Transcript
Transcript: CSPS Virtual Café Series: Talking Canada Through Numbers, with Anil Arora and Nik Nanos
[The animated white Canada School of Public Service logo appears on a purple background. Its pages turn, opening like a book. A maple leaf appears in the middle of the book that also resembles a flag with curvy lines beneath. Text is beside it.]
Webcast | Webdiffusion
[It fades out, replaced by a Zoom video call. The video window is filled with a white man who wears black glasses and a black shirt. He sits in a dining room, and behind him is a table set up with a coffee maker. A sepia-toned landscape hangs on the back wall. The man smiles.]
Taki Sarantakis [TS]: Welcome to episode 3 of the CSPS Virtual Cafe where the Canada School of Public Service brings to you rich policy discussions without politics or partisanship. I'm Taki Sarantakis. I'm the president of the Canada School of Public Service. And today we're talking about data. Data, data, data, data. What is it? Why are we all talking about it and why does it matter to you as a professional in the government of Canada? And we are really, really privileged and honoured today. We have two of the best. The first is Mr. Nik Nanos, who is the Chief Data Officer of Nanos Research.
[A cursor unpins Taki's window in the Zoom call, and two more video windows appear. In one of them, Nik Nanos waves. He is a white man with short grey hair. He wears a black suit jacket over a white button-down shirt. Nik sits in an office. Behind him is a table covered with papers. Framed pictures hang on the back walls.]
He is also an author. He is a television pundit. He is the former chancellor of Carleton University. He's a fellow at a bunch of places in the US. He's just really, really smart. He's a genius. Our next and our second panellist is Mr. Anil Arora.
[In the third video window, Anil smiles. He has medium-brown skin and short black hair. His camera is angled up slightly, so we see the upper walls and ceiling behind him. A painting of snowy mountaintops hangs on a yellow-beige wall.]
To my knowledge, he has not written a book. He is not a pundit.
[Anil chuckles.]
He's not a fellow at any US universities, but he is a genius like Nik, and they're both numbers guys. And we're really, really, really honoured for them to be spending an hour with us today. So, I want to start first with Nik. Tell us a little bit about what data is to you and maybe tell us something personal vis-à-vis data, like how you started in this business, so to speak. Why are we talking to you today in July 2020 about data? What bad thing happened to you in your youth?
[Anil smiles.]
Nik Nanos [NN]: Well, actually, it wasn't a bad thing. It was an interesting thing.
[Nik's video is pinned so his window fills the screen.]
I was asked to do a project. As some of you might or might not know, I'm a soccer buff. So, I was asked to do a project at the very beginning of my career where I had to do projections on which countries would qualify for the World Cup, and then to model out tourism travel for sports in order to help an airline recalibrate all of its routes. You can imagine from my perspective, I was able to use math in order to look at my favourite sport to figure out what the odds were of different countries qualifying for the World Cup and then linking that to economic and tourist data. You know what? I was a junkie from right after that, because what I'd learned is that data brings stuff to life, and it can be really interesting and fun.
[Taki and Anil's windows reappear.]
TS: That's really cool. And Anil, why are we talking to you today? How does somebody become a chief statistician of Canada?
Anil Arora [AA]: Well, I was going to say after that introduction that you were going to say, "there's got to be a contrast to Nik." I'm the contrast to Nik.
[Nik and Taki laugh. Anil's window fills the screen.]
Obviously it's a privilege for me to represent an entire institution, an institution that's been around in Canada for over 100 years and arguably the best in the world. In our case, it's a federal institution. It has a very clear legal structure and a mandate. And for me, it's just an honour and a privilege to be part of this amazing institution that puts out not only a baseline for the kind of work that Nik does and many others do, but also provide some insights that are really, really critical for policymaking right across our country. In my case, it was an accident, joining of Statistics Canada, some 32 years ago. I started my career out in the west working the oil and gas sector, and it was a pretty rough time in the mid '80s, and I decided I want to do something different. And so I changed my career from working in the oil and gas sector into working with data. So it's just a different kind of mining, that's all.
[Anil smiles. Taki and Nik's windows reappear.]
TS: Yeah. We're going to talk a little bit about the data over the next hour and talk a little bit about trust, a little bit about privacy. La session d'aujourd'hui est essentiellement en anglais, mais on a la traduction simultanée disponible. Also, we invite your questions. There is a little hand-raising icon on the top right-hand. Ask your questions and those will be filtered to me over the course of the discussion and we'll weave them in as we're going through.
So, I want to start with data. You guys have been talking and working with data for a long, long time, but the rest of us are just catching up. Nik, you've been in this business a long time. You've been polling forever. You've been providing data advice to Canadian companies. And now you said you've been working in data in one way or another for 30 years. Why are the rest of us talking about this now? What's happening? What is this data revolution? Maybe I'll start with you, Anil.
AA: I think data has always been there—to observations, and we put them into a format where we can then try to make sense out of this thing. Observations have always been there and they always will be there, our nature and our ability to translate them into bits and bytes. I think there are a number of factors. One is just pure computing power, the ability to be able to deal with vast amounts of information in very efficient ways. Second is the proliferation of devices that actually now create and spew data. Everything, almost, is producing data. This very setting that we're in is producing data and somebody's going to analyze it.
[Taki nods.]
