We can do better than the long form census

Posted on February 17, 2015 in Inclusion Debates

OttawaCitizen.com – News/National
February 16, 2015.   Martin Cooke

In Private Member’s Bill 628, Liberal MP Ted Hsu proposed to bring back the long form census. I appreciate any effort to strengthen Statistics Canada but, now that this bill has failed, we should be developing the statistics system of the future and stop looking back to the system as it was in 2006.

I was among the academics protesting the 2010 decision to scrap the mandatory census long form in favour of the voluntary National Household Survey. As we said at the time, the census is a critical tool for policy research. It gives us the only complete count of Canadians and where they live and some important social and economic indicators. The 20 per cent of households that got the long form were enough to give accurate estimates down to a few city blocks. The census was also a central pillar of a large national statistics system, as is the Labour Force Survey (which remains mandatory). It facilitated many voluntary surveys on health, crime and victimization, retirement, consumer finances, and other important topics.

That old system worked well. A friend at the Australian Bureau of Statistics once admiringly referred to Statistics Canada as “the world’s premier national statistical agency”– high praise indeed. It wasn’t perfect, though. Every year the census failed to enumerate a proportion of the population, disproportionately the poor, young or Aboriginal. The system relied mainly on self-reporting and respondent recall, and surveys are susceptible to the same problems of falling response that bedevil political pollsters.

The truth is that Canada was good at surveys and censuses partly because we lacked what some other countries had: a system based on administrative data. As difficult as it might be for bureaucracy-burdened Canadians to believe, there is no central database that records the major events in our lives. Rather, we have separate systems maintained by federal, provincial and sometimes municipal agencies containing disconnected data such as our health records, tax returns, crime reports, land titles, immigration records, marriage and divorce certificates, and (eventually), death certificates. This is unlike Belgium or Denmark, for example, which have had centralized registers of population since the 1960s. The Netherlands did away with its traditional census in 1971, and has since saved billions in enumerators and processing costs by using already existing administrative data to produce a “virtual census.”

Privacy is a key reason that we have kept data collected by these administrative systems separate. In the same way that Canadians (and Americans, Australians and Brits) haven’t historically liked the idea of national identity cards, we haven’t cared for the idea of having all of our data connected and in one place.

This is changing. For good or ill, we are becoming more and more comfortable with sharing our data, as we leave traces of our activities and identities across the Internet. We trade some privacy for the convenience of having Amazon remember our credit card numbers, or recommending the right book on our next visit. We might also be happy to avoid answering the same questions repeatedly, if we are assured the data will be used appropriately.

Elements of a new system based on “data linkage” are already being developed. Projects across the country are connecting health, tax and vital statistics records to other sources of data including the census and surveys, usually for a limited number of people or years, and under the tight control of privacy commissioners and ethics bodies. We are getting better at linking data that were not designed to be linked, and developing the technical, legal and administrative tools to safeguard privacy while facilitating research and access by government and civil society groups.

The national statistics system is important to our economic and social well being, and any changes should be very carefully considered. The 2010 changes have been harmful but, rather than trying to recreate the old system, I would rather see us think about what a new system might look like, and to ensure that it meets the needs and expectations of Canadians.

Martin Cooke is an associate professor at the University of Waterloo and Co-Director, Waterloo Survey Research Centre.

< http://ottawacitizen.com/news/national/we-can-do-better-long-form-census >

Tags: , ,

This entry was posted on Tuesday, February 17th, 2015 at 8:47 am and is filed under Inclusion Debates. You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

Leave a Reply