Publications, opinions, and speeches


Counting on data: the 2016 long-form census is just the beginning

Published on 26/05/2016

For a brief moment this May, we broke the internet. In true Canadian fashion the cultural phenomenon that sent us all to our screens in synch was not a Kardashian or even a new Drake single but the arrival of our individual access codes for the 2016 census. So many Canadians rushed to fill out the questionnaire immediately that Statistics Canada’s site briefly buckled under the pressure.

The real driver behind the stampede was the collective enthusiasm around the return of the mandatory long-form census. While the standard census does little more than count Canadians and collect a few vital statistics, the long-form goes far deeper, asking one fifth of households a longer list of questions covering how they live, work and move around.

In 2011, the long-form had been replaced by a voluntary National Household Survey, to the disappointment of just about everyone. For good reason: the lower response rate left significant gaps in data that are necessary to governments, the voluntary sector and businesses to make decisions. The return of the mandatory long-form was a signature commitment of the new government. And if social media is to be believed, its return was celebrated like something out of Charlie and the Chocolate Factory, as Canadians ripped open their access code letters hoping to find their golden ticket to the long-form version of the census.

The enthusiasm comes as part of a broader call for “evidence-based policy,” which includes commitments from governments across Canada. The federal government has given improving the quality, use and availability of evidence and data priority placement in Ministers’ mandate letters. The Ontario government has created a Centre of Excellence for Evidence-Based Decision-Making Support. Cities across Canada are creating open data platforms to improve services and transparency for residents.

This commitment to making better use of the evidence that we do have is welcome and necessary. But to make these ambitious visions possible, we need to treat the return of the long-form census as the starting point, not the finish line, of reinvesting in the information necessary to make better decisions about public policy.

The reinstatement of the long-form census is certainly important. The low and inconsistent response rates left us with poor data that underrepresented lower-income people and small to mid-sized communities. While there is no getting back the missing year, the return of the long-form census will help us to correct for that missing information, in time.

The return of the long-form census does not however come close to restoring the pieces of the data puzzle that we have lost in the last decade. Since the cancellation of the Survey of Labour and Income Dynamics, we know less about what is happening in the lives of low-income Canadians. The closing of the National Council on Welfare would have left us without an essential tool to track how income supports work, if the Caledon Institute had not crowdfunded a “data rescue” campaign to save it with the Canada Social Report. The cancelled National Longitudinal Study on Children and Youth could not be saved in the same way.

In addition to lost surveys and data sets, we have been losing ground in other areas. We don’t know enough about self-employed and contract workers. We don’t have clear and consistent information on what is happening in labour markets. We don’t collect the data needed to understand whether people’s rights are being systematically infringed on. And we have a long way to go before we can say we are taking advantage of the opportunities presented by new types of data collection and analysis, such as big data analytics.

To really improve the way we use evidence to inform policy decisions – and to improve the ability of Canadians to get information about our country – we will need to take steps that are more transformative than adding a few surveys to the sometimes frustration-inducing archives of Statscan’s CANSIM system. We should be thinking about how to overcome the siloes that see data about education, housing and labour markets kept in different places.

We could be making better use out of administrative data often buried in not only tax records but the federal, provincial and local bodies that deliver Employment Insurance, social assistance, housing and employment supports. These could give us a better idea of what is happening in people’s lives and whether our policies and programs are working.

As a bigger idea, we think that it’s time for a Canadian Institute for Social Information – an arms-length institution that could connect and build on existing efforts just as the Canadian Institute for Health Information has done for health care since 1994. A new platform like this could build into its organizational fabric commitments to modern practices such as meeting gold standards on open data and access, and making use of tools to visualize and interpret the information to make it useful for researchers, policymakers, media and the public. To make effective evidence-based policy possible, it is not enough for the data to exist on a server somewhere, it needs to be curated to support the practical needs of users in different sectors and roles.

Statistics Canada does some excellent work. However, its mandate leaves it providing us a better picture of Canada’s 19th and 20th century economies than it does about our 21st century needs. We have consistent data spanning decades about Canadian beekeeping. And chickpeas. And stocks of frozen and chilled meats (really).

If we could only have half as much information about what is happening to Canadian families as we do about what is happening on Canadian farms, we would be in much better position to reduce poverty and increase opportunity. To get there we will need to build on the enthusiasm for the census to put in place a wider effort to work across and between governments to make available the evidence needed to make better decisions.


Evidence-based policy


If we want evidence-based policy, we need much more than the census.