Somebody's going to review it. Somebody's going to make some sort of sense out of it, and so on. Thirdly is the expertise. I think people are really now becoming more and more data savvy because they see the value that they can look at the trends. They can look at things that others can't to use as a competitive advantage. And there's a culture which says we want to be driven by data, facts, evidence, and so on. From just a strictly financial perspective, all the richest companies in the world, guess what they do? They deal with data. So it in itself built this ecosystem which says you'd better get good at it because you know what? You could be rich one day. So all that to say we, I would argue, are just at the beginning of this revolution, the fourth revolution or data revolution, as many of these terms, because we are looking at the internet of things. We're talking about billions of devices that are just on the verge of coming out. And so, you know, if somebody isn't good at data, you'd better get good at data pretty quickly here because it's not going away any time soon.
[Taki nods.]
And not only is it the volume, but we have to deal with some of the societal issues, like you mentioned, of privacy. We have to deal with some of the social and economic impacts of data as an asset. So that's why I think it is absolutely top of mind for everybody today.
NN: Yeah, what's interesting is—
TS: Nik, do you agree?
NN: Kind of, yeah, but I think we have to put some risks on the table, too. You know, it's one thing for us to talk about the wonderful world of data. And the fact of the matter is, among the more compelling things that we can do now have been enabled by technology, the interconnectedness between different data sets that we can now do with open data—and StatsCan is a leader in this particular area—where now researchers have access to data, but now they can connect different data sets, which was very difficult 10, 20 years ago, 30 years ago when we started in this business.
And if we want to talk about geeking out, whenever we do a survey, we talk about estimated error. Well, now with computing power, I would expect that we will move to observed error as opposed to estimated error, because a lot of these statistical tests were developed, like, 50 years ago when there were no computers. There was only so much power.
[Anil and Taki nod.]
I'll tell you, the thing that I worry about right now is the fact that, as everyone becomes immersed in data and starts to use data, that it is being abused and misused. What do I mean by that? My PhD advisor used to say, "just because you can run a statistical test doesn't mean that it's valid or it's correct" because garbage in, garbage out. It's a mechanical process. It needs human insight and experience in order to know whether something is right or not.
[Anil nods.]
And I worry about people not being properly trained using data, because one thing that we have learnt is that data can be misused by organizations and individuals in order to mislead people whether on purpose or by accident. So I worry. We have a richness of data out there, but along with that and the accessibility comes a risk in terms of the misuse and misinterpretation of data.
TS: That's a very good point. So let's jump into that, because when I was a little kid, so to speak, data was a good thing. Data was something we all agreed on. Sometimes we disagreed on the interpretation, but we all agreed on of what was the basic data. Now, what we're finding more and more is some people are afraid of data. Some people are using data and some people were having now the kind of dynamic of the weaponization of data. And that's a really interesting concept when you think about it, when you take—because when most people think about data, they think of something that is kind of intrinsically neutral, something that is kind of value free, something that's observed rather than interpreted. But now we have a whole subset, a whole industry built around the weaponization of data. And Nik, you kind of introduced us to that. Anil, talk to us a little bit about how you see some of the dangers of data or some of the potential pitfalls that we particularly in public policy have watch out for.
AA: Yeah, I think Nik's absolutely right. It's kind of a bit of an irony that we've never seen as much data, and yet we've never seen as little information or evidence or intelligence that we need. It's a bit of a dichotomy and we need to be mindful of exactly the kind of points that we'll raise. First, there are inherent biases. What you measure—and there have been some famous examples of where the police go to look for drug interactions or whatever it is, where did the calls come from. And in the states, some work done on they come from primarily black schools or black neighbourhoods or whatever it is.
[Taki nods.]
So if you feed some of those biases—and they're not unique to the United States, you see them all over the place. And if you feed them now into engines and artificial intelligence and you then build, saying, "Well, where should we put more cop cars?" just to take that example. Well, of course, you're just going to go predominantly—and when you dig back and you find out, well, that's not actually where a lot of the drug deals might be occurring. They might be occurring in richer neighbourhoods. And so we need to be mindful of that.
The second thing is, even in the private sector, there was this twist between when data made its transition from being a passive resource to an active resource. So when they said you are the product that is being marketed, all sorts of people came up and people said, "Well, wait a second, when did I become the victim here? When did my say in this go away?" And the power of being able to present to a consumer, you go, "Well, wait a second, how did that engine know that I looked at the case for my iPhone or whatever it is, and now it's presenting me these kinds of things. How did all those things get connected in the background without me knowing?" And then you just look at the examples of the last election in the United States and so on.
So it's permeating at the very micro level, individual level, and then it's also permeating at the societal level. And in fact, in many cases, it's being presented to governments as a fait accompli. Governments don't have access to some of the algorithms or the data or the access or whatever it is, and so you see some of the big projects get into trouble as well. I think data is no longer used as a passive commodity because of all the things that we talked about. It is now becoming a very active component of our society. And like anything else, it can be used for good or, if not challenged or not properly governed, it can actually become a weapon as you just mentioned.
NN: Yeah. But the most common risk that I've seen, doesn't matter whether you're in the government or the private sector, is when someone has a hypothesis and wants to do research and they don't have the funds to do it properly.
[Anil and Taki nod.]
And as a result, it's kind of like having a key performance indicator, right? You have a KPI. In the perfect world, if you were designing a KPI, you would design the perfect KPI and then you would initiate the research, the original research in order to make sure that you reach your objective in measuring these key performance indicators. And you would be funded to do this properly. But I'll tell you what happens in many instances in the private sector and in the public sector is that they are not properly resourced to do the measurement. As a result, they report on KPIs that can be measured as opposed to what should be measured because they just don't have the resources to initiate the research properly. I think this is where you're reverse engineering and making—"Well, we have this data here and it wasn't collected for this purpose, but we can measure it. Let's ram it into our KPI."
