New census technology under close watch as Statcan looks to the future

Looking forward to the results from the long-form census as they come out (short-form provides overview, long-form fills in many details). Makes sense to use existing government data to extent possible (CRA income data should be more reliable than self-reporting):

With just days to go before the very first release of data from the 2016 census, there is an unusual calm outside Marc Hamel’s Statistics Canada office.

A calm before the storm, perhaps.

After all, Wednesday’s release will be watched closely by federal officials, demographers and urban planners — all of whom use the data to help political leaders make myriad decisions that affect the daily lives of Canadians.

This time around, however, some of the keenest observers will be census director Hamel and his staff, watching to see if their new census data-collection methods are hitting their mark.

Statistics Canada has been quietly working on a plan for 2026 to eliminate the mandatory short-form census that goes to every household, instead using existing government databases to conduct a virtual count of the population. The plan, if successful, could mean millions in savings for federal coffers.

The closer the census numbers are to the tests being conducted by Hamel’s team, the more likely that multiple pages of the census questionnaire will be dropped during the next count in 2021, or replaced altogether one day in the future with an electronic count of the population.

This year, for instance, the agency cut two pages about income from the long-form questionnaire and replaced the questions with readily available and, arguably, more reliable Canada Revenue Agency data. Other questions, too, will eventually be replaced with information from existing administrative databases, making it easier to collect the details that comprise the census portrait.

Hamel said the challenge for his staff is to find a way to accurately reflect the Canadian population as it is at any point in time.

“The census as we run it now is very high quality, so anything that we would come up with in the future would have to be as high quality as it is today,” said Hamel.

One particular challenge for an electronic census: address information in various administrative files doesn’t always correspond to where people actually live, making it hard to be confident people are being counted in the right places.

And what about technology?

The majority of Canadians filled out their census questionnaires online, cutting down the time required to input data, and helping to speed up the release of information. Hamel said there might be other technological changes coming for future censuses, but it’s hard to predict what that might entail when census day rolls around again in 2021.

The question that guides planning for the next census and beyond is simple: will this work the same way next time?

“Four years in census terms — for me anyway — it’s short. It’s not a long time to prepare to make sure that we get it right. But at the same time, from a technological point of view, it’s fairly long,” Hamel said.

“It’s always a bit difficult to predict how technology will evolve in a short period of time and how that might have an impact on how the census might be rolled out.”

The questions on the census are also likely to change by 2021, with consultations starting this fall on what things Statistics Canada should and shouldn’t be measuring any more. One question likely to change is about sex and gender, which this year didn’t include a third option for transgender Canadians, Hamel noted.

“Society keeps evolving, so I think that from a census point of view, the census questions and questionnaire should be evolving with it.”

Source: New census technology under close watch as Statcan looks to the future | Toronto Star

Census 2016: Western provinces’ populations are the fastest-growing in Canada – The Globe and Mail

The first batch of Census 2016 results are out from the short-form with basic demographic data. Waiting for the long-form more detailed data how Canadians are doing in relation to economic and social outcomes.

Have excerpted this from the Globe analysis, showing again that population growth continues to be increasingly dependent on immigration:

census_2016__western_provinces_populations_are_the_fastest-growing_in-canada_-_the_globe_and_mail

Roughly two-thirds of the growth in population is due to migration, or the amount by which the number of new immigrants exceed the number of people who leave Canada. The other third comes from what’s known as “natural growth,” the difference between births and deaths. Some countries, such as Germany, Italy and Japan, have already seen the annual number of deaths exceed births, meaning all their growth now depends on migration. Projections show that Canada may reach the point where migration accounts for all population growth around 2050.

Source: Census 2016: Western provinces’ populations are the fastest-growing in Canada – The Globe and Mail

Statistics Canada eyes the end of the short-form census

Other countries do this and makes sense, both from cost and accuracy perspectives. But complex transition:

The mandatory long-form census returned this year, a decade after it was last seen.

If things go as planned, a decade from now the short-form census won’t be seen again.

Statistics Canada is working on a plan for the 2026 census that would eliminate the mandatory short-form census that goes to every household and instead use existing government databases to conduct a virtual count of the population.

The plan would save taxpayers millions of dollars and provide the same information used by governments to plan roads, hospitals, schools and other public services.

