On my radar this week

Alain Giguère

CROP in the news

Our public studies

Our contents

Our Blog

Welcome to our blog, a creative space for free thinking, ideas and inspiration!

The pollsters’ mea-culpa?

Categories: CROP in the news

Posted on 10-02-18 at 2:53 p.m.

Many will no doubt tell us that we, the pollsters, got it wrong, and we will have to accept their verdict.

In our defence, we can always cite the particularly low voter turnout (66% versus 71% in 2014), the fact that Liberal supporters stayed home, and so forth ... which wouldn't be entirely untrue.

But could we really have predicted such an unpredictable about-face? Polls are snapshots taken at a specific time and especially in a specific context. You have to be wary of their predictive value. A few days later, by the time voters arrive at the ballot box, the dynamics can be very different.

We are forced to trust what people tell us. Perhaps they do not always reveal their true feelings.

Former Quebec premier Robert Bourassa coined a now historic expression, "la prime à l'urne" (the ballot-box effect), arguing that Liberal voters were more discreet with polling firms; that they were reluctant to reveal their true voting intentions.

When Bourassa coined this expression, it was "cool" to vote for the Parti Québécois - a young, liberating, trendy, Montreal (but not elitist) party. In contrast to this image of the PQ, the Liberal Party was perceived as more "conservative." People were somewhat embarrassed to admit in a poll that they were voting Liberal. I am not saying that people outright lie to us, but some tell us that they are undecided while others act differently in the privacy of the voting booth.

In such a context, political analysts in Quebec have always criticized us for underestimating the Liberal support. Even until recently, they were advising us to allocate 50% of our undecideds to the Liberals in our distribution process in order to account for this anticipated boost at the ballot box. Imagine where we would be if we had done that for this last campaign!

However, if the fact that a party is perceived as "conservative" makes voters slightly embarrassed to admit that they will vote for it and pollsters consequently underestimate the support for that party, the CAQ may well have borne the brunt of this trend in this campaign.

In my last text for L'actualité and my blog, I pointed out that CAQ supporters clearly display a certain degree of ethnic intolerance. This party has forged an image of ethnic intolerance (remember the burkini ban proposed by Nathalie Roy, who was re-elected last night). One shouldn't forget that the campaign focused largely on immigration and that the CAQ has certainly appeared intolerant on this subject.

This party is undoubtedly perceived as a right wing, conservative party. In Quebec, such conservatism may be circumspect yet freely expressed in the privacy of the voting booth.

In hindsight, we can now conclude that the about-face at the ballot box, the so-called prime à l'urne, led pollsters to underestimate the CAQ support and overestimate the Liberal support, while properly estimating support for the two other parties.

The polling industry will assuredly be pondering these results but, at this point, it is not entirely clear how to proceed.

The solution for properly allocating "discreet" and undecided voters in this new context is not obvious. To have correctly predicted last night's election results would have required manipulations that are difficult to imagine!

CROP has invested in artificial intelligence to better predict consumer behaviour. But we do so by merging transactional data with attitudinal and declarative data. In the case of an election poll, however, our only data source is what people tell us, and we are forced to rely solely on that.

This election has changed things. We will have to be very creative in the coming years to get around what appears to be a new boost at the ballot box!

Alain Giguère
President, CROP Inc.