I find many times it's done just because of convenience and budget constraints and there needs to be—I'd always say that on every research team, there needs to be someone who plays a challenge function, who says, "Well, I think there's a better way to do this," and to build that into the process, to say, "You know what, I know that we can measure this, but is this really a KPI or are we just putting this in because we have the data and it's easy?" And that's another constraint when it comes to doing proper research, because proper research to be done costs money. It costs money for the work that StatsCan does. How about this? No, I'm going to give you a plug. For the work that StatsCan does, it needs to be properly funded. Right? It's as simple as that.
[The three participants speak over each other, and their voices cut in and out.]
NN: Am I breaking up?
[Nik laughs.]
TS: No, no, no. I said you were—[inaudible].
AA: You know, Taki, just to—many of the folks who are watching us today are in policy areas, regulatory areas, and have got huge data stores. And so just to build our next point, the interesting thing is always the insight. The boring things are like definitions, standards, interoperability, metadata, paradata. And guess what? The utility of the insight and the quality of what you put out is absolutely dependent on definitions. I mean, if one data set said we're going to exclude anybody who's a senior citizen or anybody who's below 18 and the other one is about the entire population. And if you don't know that—I would challenge our colleagues across systems, say, "How much do you know about the data store that you have? What does it have? What are the gaps? What do you know? Where is the documentation? How clean is it, quote unquote? And are we just shovelling it into something else with some broad assumptions?" This is where you take that, as Nik rightly said, garbage in, garbage out. So we have to focus and that's what the data strategy, the federal data strategy that I was a co-author on last year said, is that we have to focus on the fundamentals. We have to get that right.
NN: And we tend to be wowed by big numbers.
[Taki and Nik's voices cut in and out.]
TS: So, you're both—
NN: No, I was just saying—
TS: You're both talking in one way or another about the credibility of numbers. And Nik, in your example, it was a little bit about laziness. Let's talk a little bit about being overtly nefarious. Now, you guys are professionals. You're highly credible, but there are a lot of people that play with numbers. If I had enough money and if I had enough inclination, can I basically get data to prove anything that I want?
NN: With an unethical researcher, yes. If there's an unethical researcher.
TS: Anil, is that becoming more and more of a problem in our society? One of the things that's happening is...
[Taki's audio cuts in and out.]
AA: You know, a data point doesn't have... I was just going to say, a data point itself doesn't have an ethical frame around it; it's a data point. So it comes back to the science. You actually have to know so much around that data point. When a data point carries its way through, it's all that information about that data point that actually makes it good or bad. There have been many, many studies, I think Harvard and Yale have put out studies, MIT put out a study which essentially compared really ludicrous kinds of associations and said, "Here you go." We've got to be careful about those kinds of things, because in terms of volume, there is so much data out there now that you can go look for a pattern and actually make it say that it is. But you know what? This is where I think there's an obligation on our part and this is what sets apart organizations like ourselves and Nik, because not only do we put out the nugget of information, but we put out all the qualifiers, the footnotes, the standard errors, the CBs, etc. And so this is where I would say expertise has to come into play. Knowing that it has a CB isn't good enough. To be able to understand it and to be able to say, "Wait a second," and then be able to critically look at it and say, "Well, hang on, is it good for my purpose? Is it real?" So we have to have the literacy and you have to have the curiosity to be able to differentiate between what makes sense and what is ridiculous.
NN: A lot of the problems, Taki, are not nefarious. So why don't we just use an example? Google did, for a number of years, crunch search terms in order to project the flu pandemic. And Google is a credible, large company. They have access to a massive data set that no one else has access to fully. And then they did projections and they'd say that we looked at a billion search inquiries that if someone typed in runny nose, flu and they tried to do projections and were in the news about when the flu would occur and they were never, ever right. But the people had comfort with the fact that it was a large data set. And I think that's one of the things I worry about now. And I would go into meetings and people would say, "Well, this data set has 100,000 points." And I said, "Well, I don't really care. It's not measuring what should be measured." To use the Google example, just because someone typed in cough or runny nose doesn't mean that they have the flu. The only way to really know if someone has the flu is to take one of those StatsCan laboratories into neighbourhoods and to have someone that is a professional to say you have the flu. Not, do you type flu into a search term. But these are not enough information examples that are misleading to the public.
TS: That's really interesting Nik, because we think in a lot of cases, the untrained people—and I count myself amongst those—that if you have a lot of data, it translates into telling you the truth. But you've called something like, what, 15, 20 elections in a row in terms of the winner. How many Canadians do you typically poll? Is it about a thousand? Is it about—
NN: Yeah, on our one-day election call, we'd have less than a thousand people.
TS: So you can kind of actually predict...
[Taki's audio cuts out.]
NN: Yeah.
TS: You can kind of certainly predict the behaviour of, grosso modo, a country of thirty-five million people with less than a thousand data points, and yet somebody could give you one hundred thousand data points and it could just be complete garbage.
NN: Yeah, absolutely. And this goes back to what Anil mentioned, and it has to do with the sampling procedure, the adherence to standards, the transparency, the level of effort. It's expensive to do this. If it's cheap, that's probably the first indicator that maybe—if someone spends much less than someone else in the same job, it's like having—I'd always tell people, everybody would say, "What's the secret to good research?" And I'd say, "Quality ingredients." I'd say, "You cook. If you have fresh vegetables, if you have good quality ingredients, if you follow the recipe properly, if you're consistent in your application of the technique, you're probably likely to have a consistently positive outcome." If you decide that you're trying to make cookies and you say, "I'm substituting margarine for butter." If you start to do substitutions, if you don't follow, if you're not consistent, you're going to have—And I said, people can have the exact same recipe, but the skill of executing the recipe and the quality of ingredients can lead to significantly different outcomes. One can be very pleasant and another one can be very unpleasant.