Documents obtained by The Canadian Press under the Access to Information Act paint a detailed picture of what officials hope to have in place by 2026: a digital register of every Canadian that could be updated every five years, if not annually, and a smaller long-form questionnaire.

“This approach to replace the short-form questionnaire will require a complete redesign of the long-form questionnaire,” reads the April report provided to former chief statistician Wayne Smith.

The agency said in a statement that it hasn’t yet determined its approach for the 2021 census, but made no direct reference to the 2026 count. The statement said the agency “conducts ongoing research activities to determine the most efficient way of collecting census information.”

Source: Statistics Canada eyes the end of the short-form census – Macleans.ca

Statscan fails to keep pace with seniors’ living arrangements

Valid concerns given the aging of the population:

In the wake of the 2016 census, researchers say they’re increasingly worried about limited data on a key segment of Canada’s booming senior population.
For the latest census distributed in May, Statistics Canada allowed administrators of nursing and retirement homes to complete a short-form census on behalf of residents. The agency also omitted the long-form census for all “collective dwellings,” which include hospitals, work camps and correctional institutions.

The move has irked some seniors and sparked calls from researchers for Statistics Canada to revise the 2021 census delivery as new models of senior living crop up.

“It’s something I’m concerned about with the aging of our population,” said Doug Norris, chief demographer at Environics Analytics and a census expert. “The data on our elderly population needs more attention than it’s gotten.”

Last year marked the first time that Canada had more people aged 65 and over – 16.1 per cent, or 5.8 million Canadians – than those 14 and under.

Without long-form results, Mr. Norris said researchers will lack crucial data about seniors’ income, ethnicity and education, among other findings. The data would be able to pinpoint demographic trends that have health implications, and shape myriad social policies, including seniors’ housing.

“Depending on the research and topic, it could be very important to include that group [of seniors], especially if you were doing anything health-related,” said Mr. Norris, who spent nearly 30 years at Statscan.

In the 2011 census, 378,000 people were counted in nursing and retirement homes classified as collective dwellings. This year, as in 2011, the short-form census – which captures age, gender, marital status and languages spoken – was distributed only to administrators of such homes.

Many questions on the long-form census, such as employment information, would not apply to seniors in nursing and retirement homes, said Geoff Bowlby, Statscan’s director-general of collection and regional services.

http://www.theglobeandmail.com/news/national/statscan-fails-to-keep-pace-with-seniors-living-arrangements/article31305272/ 

Stephen Gordon: The damage the Tories did with the census won’t be easily undone

Stephen Gordon on the possible long-term damage to the Census:

The census is only useful if (approximately) everyone co-operates. The same goes for lots of other things: carpool lanes, anti-littering bylaws and jury duty, to name three. The nature of collective action problems is that it’s never in one’s individual rational interest to take part in the solution; it’s better to simply free ride off the efforts of others. This is why one of the core tasks of government is to enforce participation — and this means imposing penalties for not co-operating.

This is where social capital comes in — or social trust, or social cohesion, or whatever you want to call it. It’s not feasible to governments to micromanage their citizens and enforce their co-operation in their daily activities, even if they wanted to. To a very great extent, the smooth functioning of society relies not on government enforcement, but on people’s willingness to go along with the rules, so long as they believe that everyone else is obeying them as well. Everything depends on a willingness to trust strangers, and to reward their trust in you.

It’s worth dwelling on this point, because one of the most debilitating consequences of the Conservatives’ time in office has been the creation of a constituency for whom the census is now a highly-politicized symbol, instead of being a neutral instrument for good governance. While the government can force co-operation, this isn’t the same as restoring mutual trust.

You can’t expect people to take your concerns seriously if you won’t do the same for them. To the extent that their concerns are about privacy, the most promising way of restoring that lost trust is to demonstrate the extent to which concerns about privacy are taken seriously, and to show some flexibility on the details. For example, questions about religion have been dropped from this year’s census questionnaire.

Social capital is difficult to build, and easy to destroy. The former Conservative government demolished a big chunk of our social capital when it blew up the census, and it will take time and effort to restore it. Posting selfies with census forms can’t hurt, and just might help.

