Industry Committee on April 6th, 2017

[Ian McKinnon]

I have two quick notes.

First, it is very difficult to draw a very firm line. To use the example of response burden in the technical sense, with 15 different operations a minister might say to a chief statistician, that's a huge burden. That is also a core methodological issue about how you collect the data, and a good example of where those intermesh.

[The Chair, Liberal, Dan Ruimy]

Thank you.

Mr. Dreeshen, you have seven minutes.

[Earl Dreeshen, Conservative, Red Deer—Mountain View, AB]

Thank you very much.

I had the pleasure of going to Belgium for the Blue Sky 3 forum and so you get a chance there just to see the value of data analysis and innovation, and how important it is throughout the world. Of course, there are different ways that the international community assesses information. Of course, our universities are paramount in being able to use that as training ground and then as an expansion. That's a critical part of it.

One of the things I'd like to ask—and perhaps, Mr. Sheikh, you could reflect on this—is what statistical tools were used to correlate the long-form census data with the results that were obtained from the long-form survey? The data came in two different formats, but it's similar to 60,000 Canadians being of the Jedi religion. There has to be some analysis and some way in which you relook at the data.

[Dr. Munir Sheikh]

I can give you an example of a comparison of the survey with the census. In 2011, there were some areas where questions were asked in the short-form census that were the same as in the long-form census, the national long-form survey. If you look at that small number of questions and compare the responses, you have a fairly good idea as to how good or bad the survey is. The census is the census, and everybody had to answer it.

I'll give you an example of one area. In the Toronto metropolitan area, there were hundreds of units that responded. I looked at many indicators, but let's say apartments, so how many apartment buildings existed in various areas of Toronto. The ratio of the response in the short-form census to the survey was 0.25 to 3, and that tells you the error in the survey.

There are ways of comparing those, and in fact StatsCan does it all the time. We have a lot of voluntary surveys, and we test them against the census and we adjust them.

[Earl Dreeshen, Conservative, Red Deer—Mountain View, AB]

That's one of the points I've tried to make. Statistics is a tool. You work with it; you take the best information that you have, and you go forward. I know, for example, with the homelessness surveys and so on, people were saying they may not have had that data, but people have found their ways around it and there has been success that has come through that.

With the long-form census, as far as farmers are concerned, if you get that it's as though you've drawn the black straw. They're after you constantly about why you haven't got this in yet or why you haven't done that. Prior to that it was, “Well, you're going to go to jail,” if you were hauling your own grain across the border.

If these are the situations you have, it's about the timing of them. Lots of times when these things are sent out, you're on your tractor. You may have your phones and you're all linked up, but I know this is one of the reasons why in the agricultural communities you say to give you a break. Of course, in your case, you say, “I have all this information they're asking for anyway. Now they want me to go back and drag all of this out and complete all these forms.” Yes, it's important, but again, there are a lot of different ways of analyzing that.

The other thing I want to talk about is the census questions that are there. When they come from government, from any government, they can be couched in the type of commentary that you might want. There may not be any questions on how much carbon tax you are paying right now, because they might want to talk about something different from that. They may want to talk about green tech and where that is going, but they may not want to talk about clean coal technology and all of these sorts of things because those are outside of a policy position that they have.

How do you on your council look at these sorts of things and say that you know which is the political side of this and which is the natural resource side and how best that should be presented? How do you then tell the ministers or whoever is making those decisions to back off a bit here, that you know what the difference is, and you know what the situation is?

[Ian McKinnon]

First, I'd like to distinguish between the census, for which there is a very long history.... Frankly, statistical agencies are very conservative organizations. Given the choice between changing a question over time to reflect changing circumstances and asking the same question twice so you have the same comparability over time, they always go for the comparability, or almost always. With the census, there is a very well-structured, extensive consultation with stakeholders, both with the people who provide the information and with the users.

With regard to individual surveys, there they are fairly extensive. If we were to do an energy-use survey, a lot of it would be from administrative data just explaining what it is. In that case, there are many....

Here's where policy—and not methodology—intervenes. If the government of the day says it has an important issue area that it needs to work on, be it housing prices, the debt of Canadian households, homelessness, or the role of NGOs, it can—and I come to Statistics Canada—say, “This is our policy interest. Design, and you will be funded for, a survey.” This would usually come through a department, through extensive consultations, and on it would go. There is, I think, a good example of where the broad policy interests of a government are distinct. The chief statistician can say, “Here is the way to gather information on that topic”, and they would control....The government would say, “Here's our policy interest area.”

[Earl Dreeshen, Conservative, Red Deer—Mountain View, AB]

Do you have any other comment here?

[A voice]

The same.

[Earl Dreeshen, Conservative, Red Deer—Mountain View, AB]

I have one last question. Is it important to reverse the mandatory nature of StatsCan's participation in the Shared Services Canada initiative, and would it maintain or restore management control over IT to Stats Canada including IT's hardware budget? Is it important to allow them to have that flexibility?

[Prof. Mel Cappe]

As I mentioned to Mr. Lobb, I think it is too early to answer the question. There are advantages to having Shared Services Canada. There are disadvantages. I know that when I was managing departments, and when I was deputy minister of Human Resources Development as it then was, which is now ESDC, all of the informatics for the Canada Pension Plan and the Employment Insurance Commission were done in-house. Frankly, we would have benefited from having an outside manager who was handling this for us, because we ended up cancelling a contract for $750 million because the contractor wasn't delivering. We could have had a better result if it had been done by a centralized government organization. So I've seen both—where it's better to have it under the control of the department and where it's better if the government, the whole of government, is trying to manage it.

