The Destructive Power Of Polls

All over social media, I see people posting about polls and extrapolating election results from them. Worse, they look at polling aggregators and believe the riding-level data expresses some form of reality. To me, published polls are an affront to the very concept of democracy, leading to the intellectual laziness of groupthink. Momentum, trends, horse races, bollocks! Only one poll matters: the one in which you place your ballot.

We each have to make our own analysis and decisions, not lean on the perceived popularity of our candidates. There is no value to knowing who everyone else thinks is going to win, and there is negative value in using this information to make your own decision. Being an informed voter does not mean relying on the meta-analysis of the opinions of your neighbours, it means understanding the major issues and assessing who has the best plan and ability to deal with them.

The trouble is that people crave the information polls provide. They want to know what their neighbours and compatriots think. If credible polls aren’t published, false narratives about momentum and support will be circulated in their place that aren’t based on any statistical or empirical data. Without polls, how will we know what Canadians, as a whole, are worried about?

But then, why does it matter? What are you worried about? What issue is driving your decision on how to vote?

Polling on its own is not the issue. Polls are essential to the health of the democracy. The government regularly commissions polls to get the mood and priorities of the country. Businesses use them to test messaging and popularity for their products and brand recognition. Political campaigns use them to assess their own successes and failures. All of those are internal.

In this election, public polls are telling us of a dramatic shift in support. We’re hearing projections of Liberals hitting absolute-majority territory and Pierre Poilievre’s fall from grace being so spectacular, so thorough, that he may even lose the seat he’s held in the House of Commons for over twenty years.

It all creates a narrative that the Conservatives are done and the Liberals will overwhelmingly win the election. It was only three months ago that the narrative was that the Conservatives would win the election and that the Liberals would struggle to hold onto official party status.

It was less than a year ago that Joe Biden suddenly dropped out of the Presidential race and was replaced by Kamala Harris at the last minute, changing that narrative from one of certain defeat to certain victory for the Democrats.

We must remember that at the end of the campaign, it was Trump that won the election. Polls didn’t really see it coming. The public didn’t see it coming. Believing Kamala Harris would win, more Americans stayed home than voted for either candidate.

They had followed the narrative instead of simply voting for the results they wanted to see.

That risk remains in Canada as well. If Canadians are left to believe that the Mark Carney-led Liberals will win so overwhelmingly that the entrenched Conservative leader will lose his own seat, the incentive to vote will not be there. The belief that others have it under control so you can stay home may be strong enough to impact the election, as it did south of the border.

For those wondering about their local ridings, many refer to 338Canada for what to expect, and act on that basis. This is a terrible idea. While the overall aggregated data is the collection of all available polling data and has a high probability of giving roughly accurate national numbers, it has no such local value.

The website explains its own methodology at length. It boils down to the application of national polling data against local history to roughly predict what should happen in the riding given current national or regional polling trends. To understand the impact, take my campaign in 2019. To their credit, 338Canada published their prediction overlayed with the actual results after the fact.

In Laurentides—Labelle, the riding I grew up in and represented, the 2019 forecast put the riding in the safe Bloc category. All over the riding, I heard that the Bloc was winning, and many people referred this information back to me during the campaign, using it as a basis of their own support for the Bloc — in essence: that’s where the riding is going, so I am going there too.

The final results, however, showed that while the Bloc support almost exactly lined up with Canada338’s prediction, none of the other parties had. My own support was far above even the ceiling of support Canada338 had ascribed to me based on national polling data, because no local data was used in their model. The big number was the prediction; the small number was the actual result:

Published polls do not only reflect how voters are thinking, they guide their opinions and their behaviour.

The message is clear:

Vote for the outcome you want to see.

Do not rely on polling data to make your decision.

Do not trust polls to accurately predict election-day results.

Do not wake up on April 29th regretting your vote or having stayed home because you took for granted that your neighbours had this election under control.