Source: Stephen Gordon: The damage the Tories did with the census won’t be easily undone | National Post

Census needs to reflect modern reality about gender | Toronto Star

I am sure StatsCan is already thinking about this in the context of the 2021 Census and the best means to do so (may just be an “other” category:

After 10 years, the long-form Canadian census is back. Young Canadians, primed by a decade of digital media saturation, flocked online in droves so large we took down the website.

It makes sense — and it’s not just false enthusiasm as we collectively do our duty because “it’s the law.” A generation used to sharing its descriptive statistics online (finding friends, networking, dating) would intuitively understand the benefit of the census. Understanding the sociodemographic landscape helps us know and better service ourselves. And after all, that’s what millennials want: a fairer and more representative social democracy.

Yet, as Canadians fill out the census, some gawk at the glaring anachronism of the gender binary, the idea that there are two mutually exclusive genders: males and females, who occupy distinct cultural, social, and sexual roles.

But we know this isn’t true. The recent media awakening to transgender people (Laverne Cox, Caitlyn Jenner, Jazz Jennings) is evidence that gender variance has gone mainstream.

If we recognize men and women who identify with the genders they were assigned at birth (cisgender) and we recognize men and women who do not identify with their assigned gender (transgender), then surely we agree this difference is worth recording.

As my friend quipped, “Well, they’re not asking about gender. They’re asking about sex!” His point reflects the growing awareness about gender as the patterns of behaviour and expression associated with its respective sex categories. This is good. It shows a recognition of people whose self-concepts do not match the gender assigned them at birth.

…Despite a variety of new ways to capture gender variation in the population, this simple two-step approach takes us miles further than the two-option approach of the 2016 Census:

  1. Do you identify with the gender you were assigned at birth? Yes / No / Not sure / Prefer not to say
  2. Please indicate your current gender: Male, Female, Non-Binary, Intersex, Other (please describe):

As the 2016 census has done with its categories for race, we must open up how we assess gender. I know it seems hard, but let’s no longer pretend we cannot do better.

Source: Census needs to reflect modern reality about gender | Toronto Star

2016 census drops income and benefits over faulty data given data to come from CRA

Good change and use of existing and more accurate data:

When Canadians receive their census questionnaires this May, they’ll no longer be asked to report their income and benefits — something Statistics Canada says produced subpar data.

“To substantially reduce the burden on Canadians, and improve the quality of income data compared to previous censuses, Statistics Canada will use income and benefits data from the Canada Revenue Agency for all census respondents to replace questions previously asked on the 2011 National Household Survey questionnaire,” a recently-published order-in-council explained.

Aside from the return of the mandatory component of the long-form census, which Minister of Innovation, Science and Economic Development Navdeep Bains announced the day after being sworn in last November, the 2016 Census of Population will essentially mirror the 2011 National Household Survey.

“There are no new questions on the short or long form. To ensure comparability over time, with the exception of two changes, questions asked by the Census of Population will remain the same as they were in 2011,” a Statistics Canada agency spokesperson told iPolitics.

“First, the question on religion will not be included as the census program has asked this question only every 10 years since 1871. Second, in order to reduce the time required and make it easier for Canadians to respond, income questions will be replaced with more precise tax and benefit data that have been available to Statistics Canada since 1985.”

The latter change is welcomed by Philip Cross, formerly the chief economic analyst at the agency.

When asked about it, Cross referred to a paper he wrote with Munir Sheikh — the head of Statistics Canada who resigned in the wake of the Harper government’s decision end the long-form census in 2010.

That paper, published by the University of Calgary school of public policy last March, attempted to assess the extent of the middle class plight dominating the Canadian political discourse.

And one problem it highlighted was the “disquieting” difference between what people reported as income when surveyed, as in the census, and the tax data reported by the Canada Revenue Agency.

In a nutshell, Canadians were underestimating their income.

“One reason households routinely underestimate their income in surveys is they respond as if only wages and salaries are income, ignoring the growing importance of supplementary benefits such as employer contributions to pensions or health care that are included in taxable benefits,” Cross and Sheikh wrote.

Supplementary income, they added, had risen to over 13 per cent of all labour income.

“Most of these benefits accrue to middle-income earners, something that should be taken account of when examining how their real income has fared in survey data. As well, surveys exclude irregular sources of income, such as bonuses or stock options,” they wrote.

“Income tax data are less timely but more complete.”

Will miss the religion question but it has always been on a 10-year cycle.