[Nik laughs.]
AA: If I could just add to that, Taki. It's a system, first of all. And it comes back to both your points about investments. You have to have a credible data infrastructure. I mean, you use the number about how can you get such accuracy from a thousand or fifteen hundred phone calls when we have thirty-five million Canadians, by the way, we have thirty-eight million now. But the point is that infrastructure underneath that tells you what the denominator is, that tells you what the age group breakdown is, the gender breakdown, the regional breakdown. Without it, you could never do a poll of a thousand or fifteen hundred and assign the weights of the responses to be able to say, "Well what does it mean for the population?" Now, you could still have other extraneous factors, like how many people turn out to vote and all those kinds of things. Those are separate, but it is an interrelated data system. So, you have to have the baseline. And the baseline in this case comes from the census. That is a massive investment in public infrastructure. And so I think to your point, up until now we've taken data infrastructure as almost a given. We have to now pay attention because that data is becoming so important, just like we build physical infrastructure. We're talking about social infrastructure. We need to have a strong data infrastructure. And in the current context, arguably those with a strong data infrastructure are doing better than those with weak data infrastructure.
NN: Here's something to chew on. We've decided at Nanos that our data is more valuable than our revenue in the valuation of our company. And that is kind of a turnaround when we think of what is the real value of our firm, and it's actually 30 years of collecting and curating quality data. It's not our revenue of X million dollars a year, because the revenue can fluctuate. If Nik gets hit by a truck, that might impact the revenue. But the data is actually the key value in the organization. And I think public and private sector corporations are now starting to wake up that in addition to the human capital, the data is kind of a significant piece of capital in the value of an organization, which in the past it was just something that was stored someplace. No one thought about the value of all the information that they collected.
AA: Well, in fact, just to build on that, Taki—
TS: So, let's—
AA: Well, in fact, just to build on that, Taki, Stats Canada for the first time and first in the world last year, we put out our first estimate and we know it's not perfect, but it's our first estimate of what is the total value of data in Canada. Believe it or not, we tackled that problem. And then we said, "OK, how much did we invest as a country in data last year?" And would you believe in the last couple of years we've invested more in data than in our natural resources, in our manufacturing. So, to come back to your first point, why is data so hot? Well, guess what? People are putting a lot of money into it as to Nik's point because they're seeing the value and they can now actually tangibly put a dollar figure to the value of data. And to those that say all data is the new oil and so on, I'll argue data actually gets even more valuable as you use it and link it and get an insight. So it's a very different product than any other sort of tangible good that we've seen.
TS: Yeah, when I talk about data in a public forum, I say to those who call data the new oil, you're wrong. Data is the new oxygen.
[Anil and Nik nod.]
It's something that we all need. It's something that none of us have a monopoly on. But it's something that without it, we're kind of nothing. I want to shift a little bit into some of the actual data as opposed to talking about data as maybe a bit of a profession or a bit of an institution. And it seems to me that one of the things that's really important in public policy is trust. And we're seeing more and more, we're seeing declines in a lot of the trust indices on a lot of our public institutions—and public, very broadly defined, whether they're doctors or lawyers or professors, teachers, etc. Tell me a little bit, who do Canadians trust right now? My sense is, if I were to ask this question 30 years ago, we trusted a lot more people and a lot more institutions than we do today. A, is that kind of true? Is trust eroding? Number one—and number two, who do we trust today? I don't know who wants to tackle that first. Who wants to kick us off?
[Anil motions for Nik to speak.]
NN: Sure, I'll start. The thing is, many of the trust measures that we've seen over time are actually context dependent. Why don't we use firefighters as an example? So following 9/11, trust in firefighters actually rocketed through the roof because people around the world saw firefighters and their role in 9/11. And we've also seen in the past—now we're seeing fluctuations, trust in the police. We've seen trust in the police go up. And then we've also seen trust in the police show a bit of a decline based on what's in the news. But what I'd like to put on the table, something that is more disconcerting about trust, at least in some of the research that we've done, is that in some of the research that we've done, it suggests that there's a correlation between people being vulnerable to misinformation and their level of trust in institutions, that Canadians that are more likely to believe mistruths. Especially things that are on the Internet, Canadians are less likely to trust even things like public health authorities and stuff like that because they think that there's some kind of conspiracy.
One of the new dimensions that we're starting to look at now compared to the past is: what are the drivers of trust beyond something that's in the news that might reflect well or poorly on a particular profession? But I'll just make one more comment and then I'll pass it over to Anil. The other thing is we started tracking—the nice thing about having a research company, and I don't know, Anil, if you're allowed to do this kind of thing, like, "I have an idea and I'd like to do research on that." It happens. Nik wants to do research on something. The research gets done. So, can you do that at StatsCan? Can you just say, "Hey, I'd like to do a study on X?"
TS: We have no competition—
[Anil laughs. Taki's audio cuts out.]