[The Chair, Liberal, Dan Ruimy]

Thank you very much.

Ms. Blaney, you have seven minutes.

[Rachel Blaney, NDP, North Island—Powell River, BC]

Thank you. Again, thank you so much for your time.

One of the questions that I have is around the nature of collecting information through the two different methodologies.

To use a concrete example, in the context of Bill C-36, will comparisons between data collected from the mandatory census versus data from a voluntary long-form census be statistically reliable?

[Dr. Munir Sheikh]

I think StatsCan made it pretty clear in one of the releases, which is on the website, that the two are totally different and you cannot compare the results.

[Prof. Mel Cappe]

I would just note that my students have been writing papers, and 2011 is an anomaly.

[Ian McKinnon]

Furthermore, and to go back to something Mr. Sheehan talked about, the voluntary nature of the NHS meant that for small areas and specialized subpopulations, no data were released, because they didn't have the confidence they required.

[Rachel Blaney, NDP, North Island—Powell River, BC]

Thank you.

Dr. Thomas, can you elaborate on your proposal for the appointment process for the chief statistician? Why is it important that the government engage in this process in a collaborative and transparent manner?

[Dr. Paul Thomas]

I've tried to make the case that the chief statistician position is different from that of the regular roster of deputy ministers, in my opinion. That individual needs to be part of the deputy ministerial community, meet regularly with the clerk of the privy council, and be involved in the discussions of the statistical needs of the various departments and agencies of government. In other jurisdictions—as Ian McKinnon mentioned, we talked to people in the U.K., Australia, and New Zealand—there is a separate process of appointment.

I think that we should have a more customized, tailored appointment process for the chief statistician. There was one private member's bill that even suggested that individual should become another officer of parliament or agent of parliament. That's a terrible idea, a very bad idea, but—in recognition of the status of Statistics Canada as a “statutory agency”, which implies more independence, more impartiality, and more autonomy from the central decision-making apparatus of government—I think we should change the appointment process.

It will be hard to get any prime minister, regardless of which party is in power, to give up the prerogative to make these appointments, so I'm just saying we need a different advisory mechanism while retaining the right of the prime minister to select the final name.

Someone mentioned earlier the possibility that there might be advance consultations with leaders of all recognized parties in the House of Commons. That could be done, but it will slow down the process and we already have a significant backlog of order in council appointments. These sensitive positions like chief statistician should not be held on a probationary basis for very long.

I think there's a way here to give some opportunity to get better background knowledge about someone and make a qualified appointment. Getting the right person is critical, as I said earlier.

[Rachel Blaney, NDP, North Island—Powell River, BC]

Thank you so much.

Mr. McKinnon, I'm going to come back to you.

Does the outsourcing of data collection responsibilities to an agency outside of Stats Canada—in this case, Shared Services Canada—introduce another area of risk to data privacy? I'm not talking about firewalls and hacking. What I'm really wondering about is the data collected by Shared Services Canada on behalf of Stats Canada being used by other agencies or departments for purposes unrelated to the original data collection purpose.

[Ian McKinnon]

First, in many cases, the collection itself, as I understand it, will continue to be done by Statistics Canada, so we're talking about the data processing operations. I want to demur on the question because I really do not have the technical expertise, nor does the council view itself as having the kind of expertise to make a sophisticated judgment.

As I said earlier, in addition to the technology involved—the machinery, literally—corporate culture is critical. We do have confidence in how deeply ingrained confidentiality is in the corporate culture in Stats Canada. I cannot speak for Shared Services. That's not to say they do not have it or will not develop it. I simply don't know.

[Rachel Blaney, NDP, North Island—Powell River, BC]

Thank you so much.

Those are all the questions I have, so if there is any time left, I'm happy to share it.

[The Chair, Liberal, Dan Ruimy]

Thank you.

For the final five minutes, we'll go to Mr. Jowhari.

[Majid Jowhari, Liberal, Richmond Hill, ON]

Thank you, Mr. Chair.

I welcome you all.

I want to go back to the long-form census, and I want to talk about two things: number one, whether it should be made mandatory or not; and number two, whether the scope of the long-form census should be legislated or not. Prior to you gentlemen coming in, witnesses talked about the fact that they all welcome the long-form census and the fact that it's giving us more data, but they identify as a gap the fact that it hasn't been made mandatory in this bill. Also, a few of the witnesses said it is an issue that the scope is not legislated.

I'd like to actually go around and hear from all of you for about 45 seconds each.

I'm going to start with Mr. Sheikh to get your input vis-à-vis what the issues are if it's not made mandatory, because right now in the bill it is not mandatory, and I want to understand what the issues are.

[Dr. Munir Sheikh]

The Statistics Act is totally unclear on the issue and Bill C-36 is silent on the issue.

What I am suggesting, making section 21 subservient to 22, would—in my view—solve the problem. The way it would solve the problem is that section 4 gives the issue of methodological decisions to the chief statistician, covering all areas in section 22. So as soon as 21 is made subservient to 22, the chief statistician would then be able to prepare a long-form census, and he is the one who makes the decision as to whether something is voluntary or mandatory. Of course it would be mandatory.

However, for accountability, you want to make sure the Government of Canada has the final say on it, and with section 21 around, the Government of Canada can approve it, reject the chief statistician's advice, and send him a directive at that time and say, “For the following reasons, we are not going to accept what you are recommending.” To me, that simple change would solve the problem.

[Majid Jowhari, Liberal, Richmond Hill, ON]

Perfect.

Should the scope be legislated?

[Dr. Munir Sheikh]

No, I think you need to have the flexibility to ask the best questions.