Back in July, when the kerfuffle over the long form census was fresh, I accused the Harper government of being disengenuous in their claims that changing the long-form census from mandatory to voluntary was due to privacy concerns over the invasiveness of the census form.
Recent permanent appointment of new Chief Statistician of Canada Wayne R. Smith (necessary due to the 2010 resignation of Chief Statistician Munir Sheikh over the census issue) and subsequent media coverage of Smith’s approach pretty much confirm that view.
Apparently Harper has asked Smith to “rethink” the census for 2016, specifically directing the Chief Statistician to look into how other countries conduct censuses. (Oh, you mean like this freely-available Sept 2010 article in Canadian Public Policy did?) The buzz that is arising from this directive is all about register-based methods that avoid mailing surveys to individuals at all. And this reinforces my assertion that the claims of concern over privacy are bogus.
Register-based censuses link administrative databases so that an individual’s profile can basically be mined for the answers to questions that would have been asked on a paper census. This is exciting for researchers and statisticians, because these registries are generally very accurate. However, for privacy advocates, giving the federal government carte blanche to link and mine federal databases is a matter of concern.
The Online Party of Canada is hosting a discussion on this topic in a forum on their website. The original post notes:
At first glance, it would seem that most of the vital information collected by the Canadian long-form census questionnaire (2B) is already being collected, at least in part, via other governmental sources (i.e. federal, provincial, municipal, etc.):
● Age (Q1), gender (Q2), date of birth (Q3) and most labour market activities (Q34-46) information can all be linked to our Social Insurance Number (SIN) and/or Canada Revenue Agency (CRA) data.
● Citizenship (Q10-11) via Citizenship and Immigration Canada;
● Place of Birth (Q9) and parents’ POB (Q25) via Provincial Birth Certificates or country of origin under the Immigration Act;
● Mobility (Q23-24) via Canada Post (theoretically);
● Education (Q26-32) via Provincial Departments of Education, School Boards and Post-Secondary Institutions;
● Only a few questions related to: Activities of daily living (Q7-8), Household Activities (Q33), Language (Q13-16), Commuting (Q47) and Dwelling (QH1-H8) could not presently be answered from other sources.
But then asks:
How would Canadians feel about this alternative?
No more census. Save millions in tax dollars. More accurate data.
But our personal information would have to be linked and completely centralized. What types of mechanisms would we need to implement in order to insure confidentiality and protect ourselves from government misuse or abuse of data?
Now I’ve never heard of the OPC before, but these questions are right at the heart of the matter. The only thing I can think of that they leave out is the question of whether inconvenient questions – questions not represented in official data and also those asking things the current government might not wish to collect – things such as self-reported ethnic identity or unpaid work hours – would be scrapped under a register-based census plan. Given that the unpaid work hours question was already scrapped (much to the chagrin of feminists), I expect this is not a large concern of the current government.
So, How do we feel? How would you feel about your medical records being linked to your tax filing and your hydro bills and your school records and your motor vehicle records and your vital stats records and your address as listed with Canada Post and…a bunch of other things? How would you feel about a law requiring you to register all address changes with the government? Denmark astutely points out that registering your address with one central body is highly efficient. But do we trust our government the way Danes trust theirs?
The researcher in me loves the idea of the high-quality data we could get with a registry-based system. (The researcher in me is also skeptical that Canada could pull together anything like that for 2016!) However, there would have to be some serious safeguards (including updated ethical review and data stewardship processes) put in place for the privacy advocate in me to feel comfortable with it. I’d also have to feel assured that the increased data infrastructure would be available for non-governmental researchers as well as internal government use.
Last year Harper supposedly changed the census due to its invasiveness. But that’s not the right word. A lot of us were using the wrong word, because it appears that perhaps he’d like to entertain the idea of a more privacy-invasive process, as long as it would be less intrusive into our lives. No pesky, visible forms taking up our time. No census takers knocking on our doors, asking us annoying questions, making sure the population is well aware of what info the census is aiming to collect. Instead just a quiet government data-mining operation. Invasive (possibly more invasive than now), but not so intrusive.
ps – Incidentally, if you’re interested in reading it, the Globe & Mail has published a transcript of their whole interview with Mr. Smith, and it’s pretty enlightening. For example, until reading that I had no idea StatsCan is assuming there would be only a 50% response rate to the voluntary survey that replaced the long-form census. In this fascinating read, Smith says amazing things, including:
“The one thing we know with absolutely certainty is the response rate going to fall from making the survey voluntary.”
“But there is no guarantee this data will not be usable. There is no guarantee it will be subject to major non-response bias beyond the levels we’ve traditionally seen in the census.”
He also denies that there is *any scientific reason* to expect non-response bias from groups such as Inuit, non-English speakers, or immigrants.