NN: No? Yeah? So we decided to start to track emotions that people would use to describe how they feel about the federal government, not the Liberal government, not the prime minister, but the government in Ottawa. And we used a combination of emotions, positive and negative emotions from pessimism, optimism, satisfaction, that kind of stuff, and ambivalence. And we started tracking that over time and what we found is that in the period ahead of the COVID-19 pandemic, there is a significant proportion of Canadians that would use words like anger or pessimism to describe how they felt about the government in Ottawa. Fast forward to now that we're in the pandemic, and words like satisfaction. So it's not positive and it's not "I like the federal government," but satisfaction. The thing is that we've come through a period and all institutions are imperfect. But I think for many Canadians, when they look at institutions like the Supreme Court, like the federal government, like the police, it's when they're tested that people get a sense of how much they can trust and their level of satisfaction. And I find the pandemic has been kind of changing that. And the Internet has been influencing the level of trust and mistrust of institutions, a lot of it because of misinformation.
TS: And you can also see what you said about trust being context-specific. Anil, who do we trust?
AA: Well, look, I mean, there are a lot of studies that have been done on trust. Edelman puts out the barometer. Nik has just talked about some of the work that they're doing and they're measuring it on an ongoing basis and seeing what's going on. I think—a couple of points, if I may. First of all, trust is not just a "nice to know." It actually drives public policy in a big way. Let me give you two examples. One is, obviously right now, I mean that the federal government and I would say governments writ large are enjoying a very high level of trust and in fact, trust levels that we haven't seen for quite some time in some of our own work. For example, trust in the government right now is that seventy-seven and a half percent. It's just incredible. And let me come back to the policy. Let me give you two examples. One is mandatory: wearing your masks. Interestingly enough, there's a difference whether you had a high level of trust in the government or you had a low level of trust in the government as to whether you would comply or not, which then raises an issue. Do you make it mandatory or do you appeal to the social conscience of people to go along with it? Or vaccines, do you take a vaccine or do you make it mandatory? And some of the research has shown that, in fact, if you make it mandatory, less people will take it as opposed to if you appeal to their social sense of responsibility. Similarly with closing the borders. In other words, trust is going to be a very strong—along with public health advice and so on—going to be a strong determinant of where a government decides to—and what levers it uses as well.
NN: Yeah, but the thing is, what Anil has put his finger on is what I always look for, are kind of structural views that drive other views because... If I could use an analogy, it's kind of like global warming would be a structural—I'm not talking about the opinion, but the fact—is a structural thing that influences local weather. Local weather changes every day and it goes all over the place. And in our research and the areas that we've been focusing a lot more is one of the structural determinants of a lot of these things has to do with educational attainment. In the past we'd say, "Oh, let's look at the income numbers and gender and cities and rural and stuff like that." And then we start looking at the educational numbers, educational attainment, and then we'd say, "This is a key driver of how people, how much they trust the government, how much they trust science, how much they might trust their neighbours, their worldview, their worldview on trade and whether trade is a threat or not."
And the interesting thing about educational attainment as a driver, and this is Nik's tip of the day: if you're doing a study, ask for the tables on educational attainment. That's the first table that I always look at.
[Anil nods.]
The other stuff never surprises me. But educational attainment has been something that we've been monitoring very closely for the last decade or so. And to look at that as a potential determinant. But from a public policy point of view, what does that mean? That means that if you're a government, provincial government, federal government, and you're interested in the societal good, public good, it means probably that a robust educational system is probably the key determinant to a positively and well-functioning society and access to education. This is where—we talked about this in the beginning—research comes to life, right? Because we can see if there's a correlation between educational attainment and misinformation, a correlation between educational attainment and people's views on issues, their views on marginalized groups, their views on public health guidance and stuff like that. There's a public policy solution to that that the research kind of puts a spotlight on to say, you know, what's our educational strategy? Because if we can get our educational strategy right, it will lead not just to better public policy outcomes, but a society that is willing to embrace those better public policy outcomes for everyone's well-being.
AA: So I would just perhaps add to that, we've got to be careful about averages.
[Nik laughs.]
There are determinants of attitudes and so on. Education is absolutely one of those top-notch things that we look at. But we're seeing immigration, for example. Immigrants have a different view. We're starting to see people of different ethnographic groups have differing views about trust in various policies and governments.
[Nik nods.]
There are regional differences, big time regional differences in Alberta versus Quebec versus out East on certain things. So the advice that I would give is everybody thinks that there may be an expert in questionnaire design—use standard modules on these kinds of information that are tested and that have been tested and are true. And so reach out to experts. Of course, we're happy at Statistic Canada to kind of help with some of those kinds of things, we've used them over and over again. And with it comes all the formula and the algorithms of how to analyze what you get out of it. And that's really what public policy is about, right? It's about knowing the effectiveness of the lever that you're going to exercise and how it plays out differentially across this country, because the needs are very different. And so Nik is absolutely right. You've got to dig deeper into some of these issues.
TS: That's great. So, the next area on data that I'd like to move to is obviously we're living through a very interesting time. We're in about day 110, 115, depending on how you count the COVID and the shutdown and the slight reopening. Put whatever label you want on the era. But there are a lot of interesting statistics and a lot of interesting data that that's been generating. I want to ask each of you, since COVID started, what's the one or two pieces of data that have surprised you the most? Nik?
NN: Well, going back to this whole concept of structural abuse and trying to get a grip on where things are going to go, I was quite struck by the fact that when we recently surveyed Canadians and talked about what they perceive to be the greatest impact on them personally of COVID-19 beyond things like jobs and their health and stuff like that, the two top things that people said had changed them was a greater appreciation for your family and loved ones, and an appetite to return to a simpler life and a recognition that my standard of living may not be the same in the future as it is today.