You needn’t talk money anymore

New system to release census data faces uncertain future over delays

An admirable effort to make the Census easier to sort through runs into technical problems:

An $18-million project to make it easier to sort through reams of data from the coming census has been beset by delays and uncertainty that the three-year project will be done on time.

Called the “new dissemination model,” the project is designed to make it easier for visitors to the Statistics Canada website to organize, read and play with the data statistical agency collects, be it census or jobs data, or anything else the agency measures.

The end idea is to create a more interactive experience online instead of pages of static data tables, and also to simplify and standardize how information is presented.

It was all supposed to be ready in time for February 2017 when Statistics Canada releases its findings from this year’s census.

Statistics Canada and Shared Services Canada, the government’s central information-technology department that is building the new system, said the project has been delayed, but couldn’t say by how long or if it could still be completed on time.

Shared Services Canada said it has faced “a number of challenges” hosting the new system in its data centres that it is trying to address.

Internal government documents show there was a “final go/no-go” test on the system in December 2015. Statistics Canada hasn’t made a decision on the project following the test, the details of which neither agency would disclose, and is “currently analyzing the impact of the delay” to see what the next step will be.

The project is a microcosm of the problems auditor general Michael Ferguson raised last Tuesday in a critical review of Shared Services Canada. That audit found, among other things, that Shared Services Canada didn’t always communicate well with the departments and agencies it serves, leaving some of them in the dark about projects, and confusion over who was responsible for what.

As this will likely be more for general users, expect that I will continue to use the Beyond 2020 specialized software rather than this tool once (or if) it becomes available. My only wish is for Beyond 2020 to have a Mac version rather than having to run Windows.

I have been invited to a Stats Canada usability testing session this week which will give me a better sense of the planned approach.

Source: New system to release census data faces uncertain future over delays

StatsCan to unveil new ‘efficient’ long-form census for 2016

Good example of the public service doing its job and preparing for a possible change:

When the Liberals were sworn into office in November, one of their first orders of business was to announce the reinstatement of the long-form census.

The timeline seemed very tight — the first forms are to go out to residents in the North in February.

But Marc Hamel, the census program director general, says the agency had planned for risks associated with the 2016 census. One of those risks was if a new government decided to bring back the long questionnaire.

“It had already been in the public sphere that opposition parties last year were saying, if they were elected, they would bring back the mandatory long-form census, so we had started to look at how that would be possible,” Hamel said in an interview.

The agency decided to design the questionnaire in a more adaptable format.

Rather than sending selected households separate pieces of mail with the short form and then the National Household Survey, the questionnaires were integrated into one document.

“That design was going to be efficient and it was going to work for both approaches,” said Hamel. “From that perspective, no redesign was required. We were simply able to move ahead with the same questionnaires that we had already designed for 2016.”

Also, because most Canadians fill out the census online — 64 per cent in 2011 — changing details in a computer system was not a major overhaul.

The letter that accompanies the questionnaires will allow the agency to underline that the long part is mandatory again. Census staff will also drive home the message.

Fewer people will have to fill out the long form than last time, one in four households rather than one in three with the NHS. Statistics Canada has had to print more short-form questionnaires as a result of the change.

The agency doesn’t think it will save money with fewer people getting the bigger package. It expects it will have more responses to process because of the return to the mandatory format.

The main challenge will come from adjusting to the data logistics of bringing back the long-form census. Bar codes help the agency keep track of where they drop off which forms and some of that work will have to be rejigged.

There will also be a public awareness campaign to make sure that people realize they need to fill out the forms. Hamel says the agency never really emphasizes the penalties associated with not filling out the forms — a $500 fine or up to three months in jail, or both.

“Census information is really important, and that’s where we put the focus,” said Hamel.

“What do we use the census information for, why is it important for communities, and why is it important for people to participate.”

Source: StatsCan to unveil new ‘efficient’ long-form census for 2016 – The Globe and Mail

And one of the new challenges:

Quinn Nelson wants to be counted in the 2016 long-form census, but when it comes to the question of gender identity there’s a problem: Nelson is transgender and identifies as neither male nor female.

“As a non-binary person, often when I fill out forms there’s only two options given to me and that’s not enough for me,” Nelson said in an interview on CBC’s Power & Politics.

