Industry Committee on March 8th, 2011

[Robert Bouchard, Bloc, Chicoutimi—Le Fjord, QC]

You say that we cannot expect miracles. So, according to you, will the census be of a lower quality than the 2006 census? If we were to compare both, the one in 2006 and the one in 2011, do you expect the quality of the 2001 census to be lower or equal? I would like to know your thinking on this.

[Dr. Ivan Fellegi]

I think everybody agrees that it will be inferior. That was the view of my immediate successor, Munir Sheikh, and Mr. Smith just also said it won't result in comparable data. It won't result in the same quality. So we all agree that it won't.

In my mind, the main problem is not that it's going to be useless for all purposes. I think Mr. Smith is entirely right that it will be usable for a whole variety of purposes, but we won't know which ones it should be used for and which ones it shouldn't, at least not consistently, because bias is intrinsically unknowable. We can get some indications by comparing to other data and so on, but intrinsically we won't even know the response rates of the various different groups affected.

We will have an overall response rate, but we won't know what proportion of aboriginals, what proportion of low-income people, what proportion of visible minorities, and what proportion of recent immigrants did or did not respond, let alone how well the ones who responded represent the ones who didn't respond.

So there is going to be, one, a deterioration and, two, a great deal of uncertainty. While the data will be usable for a whole variety of purposes, we won't be certain for which ones they are usable and for which ones they are not.

[Robert Bouchard, Bloc, Chicoutimi—Le Fjord, QC]

Mr. McKinnon, I hope I will quote you correctly. You said that, with the new methodology, we will lose our benchmarks. If Bill C-568 is passed, do you think we will have better benchmarks? Will we be able to collect data providing us with better benchmarks than what would be collected with the 2011 census, because of the changes that have been brought about?

[Ian McKinnon]

Again, with the caveat that I agree entirely with the other witnesses that the 2011 census is too far advanced operationally to change now, if this bill were to become law, then we would have a situation whereby the capacity to have benchmarks across a wide array of variables would be restored to what we have known over the past 100 years with high-quality census data, yes.

[Robert Bouchard, Bloc, Chicoutimi—Le Fjord, QC]

I will put my question to Mr. Fellegi and Mr. McKinnon.

Let me go straight to the point: can you tell us briefly why, according to you, the chief statistician, Mr. Sheikh, resigned? You probably know him. Can you tell us the main reason why he resigned?

[The Chair, Conservative, David Sweet]

Mr. Bouchard, we're quite a way over your time, so if you want to hold that question and have it answered in another round or something, you can do that.

But now we'll go on to Mr. Wallace for seven minutes.

[Mike Wallace, Conservative, Burlington, ON]

Thank you, Mr. Chair.

I thank our guests for coming today to discuss this issue.

Mr. Smith, I want to follow up a little on your opening statements about how it's going in the north. You've been out in the field, I guess you'd call it, for the last number of weeks. Did you say you have about an 85% return rate thus far?

[Wayne Smith]

In the north, we fly people in. Essentially, we go door to door; we actually do the enumeration. The interviewer does the interviewing. Also, instead of doing a sample of one in three, we actually go to every household.

[Mike Wallace, Conservative, Burlington, ON]

You do every household?

[Wayne Smith]

So ultimately in the north we'll enumerate about 22,000 households in remote areas. So far, we're getting cooperation from 85% of the households.

[Mike Wallace, Conservative, Burlington, ON]

Based on that methodology before, in the last census, what kind of return rate did you get?

[Wayne Smith]

I don't know it offhand, but it would have been slightly higher than that.

[Mike Wallace, Conservative, Burlington, ON]

Slightly higher than that, so just for my education—I don't know if I'm right or not—if the item from StatsCan carries penalties, that helps define what a census is, because there are penalties for not doing it. Anything else is a survey, because there are no penalties for not doing it. Is that basically the difference?

[Wayne Smith]

No. The census necessarily is mandatory under the law. Under the Statistics Act, any other survey can be declared voluntary. It's mandatory by default; it can be declared voluntary.

[Mike Wallace, Conservative, Burlington, ON]

Something that is mandatory normally carries penalties with it, though, under the law. Is that correct?

[Wayne Smith]

That's true.

[Mike Wallace, Conservative, Burlington, ON]

Our removing the penalties, which is basically what we've done, forces it to be voluntary, then, in a sense.

[Wayne Smith]

The census of population itself remains mandatory, and people who refuse to participate would be subject to penalties. The national household survey has now been declared voluntary under the Statistics Act, so there would be no penalties for refusal.

[Mike Wallace, Conservative, Burlington, ON]

Okay. I appreciate that.

Now, what is StatsCanada doing to promote the filling out of both the census and the national household survey? Are you doing anything different in terms of promotion in terms of the national survey?

[Wayne Smith]

We're doing what we always do. The most important thing, as I mentioned in my opening remarks, is broad-based support. A few weeks ago, we began a national outreach campaign. We're going out to municipalities, businesses, and various organizations, and to the media as well, asking for their support in the promotion of the census. Our experience suggests that it has much more impact when another person says “this is important” than when Statistics Canada says it.

So far in our campaign, we're getting very good support across the country. We've had a provincial government make a public commitment to spend $500,000 on advertising. We've had various kinds of commitments from various organizations across the country. It's going quite well.

[Mike Wallace, Conservative, Burlington, ON]

I think you said to a previous speaker that a 60% return rate would be a wash. But for argument's sake, let's say that the rest of Canada follows what the north is doing and that we have an 85% return rate. That would be more data for StatsCanada than you would have had in the past, would it not, because we've increased the number of surveys that we put out there?

[Wayne Smith]

The responding sample would increase, which would mean that the sampling error would actually be less in the national household survey than we saw in the 2006 census. It wouldn't remove the risk of non-response bias—

[Mike Wallace, Conservative, Burlington, ON]

Right.

[Wayne Smith]

—but it would certainly reduce it. The general view is that the higher the response rate, the less the risk of non-response bias. But I just want to repeat that the non-response bias is a risk, not a certainty.

[Mike Wallace, Conservative, Burlington, ON]

Right.

Has StatsCan been able to do anything...are there any processes or anything to mitigate the non-response bias that may or may not occur?