And what I'm seeing—as a researcher, I am regularly humbled by what I see because I look at stuff and then it gives me a sense of what's really going on out there. And what's clear is that there are a lot of Canadians—this is a different kind of transformative experience because it's not a shock, because it's ongoing. It's like... It takes a long time to digest Thanksgiving turkey, right? It's not a quick hit. And as a result, there are a number of Canadians that are fundamentally re-evaluating what's important to them because they've been self-isolated and then it's kind of like, "You know what? I should be spending more time seeing my mom. Do I really need all the things that I need? Do I really need to do all the things that I've been doing in the past? And I need to be prepared to potentially have a lower standard of living."
From my perspective as a researcher, whenever I put a spotlight on something like that, I find that it's much more profound than, "Oh, do you think you're going to get the coronavirus?" Or, "Can you pay your mortgage next month?" And I think there are some profound changes and I don't know where they're going to take us, but I think we're at a moment in time where there is a significant part of the population—and it doesn't have to be everybody. If one out of every five Canadians decide to change their lives as a result of the pandemic, that's a massive upheaval in how our society will work. I think we're at that place right now because there are enough Canadians that are thinking about how they lived and they're now thinking about how they want to live in the future.
TS: Anil, what surprised you in the last 100 or so days?
AA: Well, a few things jump out. I think, again, as Nik so eloquently said, we're kind of trained not to go in with too strong judgments because you want to let the data talk. So you try to kind of not be surprised going in for fear of biasing the context within which you put things out. But a few things sort of resonate. I think that's what the data show us, is going back to fundamental first principles and the caring that Canadians have for each other. How youth are more worried about their parents and loved ones and so on than they are about themselves, for example, and that they all show that out. Why does that matter? Well, because they're going to behave that way as well, whether they wear masks or going and increasing their level of risk to themselves. So, these things have real meaning in terms of what a resurgence might look like or what a second wave potentially, hopefully not, might look like.
Businesses and their resilience—some of the work that we did with the Chambers of Commerce in partnership showed us the vulnerability of different sizes of businesses, small businesses, which are a huge driver of our economy, essentially telling us 40 percent of them would have trouble staying afloat more than two or three months kind of thing. When you think about how powerful an engine they are of growth and jobs and the vulnerability that's there, the fact that help—and we always talk about it in the theoretical, you know, social and economic and environmental issues are interrelated, while you can see how they're completely interrelated.
And the last thing I'll say is that what's come to light in the last while is the fact that we have gaps in data. Obviously, everybody's been watching the debate about disaggregated data on race and other things. And people have equated the gaps in the data as a kind of systemic racism. So in other words, how is it that in this day and age we have huge either geographic parts or vulnerable populations in this country that are just not captured? Nowhere do they exist and so we can't talk about how some things are impacting a certain group. People have repeatedly written to me and said, "This is 2020, man, how can we not know about these things?" And I would say just a couple of policy areas that I think—they're not a huge surprise right now, but they certainly were when they hit, which is the vulnerable looking after the vulnerable, the extent to which we were looking at low wage earners in crucial positions that were essential. Long-term care facilities and so on. I think these are really important in the policy decisions. And we're going to have to, in our case, provide data. But they certainly are a wake-up call to policymakers to say, "What are we going to do about this? What is the role of standards? What's the role of funding? What is the role of data in many of these areas?"
TS: And let me ask, from the surprise to the slightly maybe disturbing, has there been any kind of data that's emerged over the last hundred or so days that kind of scares you or where you go, "Ah, I'm not liking the way this trend is going?"
[Nik chuckles.]
NN: Well, we're going to talk about a statistician's pet peeves on data because we're hostage to convention and people are hostage to habits in terms of the data that they turn to. And if I might put what some might think is a provocative idea on the table, it's that we look at something like the unemployment rate—and I always call it joyless prosperity, the unemployment rate would be at four or five percent. Why doesn't it feel so wonderful if the unemployment rate is at a twenty-year low? And I think because the media tend to focus on the unemployment rate, because tradition dictates that we're trained to focus on the unemployment rate, I find in some cases we're not looking at the right number.
And from my perspective, one of the best numbers I like that comes out of Statistics Canada—I don't look at the unemployment rate. Anil, I look at your measurement of the participation rate, which measures basically able-bodied people and whether they're participating in the workforce. I think the latest numbers have it like six out of every ten are participating. I think that the twenty-five-year average, I think the high is sixty-seven percent. So I would hazard to say that if we said that four out of every ten able-bodied Canadians are not participating in the workforce, if we wanted to build a better society from scratch, would we start off by saying that four of every ten people who could work are not working? Would we accept that? Because basically we're accepting that. The things that worry me many times is that we focus on statistics that give us a false sense of security when we don't look at the number and say, hey, four out of every ten Canadians who can work are not participating in the workforce. Is that—maybe that is good. But, you know, as a layperson, it doesn't sound good. In my kind of ideal world, I think anyone that can be able to work should be able to have the opportunity to work and that we should be encouraging that. So what I find kind of disturbing is a focus on numbers, because we traditionally focus on them when there are other numbers that might be better indicators of how healthy or unhealthy we are as a society.
[Anil and Taki speak over each other.]
TS: Anil, unless you've got something that's scaring you, you want to jump off that last point?