In November, Nelson wrote an email to Navdeep Bains, the minister responsible for Statistics Canada and the census. Nelson didn’t want to violate the law by not filling out the questionnaire.

The University of Calgary sociology student also wanted to make sure Statistics Canada was going to provide an accurate reflection of the country.

The census assumes that 100 per cent of the respondents can answer that they are either male or female, “and that’s not accurate,” Nelson said.

“The census is used by a lot of policy makers, sociologists and government officials to make decisions. They really need to know what their population is. That’s the point of the census.”

….Bains hasn’t responded to Nelson, but Statistics Canada did. Deputy chief statistician Connie Graziadei said the 2016 census questionnaire had already been approved and published, but there is an option for Nelson.

“I was told to answer neither, to leave the question blank; also to answer in the comments why I found the question inadequate.”

Transgender student says some Canadians need 3rd option for gender on census

Liberals will find key to undoing Harper’s agenda in his infamous ‘firewall’ letter | Ottawa Citizen

A good in-depth and must read piece by Andrew Potter on how the ‘firewall’ letter was implemented from Ottawa, and the tactics behind implementation of the ideology:

Data: It wasn’t privacy, as Tony Clement said, or freedom, as Max Bernier argued, that was the real rationale for killing the mandatory long-form census. It was to throw a whole lot of noise into the demographic signal that the census had been giving for decades. That is also why Statistics Canada as a whole was gutted over the course of the Harper years. Without accurate data, social planners are flying blind.

Expertise: No government in living memory has been as hostile to experts and to evidence as the Harper government. But as Laval economist Stephen Gordon recently argued, it wasn’t all forms of expertise and evidence that gave the Tories hives – plenty of their economic initiatives were rooted in the best available evidence. What the Tories were allergic to was expertise that steered the evidence in directions they didn’t want to go – “committing sociology,” in Harper’s wonderful turn of phrase. That is why scientists were muzzled, policy shops were shuttered and bureaucrats were ignored.

Money: Here is the meat in the sandwich. When it comes to social planning, the ultimate source of Ottawa’s power is the spending power. And this is where Harper had his greatest success. By the end of his tenure as prime minister, Ottawa’s spending, as a share of GDP, had fallen to levels not seen since the middle of the 20th century. And the spending that does remain is overwhelmingly devoted to either just keeping the lights on or takes the form of transfers to the provinces and individuals.

Harper’s policy genius here was the two-point cut in the GST, which currently costs the federal treasury about $12 billion a year. Harper’s political genius was the creation of an all-party and pan-Canadian consensus around the virtues of a balanced budget at that historically low-level of federal spending.

No data, no experts and no money. Starve the beast, but make it blind and deaf at the same time. This is Harper’s “Ottawa Firewall” in a nutshell.

‘Flat-tire federalism’

As long as Harper was in power, this firewall against centralized social planning was bound to be highly effective. The question is, what remains of this agenda with a Liberal majority in power in Ottawa?

The long-form mandatory census is back, just under the wire. Another missed census in 2016 would have gummed up the data for generations, but as it stands, it looks like the 2011 asterisk will remain just that.

The scientists have already been unmuzzled. The public servants have been asked for their advice. The policy shops are staffing up and stocking the shelves and will be open for business soon.

But what about the money? This is where things get tricky for the Liberals. Their commitment to running three relatively small deficits to build infrastructure and kick-start growth caught everyone in the chattering classes off guard, and turned out to be a political winner.

But the promise was to return to balance by the last year of their mandate. That is, they accepted the basic premise of balanced budgets at more or less current levels of federal revenues (their tax plan calls for additional revenues of just $3 billion). This isn’t nearly enough, and there is not enough economic good weather in the offing for Ottawa to grow its way to good times.

An Ottawa with lots of data and lots of policy ambitions but no money is going to be pretty ineffectual.  At some point, the Liberals are going to have to tackle the revenue problem. Without money, without the fiscal capacity to get things done, all the data and expertise and policy advice is just squiggles on a page and vibrations of air molecules.

A federal government that is nicer, less controlling, more transparent but still broke is not one that has much capacity to bother the provinces with socialist schemes. And if that’s where things remain, then Harper’s long-term victory will be cemented, regardless of who is in power.

Source: Liberals will find key to undoing Harper’s agenda in his infamous ‘firewall’ letter | Ottawa Citizen