Go to the Globe and Mail homepage

Jump to main navigationJump to main content

Statscan's former chief statistician Munir Sheikh is pictured at his office in Ottawa on Tuesday July 29, 2008. (Sean Kilpatrick/Sean Kilpatrick/The Globe and Mail)
Statscan's former chief statistician Munir Sheikh is pictured at his office in Ottawa on Tuesday July 29, 2008. (Sean Kilpatrick/Sean Kilpatrick/The Globe and Mail)


As every statistician knows, 50 plus 1 is not always clear Add to ...

What is common among the recent NDP bill, the Clarity Act and the recent Globe articles on “clarity” in the context of Quebec separation from Canada? They are all unclear when they talk about clarity.

There are two parts to clarity, as all participants in the debate seem to agree by now: the “clarity” of the question to be asked; and the “clarity” regarding a majority voting for separation. While the first part seems to be more or less clear, there is no clarity among these views regarding a “clear majority”.

What on earth is clear majority?

In the debate, there seem to be two answers: the NDP answer which basically says that they are singularly going to focus on the issue of the clarity of the question and, on the issue of a clear majority, they are going to ignore the “clear” part and concentrate simply on the majority. This is not good enough because clear majority, while its meaning may be unclear, simply is not the same thing as the majority.

The answer from the Clarity Act is: we understand that “clear majority” is different from a “majority” but we have no idea what the distinction is. We will leave it to future Parliamentarians to sort it out. This is also not good enough.

The determination of a “clear majority” is largely a statistical issue and hence we should seek help from statisticians. This would be a clear example of evidence-based decision-making on an issue which, at least to me, is the most important issue facing Canada.

As an example, assume there are 100 voters and that each and every one eligible to vote does indeed vote on a clear question. Of these, 51 vote yes and 49 no. I would argue that this is a “clear majority”. It is a majority and it is as clear as it could be.

If that is accepted, one way around this unclear debate is to make participation in a separation referendum mandatory and use the simple 50 plus one rule.

However, it is not that simple because, as statisticians know, even with a mandatory rule, not everyone can vote. There are sick people, those who are travelling, some who cannot be located easily, etc., etc. Even with mandatory voting, it is conceivable that a 50-plus-one formula may contradict the unknown majority view.

It gets more complicated if voting is not mandatory. Suppose only 60 of the 100 vote and 31 vote yes. Is this a majority? For election purposes yes but for breaking up a country? It is indeed a majority – even in these circumstances – but a “clear majority” it is not.

I would offer the following simple definition of clear majority: 50 plus 1 of all those who are eligible to vote agreeing with a question posed to them.

While the definition is simple how do you make it operational? This is where a statistician would be helpful. Here is what I would propose.

First, make participation in the referendum mandatory. Surely, if we are thinking of breaking up the country, the province wanting to break up should at least listen to the view of all its current voters. Some may argue future voters should matter as well but I have no idea how to capture that.

Some would argue as well that there is no need for mandatory voting as those who do not participate are expressing their view that they are indifferent to the outcome. That is not always true, as statisticians know. Survey data show that those who are championing a cause are more likely to vote than those who are not, thereby biasing the results.

Second, we need to account for the views of the missing voters in a mandatory referendum. Knowing the experience with the census data collection, this could be about 5 per cent of the voters, a large number if the target for success is 50 plus 1. This would be a more complex and more expansive exercise than counting people in a mandatory census: this is because the census is on a household basis while the referendum count must be completed on an individual basis.

In counting the missing voters, we could follow up those who have not voted. We need to keep in mind they would know they have power as their vote could be crucial which would lead politicians to concentrate on them, thereby encouraging even those who want to vote to hold off. This could be corrected with hefty penalties for those who do not vote in the first place.

Even this may not capture all the voters. We would need to determine the cause for this and make appropriate statistical adjustments.

This sounds pretty complicated. It is. But it is a necessary evil to get a view regarding a “clear majority” of 50 plus 1 of those who are eligible to vote.

Fortunately, there is a simpler way possible: get a statistician to estimate the margin of error caused by those who may not vote in estimating the yes vote and add it to the 50 threshold. This means that a “clear majority” of 50 plus 1 would in practice mean 50 plus X (for the margin of error to be estimated) plus 1.

Munir Sheikh is a former chief statistician of Canada. He is a distinguished fellow at the School of Public Policy, Queens University.

Report Typo/Error

Follow us on Twitter: @GlobeDebate

Next story




Most popular videos »

More from The Globe and Mail

Most popular