AA: Just two points. One is just quickly, when it comes to the labour force, we put out a report. And it has you know, if you just look at the last few months, not only do we put out, as for the international definition from the International Labour Organisation as to what constitutes unemployment and so on, there's a ton of other information that one needs to look at. We put out data on what was the participation rate? What was, obviously, the unemployment rate, that has some merit and some value. But really we've been putting out what is a degree to which people are actually working in the utilization of labour, and there's a percentage that tell the number of hours. And, who's going to go look for a job when people are being told to stay home? So obviously there's going to be a difference in the denominator that's going to have an impact on the rate. You've got to go beyond the headline numbers—
NN: Yeah.
AA: It's there. It's there for the looking. And so I think that's to Nik's point. Go beyond. Don't be lazy and just look at the number and go, "Everything must be fine." Dig into and understand some of these things, because believe it or not, the changes in the numerator and the denominator that both constitute the rate are saying something and you've got to go beyond. But to your point about things that are important for us to watch for, the government spent a lot of effort in trying to deal with the inequities, whether it's gender, whether it's visible minorities, whether it's vulnerable groups and so on. And I think there's a huge, huge concern about an equitable recovery when it comes back to the sort of play between health and jobs in the economy.
Things like daycare are going to be a huge determinant about the ability for many women to be able to come back to their jobs. We've seen in the past people who have—and I'm now talking generally from the 2008 recession—many people who did not get back into the workforce many years out. Forty-five percent of them that were laid off and so on had a really tough time getting back in. And in fact, fifteen percent never made it back into the workforce. So we're going to have to watch for some of the more long term effects. And the fact is that there are going to be other things that are going to determine the economic recovery. You've got to look at some of the social aspects in order to figure out what are going to be the economic impacts of some of these things. It's not a solution and answer that's simple. We're going to have to go beyond it and look at the factors. Thirty percent of Canadians have an underlying health issue. Think about that. What is the level of comfort and trust and confidence that they're going to have in getting into an elevator or into public transit? We have to look beyond that, I think, to say, "Oh, yeah, here's the answer to that problem." It's more complicated and that's what the numbers are telling us.
TS: Yeah, and you and Nik both raised—Nik directly and you a little indirectly—an interesting point which is as policy analysts, we go back to the same numbers over and over again. OK, what's the unemployment rate? What's the GDP growth? What's the number of Canadians who have a university degree? We kind of go back to the same four or five or six or ten indices or whatever the case may be. What are some of the indices we should start looking at now? What's telling us real things about the economy? Should we be looking at more inventory levels? Should we be looking at investment flow? Should we be looking at ownership? I know you talked about how much—I would venture to guess that not a lot of policy wonks listening to this argument or this discussion today would know your point earlier, that we've invested more in data, we have more value in data than we have in the automotive sector or in the oil sector or whatever the case may be. What are some of those other indices that you would tell your fellow policy people to say, "Take a look at this and this and this," if you're doing economic policy at Transport Canada or at ESDC or at Natural Resources.
AA: I'll maybe kick it off and Nik, please jump in and correct me or add your perspective to it. I can equate it to like driving a car, right? It all depends on where you want to go. You don't just get in and hit the accelerator and go, "Gee, I wonder what the RPM is doing." You have to know that our society is a complex vehicle that is an intricate dance, if you like, of different indicators. So, you have to know where you're going. Are we going back to exactly where things were? Are we going to make this a pivot toward certain things, in which case then you're going to have to look at different types of indicators. If we know, for example, that's in the oil and gas sector, the drivers there are very much global in terms of demand and our ability to meet the demand. That's going to have a huge impact on regional dynamics, regional economies and so on.
But there's a lot of talk about "greening" electrification, obviously. And so, policymakers, it always starts with the question and a direction. And then you go in and say, "OK, what are the indicators that are kind of helping me get to that direction?" So if it's about inclusive growth, you're going to have to start to look at different ethnographic variables. If you're in a province like Alberta or Manitoba or Newfoundland where a lot of your income is dependent upon what's happening globally, you're going to have to start to look at what's happening on the geopolitical scale and what's happening globally. If you're in the manufacturing area and you're talking about personal protective equipment and we're trying to simulate certain things, you're going to have to look at a whole different set of indicators. So, yes, it is complicated, but it starts with the direction and what is the question that we're trying to answer, and then searching to say, "What are the best indicators that are going to help you make that decision that you want to drive forward?" Just as a start. I mean, we could have a whole session for this.
NN: Yeah. I know. I know for us, one of the indicators that we focus on a lot has to do with consumer confidence. We measure it every week and it's a leading indicator. It's about six months ahead of the GDP numbers in terms of where things are going. And what's interesting is as part of the composite that builds into that confidence index, one of the elements has to do with the value of real estate in your neighbourhood. You have to remember that for many Canadians, their single largest investment is their home. For those that are homeowners, it's not having stocks and it's not owning a business. It's their home. And when they feel that their home retains value or increases in value of homes in their neighbourhood, they feel good. If they're worried about that being at risk, then they don't feel good. One of the things to watch is real estate.
The interesting thing about real estate is that it also has to do with access. It has to do with expectations. One of the things that we track fairly regularly is we ask Canadians whether they think that the next generation will have a higher standard of living, the same standard of living, or lower standard of living. And over the last number of years when we've tracked this, it's usually by a factor of three to one where people think that the next generation will have a lower standard of living. And this is before the pandemic. I almost cringe to ask the question now because it could very well be much worse. But think of the public policy environment. If our citizens are walking down the street and they're thinking, "My kid is not going to be able to have the same standard of living, or I don't have confidence that they'll have the same standard of living, that they might not live in a house, a similar house. They might not go on the same type of vacation."
And I think if you're a policymaker, that your first step is kind of walk in the shoes of the people that you're designing the policy for, and this goes back to what Anil was talking about, marginalized populations. The data allows you to walk in the feet of people that we're all trying to help.
