polls for the media. This was a tiny amount compared to the 113 national polls that were released in the “Free Trade election” of 1988, in which Mulroney was re-elected.

The number of national polls in each election has since remained high, with the 2019 federal election being no exception. According to Éric Grenier, who provides polling analysis for the CBC, just under 120 national polls were released to the public during the last campaign. But do polls accurately measure voter intentions? By my own analysis of the 12 national polls released in the final five days of the 2019 federal election, on average they were a mere 1.58 points off in predicting the Liberal vote.

What about other Canadian elections? Is polling accuracy the norm? Generally, the answer is yes. In my own analysis of a total of 19 elections, both provincial and federal, from 2015 to the end of 2020, there were 81 province- or nation-wide polls released to the public in the final five days of the vote. On average, the polls were only 3.1 points off in predicting the winning party’s vote.

However, while the polls were right for most of these 19 elections, they did poorly in a few. Among the worst cases were the most recent provincial elections of Quebec (2018) and Alberta (2019). Six polls were released in the final days of the Quebec campaign; on average, they were off by 5.2 points in predicting the vote for François Legault’s victorious CAQ.

In the Alberta election, there were seven polls released in the final few days leading up to Jason Kenney’s United Conservative Party victory; on average, these were off by a whopping 6.8 points.

Most recently, the 2021 Nova Scotia provincial election raised concerns among poll watchers. Only two firms, Forum Research and Mainstreet Research, released polls in the final days of the campaign, with both pegging the Liberals two to three points ahead of the PCs. Forum Research reported in its media release that the “Liberals are expected to form a minority government,” while Mainstreet simply reported its polling numbers rather than making predictions.

To the surprise of many, when the votes were tallied, Tim Houston’s PCs scored an upset victory with a two-point lead over the defeated Liberals.

In 1963, Lester Pearson’s Liberal Party won a minority victory over John Diefenbaker’s PCs. Hankering for a majority, the Liberals wanted more. Their pollster, Oliver Quayle, an American based in New York, advised Pearson in 1965 to call an election. Both the Gallup polls and Quayle’s internal party polling showed the Liberals in the lead.

But polls are only a snapshot in time. To the party’s frustration, the Liberals won only two additional seats in the House of Commons, thereby producing another minority government.

Now that we are in the home stretch of the 2021 campaign, one must wonder at Prime Minister Justin Trudeau’s folly in calling this election. No doubt, Liberal strategists were looking at their own internal polling numbers and those in the media, thinking now was the time to get their majority victory. Indeed, at the campaign’s outset in mid-August, the national polling firm Leger reported the Liberals were holding a five-point lead over the Conservatives.

Like a mirage, the Liberal lead has disappeared. Just over a week ago, Nick Nanos released his four-day rolling poll figures, showing the Conservatives with a small lead over the Liberals. So far in this campaign, we have seen more than 80 national polls released to the media. Many more will come, and the horse race is on.

The tally on Sept. 20 will determine whether the Liberals miscalculated in calling this election. Based on the most recent polls, they will be lucky to remain in power.

Christopher Adams is a past pollster and adjunct professor in political studies at the University of Manitoba.