Official Languages Committee on March 12th, 2020

[Anil Arora]

We respect the process.

The only thing I can tell you is that we're ready. If this is the decision, it will be the first time these five questions will be in the census questionnaire. They will increase the quality and accuracy of the data as never before.

That's not the only thing Statistics Canada will do. We are prepared to work with our partners to develop a postcensal survey.

In addition, we are increasing our efforts to focus on administrative data. We are targeting a data ecosystem, as the need will exist after the 2021 census. We are therefore building a system that will serve users long after the 2021 census.

[Chris d'Entremont, Conservative, West Nova, NS]

Since there are two census questionnaires, a long one that goes to 25% of the population and a short one for everything else. So it is the government that will tell us whether the questions will be in the short questionnaire.

[Anil Arora]

Forgive me, but I didn't understand your question.

[Chris d'Entremont, Conservative, West Nova, NS]

Will the questions be included in the short questionnaire?

[Anil Arora]

In Canada, we have a census that consists of two questionnaires: a short one and a long-form questionnaire. It is mandatory to complete one or the other of the two questionnaires. That is the very definition of a census. Asking a sample of households to complete a long-form questionnaire is a feature of our census.

I'd like to stress again the importance of data quality. You make important decisions on a variety of issues for Canadians and it is essential that we provide you with the data you need to make those decisions. So we're concerned about providing high-quality data.

[The Chair, Liberal, Emmanuel Dubourg]

Thank you, Mr. Anora.

The next five-minute period will be shared between two members of the committee. We will begin with Ms. Lambropoulos.

[Emmanuella Lambropoulos, Liberal, Saint-Laurent, QC]

Thank you, Mr. Arora, for being with us today.

I have a couple of questions. I'm an anglophone from Quebec. I noticed the questions that are in your survey, that are in the census so far—only questions 16 and 17: did this person do any of their primary or secondary school in an English language school in Canada? Obviously, this excludes the grandparent clause. Obviously, in Quebec we're allowed to send our kids to an English school even if the parent didn't go to an English school but the grandparent did. Is there any way of including some type of question to ensure that all rights holders...and to ensure that we get an actual number of eligible students who would be able to go to the school?

[Anil Arora]

That's our intent with the questions that we've tested. It is to make sure that we get the quality of information and the numbers that we need for very small regions, to make sure that we have a good sense of the number of rights holders. Obviously, that's our intent. That's what that module was built, designed and tested to do. It is to ensure that we have that.

The second part, as I keep saying, isn't just what that total eligible population is. It's also about being able to go further and ask how many would actually take advantage of that. It's not one or the other. You need both.

We also need to find out how many people are actually going, where they are going and what the distances are.

It's those three things that we want to try to work at. This will be the first time in the census—once a decision is made to include these questions on the census—that this kind of quality and detail of information will be there for all users. Our aim is to continue to work with stakeholders and with you to make sure that information is kept up to speed and that we continue to strengthen it and add to it over time.

[Emmanuella Lambropoulos, Liberal, Saint-Laurent, QC]

Would these types of questions be included in both the short and the long-form survey?

[Anil Arora]

I'm speaking in general here because that's the heart of this debate. The question that's being asked is: Can we give good quality information at low levels of geography from the one in four sample part of the survey? There are so many programs already in Canada that use that information to make important decisions. It isn't just a homogenous population; it's heterogeneous or homogeneous. We do that. That's what we do all the time. That's what the science of the statistical rigour is. We want to make sure that we provide that detailed information to everybody.

The last point I'll make is that if you think that's theory.... I was just asking colleagues how many times people went those local levels. We had over 12 million visits to our census profiles that provide information at that very local level. We know we're providing it, we know it's of high quality and we know Canadians are consuming it.

[Emmanuella Lambropoulos, Liberal, Saint-Laurent, QC]

When provinces counter that and say they don't have an exact number because they're not following the 25% of people who have been surveyed, they have no recourse afterward. I think that's the main problem here. That's why we're all fighting for the short-form census. This would give stakeholders—people—the ability to go and fight the province and say that this is the exact number of people who are eligible.

[Anil Arora]

First of all, there are many programs where there's a need for very high-quality data. The unemployment program is one example I've given that we benchmarked from the census to our ongoing surveys. There are a lot of programs, such as employment equity. There are many legal requirements that are based on the long form.

I'm trying to tell you that we do not want to inadvertently leave Canadians with the impression that the information that comes from the long form is untrustworthy, is not of high quality or is not something you can trust at low levels of geography. I'm trying to explain that this is something we have done for 50 years in Canada on a whole host of areas. That information is not theoretical. It's something that we provided to you as legislators and as decision-makers. You make important decisions already on a whole host of laws based on that quality of information.

I'll leave you with this last point. If, in fact, that information was not reliable, we would see huge variances in the results from one census to the other for those local levels, but we don't. We're not the only country that uses a sample to do the census and to provide high-quality data at low levels. It is something that the world over.... It's a sampling technique that provides very high quality information. There are 3.7 million Canadians who fill out the long form and the quality of the information is very high.

That's all I'm trying to say.

[Emmanuella Lambropoulos, Liberal, Saint-Laurent, QC]

Thank you.

[The Chair, Liberal, Emmanuel Dubourg]

Thank you very much. Your time is up.

Mr. Beaulieu, you have the floor for two and a half minutes.