[Anil nods thoughtfully.]
And these are some of the indicators that I look at, because what they do is they help put a human face on the environment, the policy environment, because we could have the most brilliant public policy. You could have the most brilliant public policy. But if it does not align fundamentally with citizens and if it's not sensitive to the situation that citizens are in, it's not going to work, even if it is brilliant.
[Anil nods.]
TS: We're down to our last couple of minutes, which makes me very sad, but I want to ask you in closing to give us some advice in terms of what are we missing in, let's call it the infrastructure of our data. What do we need to continue to benefit from the data revolution in Canada? Do we just need more data? Do we need more data literacy? Do we need to put more kids through data science at university? What would help strengthen our data infrastructure in Canada as we compete in this world against Israel and Denmark and Singapore and Norway and Taiwan and all these countries that seem to me from a distance to really take data seriously at the state level as a strategic advantage? You want to start us off, Anil?
AA: In 30 seconds or less? So, first—
TS: No, you can go a little bit farther.
[Anil smiles.]
AA: Sure. So, first is I would encourage colleagues to take a few minutes and read the Federal Public Service Data Strategy that I was privileged to co-author with a couple of colleagues. And it actually lays out, at least within the government, let's start here. And it talks about the fact that we're going to need the expertise. We need to build the talent to be able to do this, because this is not just a hobby. You actually have to know what you're doing, as we've discussed many times here. You can actually do some serious damage if you don't know what you're doing.
Two is you have to have the infrastructure. I'm not talking about—people immediately go to the pipes and the bandwidth and so on. I'm talking about the data infrastructure. I'm talking about things like interoperability, standards, definitions, metadata, all those kinds of things. And therein actually has to be responsible use of data. People have to have trust in the government and how we go about using data. So, privacy, confidentiality, transparency: all these are important to build into that infrastructure piece that I'm talking about. We need to govern data properly. And I would say, there's room in that strategy, there's some work and some suggestions there on data governance.
And we have to treat data as an asset. We have to value it. And if we don't value it then we can't measure it, and you can't see the impact of it not being there. Well, then it's just going to be a nebulous commodity. So I think those are the areas and I encourage colleagues to look at. The second thing I would encourage colleagues to look at is the work that we're doing with the Standard Council on the Data Governance Collaborative. It's a wonderful opportunity to put the 300 people from provincial government, private sector government departments that have come together to say, "OK, what business this advantage for Canada look like if we have the right standards in place? How could this actually be a real advantage?" And that is something that is really exciting when you look at what's happening in the world, whether it's the neighbour to the south, what's happening in China, what's happening in Europe. Canada could actually be a leader in this space. So, I'll just stop there just in the brevity of time.
[Taki gestures for Nik to speak.]
NN: Three things super quick. First of all, we cannot compete globally if people are afraid of data. It'd be like saying we're afraid of using computers. We'd be nowhere as an economy. And if we can create a culture where Canadians are not afraid, it doesn't mean they have to be experts, but people are not afraid of data. That would be a big first step. The second big step would be to train Canadians, to train people in the public service to be able to ask the right questions when data is put in front of them. They don't have to have a PhD in stats, but they need to be trained on asking the right questions. I think that would be a massive step forward just to basically make people better consumers and users of data, because they will be asking the right questions.
And then finally, and this touches on what Anil said, I would like to see more transparency because I find many times I see research and there's no transparency in terms of the limitations of the research.
[Anil nods.]
Researchers have an obligation to report the limitations. Who were we not able to reach? What do we have to watch out for? Researchers should not be embarrassed to talk about the limitations of their research. And many times they don't do that. So, you know what? We need a new data culture where we're not afraid of it. We need to have policy analysts trained so that they can ask the right questions, and we need to have more transparency. I think if we can drive those things through—we have a world-class system now, but we can be at the front of the pack.
TS: Gentlemen, as we talked about the beginning data, data, data. And, really, we're just at the beginning. If we think that we're overwhelmed with data now, just wait. We are just on the cusp where your refrigerator will talk to your stove, which will talk to Loblaws, which will talk to Wal-Mart, which will talk to your front door, which will talk to your freezer. We're almost in the world of things talking to other things and throwing off massive amounts of usable data in real time. And the challenges that that will bring forward to public policy is not an insignificant thing. It will probably rival the internet in terms of how important and how kind of transformative it will be, not just for society, but for government and governance and public policy.
So, gentlemen, as two wonderful friends of public policy, Anil, within the public policy apparatus, and Nik, just slightly outside of public policy apparatus, but you come in and out and help us so often with your insights. A wonderful, heartfelt thank you to both of you for spending this hour with us, for helping those on the line understand data a little bit more. And for those of you that are friends of the Public School, these two gentlemen are friends of the Public School and you'll see them again. We're going to give Anil a little break because we've been wearing them out past couple of weeks.
[Anil laughs.]
But they will both be back for as long as they can bear our invitations. So thank you once again, and thank you for being friends of the public service. Bye, guys.
NN: See you later. Bye-bye. See you, Anil. See you, Taki.
AA: Keep up the great work. Yeah, thank you, Nik.
[Nik and Anil's goodbyes echo as the Zoom call fades out. The animated white Canada School of Public Service logo appears on a purple background. Its pages turn, closing it like a book. A maple leaf appears in the middle of the book that also resembles a flag with curvy lines beneath. The government of Canada Wordmark: the word "Canada" with a small Canadian flag waving over the final "a." The screen fades to black.]
