Official Languages Committee on March 21st, 2018

[Jean-Pierre Corbeil]

Absolutely. We have Stephen Thompson, the director of strategic policy, research, and public affairs for the Quebec Community Newspapers Association. We also have Eva Ludvig, an education expert, and Jack Jedwab, who is also from Quebec.

[Dan Vandal, Liberal, Saint Boniface—Saint Vital, MB]

Mr. Chair, how much time do I have left?

[The Chair, Liberal, Denis Paradis]

You had six minutes, but now you have just five left.

[Dan Vandal, Liberal, Saint Boniface—Saint Vital, MB]

From your answer to Mr. Samson's questions, I conclude that the greatest risk in this whole process pertains to the placement of the questions, which could be a problem.

Can you elaborate on this risk and on how we can address it?

[Jean-Pierre Corbeil]

As I said before, you have to understand that where the question module is placed in the questionnaire often has an impact on the answers given to subsequent questions. For instance, information about education is followed by information about the workplace and various aspects of employment.

What we have to verify is the extent to which the place of the questions might destroy comparability with the information we have gathered over the years with regard to education and employment. So we have to make sure that any new questions that are introduced do not destroy comparability, because that can happen. When questions are added, there is a risk that people will switch off and stop answering subsequent questions in the other modules.

When we conduct qualitative and quantitative testing, we can play with the placement of those questions to ensure consistency and that it is logical and can be understood by respondents.

[Dan Vandal, Liberal, Saint Boniface—Saint Vital, MB]

Would you like to continue, Mr. Samson?

[Darrell Samson, Liberal, Sackville—Preston—Chezzetcook, NS]

Yes, I have a few more questions, thank you.

I asked you earlier whether the language-related questions in the short form will be included in the long form. You said they will. Personally, I would rather see them in the short form so that 100% of people answer them. If the questions are added as planned, will they be in the short form? That would guarantee that they will also be in the long form.

[Jean-Pierre Corbeil]

I hear your question loud and clear. My answer is that we cannot do things purely according to preferences. We have to have scientific certainty that we will obtain the strongest data possible. That is why we will use the best methods possible.

As I told Mr. Choquette earlier, we will be able to demonstrate the most effective approach in order to enumerate this population as accurately as possible.

[Darrell Samson, Liberal, Sackville—Preston—Chezzetcook, NS]

I am not a scientist, but I see some problems with that.

Once you determine which questions will give you the information you need, whether you put them in the short form or the long form should not make any difference in theory. If you put them in the short form, that will guarantee that 100% of the population will answer them, because they will also be in the long form.

In my opinion, these questions should be in the short form. I would like to hear you say that you intend to put them in the short form. I don't see what difference it would make.

[Jean-Pierre Corbeil]

I cannot tell you what I intend to do because it must be based on a robust analysis.

[Darrell Samson, Liberal, Sackville—Preston—Chezzetcook, NS]

You will have to do that analysis and delve deeper into the matter. How will you determine whether to put the questions in the long form or the short form? I have not asked anyone about this, but if I asked everyone around the table, I think everyone would want them to be in the short form. That way they would be in both questionnaires and the response rate would be 100%.

[Jean-Pierre Corbeil]

We are working on it.

[Darrell Samson, Liberal, Sackville—Preston—Chezzetcook, NS]

Can you give us that information?

[Jean-Pierre Corbeil]

I cannot tell you right now. All I can say with regard to our choice is that we will not simply inform you that we have made a decision. We will provide proof to demonstrate which approach is the best.

[Darrell Samson, Liberal, Sackville—Preston—Chezzetcook, NS]

Okay, but I cannot imagine that it would not be preferable to include it in the short form, since the information would then be in both.

In conclusion, I have an important question.

Regarding the committee that has met just once so far, you said that you had considered the questions and analyzed and discussed them. People might have initially been in favour of using the short form or the long form, but I would like to know the group's position at the end of the meeting. I would also like to hear your opinion of the process.

[Jean-Pierre Corbeil]

Most of the people in attendance could appreciate the complexity of the issue. At the end of the meeting, we reached a consensus on the approach we will use in our qualitative and quantitative testing. When we provide you with the questions—they have in fact been amended since our meeting on January 25—we will provide them to the members of the advisory committee at the same time. We will then have another meeting to discuss them.

[Darrell Samson, Liberal, Sackville—Preston—Chezzetcook, NS]

Thank you.

Mr. Arseneault, you may take over.

[René Arseneault, Liberal, Madawaska—Restigouche, NB]

Mr. Corbeil, I am not a great scientist much less a statistician, but I am wondering how can you convince me that 100% of a population is not a better sample than 25% of the same population. Please explain that to me.

[Jean-Pierre Corbeil]

In five minutes?

[René Arseneault, Liberal, Madawaska—Restigouche, NB]

No, I think one minute would be enough. Just tell me that it is impossible that a 25% sample of the population is more credible than a 100% sample. It is impossible.

[Darrell Samson, Liberal, Sackville—Preston—Chezzetcook, NS]

It is impossible. There might be a 4% or 5% difference. In politics, there is always a margin of error in surveys.

[Jean-Pierre Corbeil]

Statistics are based on that. At Statistics Canada, we conduct extremely high quality surveys with extremely large samples, but none is as large as the long-form census. With the long form, we ask questions to 7 million people or more.

In 2021, if we ask those questions to 10,000 people, will we get less accurate data?

Even for small geographic areas, we do not think that will be the case. We have in fact demonstrated that we now have extremely accurate data. Since we address a fairly small population and ask a certain number of questions, as in this case, we want to be sure we choose the most reliable and effective methodology. That is what has to be clearly identified. That is all I can say on the matter.

[René Arseneault, Liberal, Madawaska—Restigouche, NB]

Okay, but in the case we are discussing, the enumeration of language minorities, we are talking about a very small population outside Quebec and a very small population inside Quebec. These are not the types of questions that apply to all Canadians, such as whether they are in a relationship, whether they are married or whether they have children, and how many.

You are saying that the short questionnaire will reach 75% of the targeted people and that the long questionnaire will only reach 25% of them. If I understand correctly, if those questions are included in the short questionnaire, they will automatically appear in the long questionnaire. However, 100% of the questionnaires affect Canada as a whole. We are not trying to gather Canada-wide information, but rather to enumerate minorities. They must be sought out and reached.

How can you convince me that, scientifically speaking, a 25% sample size can be better than a sample involving 100% of the people we want to survey?

[Jean-Pierre Corbeil]

All I am saying is that Canada's census is intended for Canadian society as a whole....

[René Arseneault, Liberal, Madawaska—Restigouche, NB]

That is what I am saying to you, as well, but the situation is different when it comes to enumeration. I made that distinction by giving you the example of enumeration. We are not talking about issues that affect all Canadians. We want to enumerate linguistic minorities.