Canada’s 2011 long-form census fiasco was brought to the fore again last week with the introduction of Bill C-626. The private member’s bill seeks to amend the Statistics Act to mandate the long-form survey and provide the Chief Statistician with greater administrative autonomy. While there’s no debating its voluntary replacement rendered useless results at exorbitant cost, the proposed legislation, even if successful, won’t necessarily assure the long-form census’ future success.
It’s been written those who cannot remember the past are condemned to repeat it. Although never mentioned in the ongoing public discourse, what’s happening with the long-form census is basically the Survey of Family Expenditures (FAMEX) all over again. And that didn’t turn out well.
A note before proceeding: Contrary to popular media hero-villain narratives, in interactions between the federal government and its bureaucracy, rarely is one solely responsible for poor decisions or outcomes. The demise of FAMEX and the long-form census demonstrate as much.
As the federal Privacy Commissioner’s Annual Report to Parliament 1997-1998 notes, “regular FAMEX surveys began in 1952.” Despite appearing to contradict the plain language, Statscan advances that Section 7 of the Statistics Act (along with Section 24 (2) of the Interpretation Act) can be interpreted to mean “responding to all Statscan surveys is compulsory unless the Minister decides otherwise.” On this basis, response to the FAMEX, along with the long-form census, was deemed compulsory.
Statscan decided to delay FAMEX in 1979, in part because it “had caused householders to complain to MPs about the lengthy questioning and frequency of Statistics Canada surveys.” (Statscan accused of delaying poll to help Liberals, The Globe and Mail – November 3, 1979) Participation in FAMEX became voluntary starting with the 1984 survey – taken shortly after Canada elected a Progressive Conservative government.
The Industry Minister for the current federal Conservative government, first elected in 2006, claimed he received that year 1,000 messages from Canadians upset with the 2006 long-form census. In 2010, his successor decided to act. Participation in the renamed long-form survey became voluntary starting in 2011.
The FAMEX “reverted to compulsory in 1996 when Statscan felt that the response rate, which had dropped to 74 per cent, was too low.”
The voluntary 2008 long-form census test conducted in preparation for the 2011 survey had only a 45 percent response rate. While Statscan reported the 2011 National Household Survey (NHS) had a 69 percent response rate, there were claims of rather questionable practices employed to boost the rate (which the agency never denied) – and, along with it, public confidence.
Either way, the voluntary 2011 NHS response rate was lower than the ‘all-time low’ voluntary 1992 FAMEX.
Reverting FAMEX to compulsory in 1996 went anything but smoothly. The questionnaire was 39 pages long, took more than 3 hours to complete and was characterised as ‘grossly intrusive’. Statscan insisted the survey be administered in-person; respondents were asked for details about their “…sanitary and incontinence supplies…” and “condoms, syringes, etc.” usage [notably, these highly personal questions were not part of the last mandatory (1982) FAMEX]; and the enumerators were external contractors with only basic security checks.
What irked many was that Statscan intended to sell the information it was compelling respondents to provide: “In its promotional material to business, Statscan extols its ability to tailor the data from FAMEX “to meet your specific needs.””
“The (Privacy) Commissioner concluded… (Statscan) had not been frank with respondents about why it was collecting the information and how it would be used. On that basis he considered the complaint well-founded.” Unfortunately, he accepted the agency’s offer to allow for self-enumeration as sufficient redress and neglected to address other concerns raised.
After Statscan initially threatened 1996 FAMEX refuseniks with prosecution, and with it possible imprisonment, the (then Liberal) federal Industry Minister at the time intervened to assure no one would be going to jail. (StatsCan survey violates privacy, commissioner says Agency threatened jail for man who refused to answer, The Toronto Star – July 10, 1998) In 1997, the FAMEX was discontinued and replaced by the much less detailed – and voluntary – Survey of Household Spending (SHS). The lack of detail, along with lower response rates, rendered SHS far inferior to FAMEX.
The campaign to boycott the 2006 long-form census began in earnest in 2003 when social advocacy groups learned of Statscan’s plan to contract out part of the survey processing to US defense contractor Lockheed Martin. This was shortly after the USA PATRIOT Act, which expanded access by the FBI to information held by US companies, was passed.
Without commenting on the merits of either the Statscan rationale, the (then Liberal) federal Industry Minister’s decision, the contracting process, social advocates’ concerns or the agency’s response to them, it suffices to say the timing and context of the Lockheed Martin contract did not inspire public confidence.
Seeming to have learned little from its FAMEX experience a decade earlier, Statscan management proceeded once again with a high-handed approach with the census. It referred 64 Canadians who refused to respond to the 2006 survey to federal prosecutors. (It subsequently referred 54 who refused to respond to the 2011 survey, which only included a short form.)
Unlike a decade earlier with FAMEX, the Privacy Commissioner did not intervene when it came to the census. Rather, her office could barely find a complaint on file – emphasis on ‘find’. [Turnover at the OPC at the time resulted in the questionable handling of complaints, many ending up either lost or closed without notice.] While the federal Finance minister’s ‘1,000 messages’ claim may have seemed questionable, in light of the number of Canadians threatened with prosecution, the Privacy Commissioner’s claim seemed at least as suspect.
Also, unlike a decade earlier with FAMEX, the federal Industry minister chose not to intervene to assure 2006 census refuseniks would not go to jail. Some have suggested that allowing prosecutions to proceed served his government’s interest in advancing the cause for eliminating the compulsory long-form.
As mentioned, those who cannot remember the past are condemned to repeat it. In the case of the federal government and its bureaucracy, questionable competence, poor judgement and a tendency toward inertia likely contributed to the long-form census’ demise as much as faulty memory.
As to what happened to the survey that replaced FAMEX, despite the shorter form, SHS response rates continued to decline, to the point Statscan stopped making the data publicly available in 2010. That ‘all-time low’ 74 percent response to the 1992 FAMEX now seems stellar when contrasted with the 43 percent response to the 2012 SHS (latest info available).
While much has been made of the importance and significance of the long-form census, results of the discontinued FAMEX and its replacement SHS were/are used to compile non-trivial economic indicators like the CPI and GDP. If the 2011 NHS response rate rendered it unreliable – ‘useless’ in the view of most economists – then what’s to be made of the 2012 SHS?
What Bill C-626 would effectively do is set the clock back to 2006, when circumstances already had the long-form census on course to cancellation. A great deal has happened since then to further undermine public confidence in the federal government and its bureaucracy. Just about every piece of ‘anti-‘ federal legislation, under the guise of protecting Canadians from child predators, online bullies, terrorists, etc., involves expanding government surveillance and further encroaching on Canadians’ civil liberties.
In preparation for the 2016 NHS, Statscan proceeded to test Social Insurance Number collection without specific authorisation from Treasury Board, without review by the current Privacy Commissioner, and contrary to advice it received from previous Privacy Commissioners – likely further undermining public confidence and trust.
To paraphrase a comment seen often in response to articles supporting reinstatement of the long-form census: “Why should I volunteer personal information to a government that increasingly collects and uses it without my consent?” That increasing distrust isn’t just undermining support of the long-form census; it has negatively impacted response rates for many critical economic surveys. And it’s not just a Canadian problem.
There’s a point beyond which data can’t reasonably be adjusted for non-response using statistical methods. The SHS, introduced in the aftermath of the failed effort to revert FAMEX to a compulsory survey, is beyond that point. Given the FAMEX’s eerie parallels to the long-form census, it’s far from certain that simply mandating compliance and giving greater autonomy to Statscan will result in successful reinstatement of the long-form census.