[Mario Beaulieu, Bloc, La Pointe-de-l'Île, QC]

Good.

I mentioned I'd be asking questions other than about the rights holders. Maybe I'll come back to that later.

There's a question I wanted to ask you. Before 2005, there were multiple answers to the question of the language spoken at home. If people answered that they spoke both English and French, half of the answers were sent to the English side and half to the French side. Those who answered that they spoke English, French and another language were divided into three categories.

In 2016, you analyzed the results in such a way that someone responding by "English and French" was counted on both the English side, the French side and the allophone side. Therefore, compared to 2011, in Quebec, when there were 81.2% of people speaking French at home, the figure rose to 87% of people speaking French at home. These are Statistics Canada's results. If we look at the results for anglophones in Quebec, there was a jump to 19%. The results for allophones were even higher. This gave a total of approximately 121%.

Why did you make this change in your method of analysis?

[Anil Arora]

I think I understand the question.

The possibility of multiple responses is not unique to this question. There are several other questions like that. There may be situations where there is more than one answer. The question is how we deal with this and how it relates to the situation of the rights holders.

As soon as someone answers that they speak French, we use that as data. It's counted. So we use an order to determine whether the answer will be counted or not. When there are several answers to a question, if one of the answers is French, that counts as one of the characteristics.

[The Chair, Liberal, Emmanuel Dubourg]

Thanks, your time is up.

Mr. Duvall now has the floor.

You have two and a half minutes.

[Scott Duvall, NDP, Hamilton Mountain, ON]

Thank you, Mr. Chair.

Mr. Arora, I guess everyone is trying to ask if the same questions are going to be put on the short form and the long form. You're talking about high-quality data. How do you get high-quality data if you're missing 25% of answers?

[Anil Arora]

I'll just repeat what I said before. I'm very concerned if people think that any of the data—through that logic that you've put—that comes from the long-form census is missing 75% of the responses or that it is not of good quality, because you as legislators make decisions on a whole host of programs based on the quality of the data that we get from the long form.

I cannot explain it strongly enough. We know the total population, and it is a systematic sampling technique that is used the world over where, in our case, every fourth household, with a random start and with a mandatory requirement, fills out a whole host of questions that are on the long form. When we get those responses, we have weighting techniques to make sure that the one house that says “yes” or “no” represents all four in that region. That is how we come up with the full population.

So, it's not that 75% are missed. It's that if you get selected, in a very statistically sound manner, to fill out the questionnaire, you are representing that whole area. That is a method by which we can ask questions of one person, and that response essentially represents others.

I don't want to make light of this, but we don't drain the entire transmission oil to find out whether it's good or not. We take a sample of it and say, “Yes, it's of good quality”, and then either it's time to change it or it's not.

The sampling technique is something that is germane to statistics. Asking a question, with that response essentially serving as a donor to the rest of the population, is what we do all the time.

[The Chair, Liberal, Emmanuel Dubourg]

Thank you.

Colleagues, we are moving on to the third round of questions, and you will have five minutes.

We will begin with Mr. Généreux.

[Bernard Généreux, Conservative, Montmagny—L'Islet—Kamouraska—Rivière-du-Loup, QC]

Mr. Arora, what you just said is true. However, what you fail to say is that the way in which the sampling is working is good for you, but it's not good for the Supreme Court of Canada and the rights holders. That's the reality. Unfortunately, the Fédération nationale des conseils scolaires francophones, or FNCSF, which testified before the committee two days ago, as well as all the other witnesses who were present with them, all say the same thing. That is that all official language minority communities in Canada and rights holders cannot get their rights recognized, because they do not have the right figures to be able to judge the situation. When they go to the Supreme Court to defend their rights, they are always told that they do not have enough statistics.

We ask you to take note of the necessary statistics. I think and we think—tell us otherwise if it is not true—that the questions should be in both the long and short forms. You can't have them in one or the other. We think it is necessary to have them in both forms. Tell us why we are wrong to think that.

[Anil Arora]

Users have expressed their requests, the courts have made their decisions, and three elements are required. First, what is the entire population in a given region? Second, how many people will prefer or use a service? Third, what will be the changes in the population?

The judges, in their rulings, were very clear. These three elements are necessary in order to respond to user requests. We will provide these three elements by working with the public. As I said, we are dedicated to providing high-quality data on the total population, also from a postcensal survey, to the people who will use the services and the administrative data. So these three elements are necessary.

[Bernard Généreux, Conservative, Montmagny—L'Islet—Kamouraska—Rivière-du-Loup, QC]

All right. You're saying the Supreme Court says we need all three. Why, in previous censuses, was it not possible to provide this information so that the Supreme Court could make decisions in favour of minority schools or official language minority communities? This is the first question.

Next, as part of your process and analysis, what recommendations will you make over the next few weeks and months to Minister Navdeep Bains for inclusion in the long and short census forms? What will your recommendations be?

[Anil Arora]

We will present recommendations that will provide high-quality data that will meet the expressed needs. That is our goal.

I don't understand why we would provide a recommendation that would not have that objective. Our goal is always to provide high-quality data.

[Bernard Généreux, Conservative, Montmagny—L'Islet—Kamouraska—Rivière-du-Loup, QC]

I'll repeat what I asked before.

Will the final decision to include the question or questions in the long form or short form rest with Minister Navdeep Bains?