Industry Committee on March 8th, 2011

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

Allow me to interrupt but Mr. Smith wants to add something.

[Wayne Smith]

I guess I need to disagree with my colleague in terms of what the literature actually shows. One of the leading experts on non-response is also the head of the U.S. Census Bureau and therefore my colleague. He's also a member of Statistics Canada's advisory committee on statistical methods. Recently, he wrote an excellent paper on the whole topic of non-response, which I would invite members to consult.

He said a number of very important things that I think are very relevant to this debate. He said that non-response can but need not induce non-response bias in survey estimates. He said the non-response rate of a survey alone is not a very good indicator of the magnitude of the bias. He said the risk of non-response bias, not non-response bias itself, is reduced with decreasing non-response rates. He said that ultimately there is very little empirical support for the notion that low response rate surveys de facto produce estimates with high non-response bias.

I just want to come back to the point that we cannot predict what the outcome will be before we start.

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

That is what I wanted to say. Furthermore, we have raised the number of potential respondents to 30%, if not 35%. Previously, 20% of households were obliged to respond. So, since the number of potential respondents is higher, do you not think that the potential risk that you foresee will be minimized? If not, why not?

[Dr. Ivan Fellegi]

Because the fact that we increase the sample size is not going to make those who don't want to respond want to respond, and they are not representative of the rest of the population. The fact that we are increasing the sample size just gets probably more middle-income, white, third-generation Canadians into the sample, as opposed to high-income or low-income people, aboriginals, and new immigrants.

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

Since the risk is real in relation to groups such as the First Nations and people living in an urban environment, that is to say groups that might not respond, has Statistics Canada taken any special steps to try and get the highest number of respondents possible within those groups for the next voluntary consensus? Naturally, you know that those groups tend not to respond.

[Wayne Smith]

There will be several parts to my answer.

As I said, we looked at the study mentioned a while ago and that we put on our website. We tried to determine which groups might be less likely to respond. We determined where they are in order to focus our efforts there. Also, there are all the steps I have mentioned that could be taken to compensate for non-response situations.

Here is another important point. We have never assessed, for most of the variables, the degree of response bias in 2006. So, we do not know the standards which have been established, nor the problems existing in those communities.

Finally, about the reasons why those groups are underrepresented, I have strictly no evidence that they are more likely not to respond. Even though it is commonly held opinion, I have no evidence that they refuse to respond. So, the fact that the survey is voluntary will not really have any effect on the level of representation.

[The Chair, Conservative, David Sweet]

Merci, Mr. Smith and Monsieur Généreux.

Mr. Smith, in your earlier answer, you mentioned a publication regarding response bias. Who was the author of that, just for the clerk's...?

[Wayne Smith]

The author was Robert M. Groves. I have another copy of the article, which I can leave with the clerk if you wish.

[The Chair, Conservative, David Sweet]

That would be great. Thank you.

I don't know what the source of the feedback we're hearing is, but anyway, we'll move on to Mr. Masse now for five minutes.

[Brian Masse, NDP, Windsor West, ON]

Thank you, Mr. Chair.

This is the whole point, though, that we don't know, so really what we have is a $685-million crapshoot, because we don't know, if we can't nail down exactly what the reliability is going to be, how valuable the data will be or if it is even going to stand the test of time.

I'd like to ask the panel members to put aside for the moment the national survey versus the veracity of a real census. How important is it for the stability of this data for other types of research? I think that's one of the things that gets lost the usage of the material coming out and how it can actually be used for other scientific research.

[Ian McKinnon]

Let me start with that, Mr. Masse.

The census underpins all of Canadian social data. It is the benchmark that we use to adjust and weight any subsequent surveys that are done. It is absolutely fundamental. For example, knowing our unemployment rate accurately depends on drawing a good sample for the labour force survey and knowing that the responses there reflect the general population. For that, you need a benchmark, and having a benchmark that is relatively current is extraordinarily important.

It also is almost our only source for micro-area data. Any question that requires accuracy at a very small level of geography, or among very small groups of people, even if they're widely dispersed, requires the kind of volume that only a census can generate.

It's not efficient to use it for everything. That's why survey programs were introduced, but it does underpin, and that's why in fact the current deliberations are so very important.

[Dr. Ivan Fellegi]

I can only add that there has been an enormous range of concerns expressed, from banks to institutes of public policy, think tanks, health organizations, religious organizations, and I don't know how many of the major cities of the country. Something like 50 officially have taken a position on this issue. The list is long and encompasses almost every kind of user.

The census is the single most widely used... It's not just for research, but for decision-making, for evidence by businesses that want to locate outlets or to locate plants. They need to know what their potential labour force is or what their client groups are, and that's by small area, typically.

It's an enormously widely used information source. That's why I agree with Mr. Smith that the only thing we can know is that the risk of bias will increase substantially. We can't say that it will be biased, but without testing, it shouldn't have been sprung, given the wide use of the data in this country.

[Brian Masse, NDP, Windsor West, ON]

That's a normal practice for just about everything. If you build an airplane, you test it before you put the passengers on it; you don't just send them up.

We also have another private member's bill, one that I've tabled, that would make the chief statistician independent from politics. Whether we have a voluntary or non-voluntary census or national household survey, would that bill be a benefit for creating more scientific and basically structured consistency to ensure that our data management is reliable?

[Dr. Ivan Fellegi]

Well, I have maintained and recommended that the fundamental principles of official statistics that were adopted by the United Nations should be basically incorporated in the Statistics Act. That would make decisions of methodology independent of political interference.

[Brian Masse, NDP, Windsor West, ON]

Mr. McKinnon, do you have any comments on that?

[Ian McKinnon]

I have nothing to add to what Mr. Fellegi said.

[Wayne Smith]

It would be highly inappropriate for me to comment on that one.

[Brian Masse, NDP, Windsor West, ON]

I thought maybe one of the notes you got forwarded to you might have said yes. I was hoping.

No, I'm trying to look for more scientific answers than anything else.

[The Chair, Conservative, David Sweet]

Mr. Masse, I'm sorry, but you've run out of time again.

[Brian Masse, NDP, Windsor West, ON]

That's fine. Thank you, Mr. Chair.

[The Chair, Conservative, David Sweet]

Now we go on to Mr. Lake for five minutes.

[Mike Lake, Conservative, Edmonton—Mill Woods—Beaumont, AB]

Thank you, Mr. Chair.

I'm curious to ask a possibly technical question. Maybe you can help me out with this. There was some discussion about bias earlier, and I just want to clarify. Was it the case that 94% filled out the long-form census in 2006?

[Wayne Smith]

It was 97.3%.

[Mike Lake, Conservative, Edmonton—Mill Woods—Beaumont, AB]

Okay.

[Wayne Smith]

Oh, I'm sorry; you said the long-form census. I take it back. It was 94%; you're right.