National Statistics Council on the Census

The following statement was released this morning. The key point is that the mandatory long form census should remain for 2011, and some changes are proposed moving forward including removing the never-used penalty of prison for non compliance.

Seeking Solutions

The National Statistics Council, the senior, external advisory group appointed by the government of Canada to advise the Chief Statistician, is deeply concerned by the effect of the announced changes to the 2011 Census. We believe that the changes will harm the integrity and quality of the Canadian statistical system. At the same time, the Council recognizes that concerns about intrusiveness and confidentiality should be addressed.

It is urgent we find solutions that protect the quality of the information Canadians depend upon while responding to concerns over the way in which the Census is conducted.
What is at risk?

First, the proposed, voluntary National Household Survey will suffer from significant respondent self-selection bias. Even with substantial efforts to mitigate the inevitable decline in response rates, this will degrade the data upon which much of the Canadian statistical system is based.

The proposed changes will likely result in Statistics Canada’s not being able to publish robust, detailed information for neighbourhoods, towns or rural areas. Much of the analytic work done by municipalities, private firms, health agencies, highway and transportation planners, school boards and large numbers of other groups that depend upon small-area knowledge and data will no longer be possible.

Our second concern is the potential loss of vital benchmark information. The mandatory ‘long form’ means that Statistics Canada has an accurate benchmark for the demographics of populations who are difficult to reach or who are less likely to complete a voluntary survey. This, in turn, means that sampling and weighting strategies for subsequent, voluntary surveys can compensate for differential response rates and produce more reliable information.

The importance of having Census benchmarks available is readily apparent when one considers some of the populations that we know are more difficult to reach – young people making the school-to-work transition, urban Aboriginal populations, the affluent, and new immigrants.

Without solid benchmark information, subsequent surveys and analysis rest on an uncertain foundation. The Bank of Canada cautiously stated that, while they do not use long-form data directly, they feel they will have to evaluate “the impact that any proposed change would have on the reliability and the quality” of economic data they use. The Bank’s statement exemplifies the repercussions the changes may have over the broader Canadian statistical system.

The National Statistics Council also recognizes the concern that Canadians not be overburdened by governments compelling them to respond to onerous or intrusive demands for unnecessary information. On a number of occasions, the National Statistics Council has urged changes and worked with Statistics Canada to reduce such respondent burden. With respect to the Census, the Council has strongly supported changes to data collection methods that enhance privacy such as mail-in and on-line options.

In addition, the Council strongly supports Statistics Canada’s commitment to the complete confidentiality of respondent information and it recognizes the agency’s undisputed success in reaching this goal. The Council shares the Privacy Commissioner’s Office view that Census questions are ‘inherently privacy-invasive’ and that the questions must be kept to what is necessary for good government and that the information gathered must be protected with the appropriate safeguards.

This focus on minimizing intrusiveness and protecting privacy is important to retaining the confidence of Canadians. We are satisfied that Canadians trust Statistics Canada and its procedures and that Canadians provide answers they would be unwilling to provide to a private survey firm. The Council also believes that confidence must be sustained through ongoing actions.

In a matter of a very few weeks at most, it will be impossible to change the 2011 Census or the National Household Survey. Meanwhile, debate over the future course of the Census has become heated without moving towards a resolution that meets both concerns about privacy and intrusiveness, as well as the need to maintain the quality of Canada’s statistical system. What then do we recommend?

The National Statistics Council recommends:

1.That, as part of a formal consultation process beginning with the 2016 Census, Statistics Canada examine each Census question to ensure that it, at a minimum, meets one of the following tests for inclusion in the Census:
a.It is required by legislation or Cabinet direction,
b.It is needed for small-area data uses for which there is no alternative data source,
c.It is needed to create benchmarks for measuring difficult-to-reach groups and ensuring that subsequent surveys or data derived from administrative sources can be sampled or weighted to reflect accurately the overall population,
d.It is needed to assess progress on issues of national importance, for example the economic integration of new immigrants, or
e.It is to be used as a basis for post-censal survey sampling of relatively small or dispersed groups, for example, urban Aboriginals or people with health conditions that limit their activity.
Even if a question met this requirement, it would still face tests of its overall importance to the Canadian statistical system and the needs of data users as weighed against cost and the intrusiveness of the question.

2.The Council is aware that other countries have conducted successful censuses without people having to face the potential of jail as a punishment for not filling out census forms. We, therefore, recommend that the Statistics Act be amended to remove jail sentences as a possible punishment for not filling out the Census. At the same time, the Council recommends that jail continue to be a punishment for those who wilfully break confidentiality provisions.

3.That the Census for 2011 include the long form being used for 20% of the population as the only way, given the very short timeframe, to safeguard the quality of the Canadian statistical system.

4.That the question series on household activities (question 33 in the 2006 long-form Census) be dropped as it was the question that occasioned the largest number of objections among the substantive questions and since it fails to meet any of the five tests outlined in point 1.

The National Statistics Council believes that these steps, taken together, can respect the valid concerns voiced by Canadians about privacy and intrusiveness, while ensuring that the vital information that currently flows from the long-form Census can be maintained and continues to serve Canadians’ needs.

Ian McKinnon,

Chair, The National Statistics Council


  • Seems like a rational course of action. The trouble is, I am not so surer the other side is wanting to be rational, purely ideological. I wonder- Harper has let this drag on for so long now and the vote damage he has incurred is probably peaked- so my guess is he will not back down.

    Makes you wonder if the country is some kind of test bed for the ultra right and the nature of governing within a permanent recessionary future of modernity that they must be planning out.

    No numbers- minimize the backlash.

  • Thinking about what could transpire tomorrow when, Mr. Clement faces questions from many, I truly do wonder about the division of labour as we strive towards the information age.

    How can it be, that such a decision that runs through the decion making fabric of our society could be left this wide open to such callous ideology.

    Potentially a better question is, given the diversity of the division of labour, and all that the census data runs through, how can the odeology actually not comprehend that to follow such misguided pathways, could lead us to such costly outcomes of inefficiencies.

    How can an ideology developed amd maintained within one man- Mr Steven HArper, who being an economists, cannot see how wide ranging, costly and inefficient such a ideological deciion he is making. Ideology at what cost?

    Do we ratchet down our standard of living to accommodate his notion of what privacy is? For a right wing ideologue, one would think, given his economic training (!??) productivity would be at the center of his thoughts.

    How can being this blinded by ideology allow such a leader to be in such high places.

    So many decisions running across so many labour processes effecting so many task and processes, that within the information age, one would have thought our supreme leader could at least come to a compromise.

    If he does not see the light here, it will be a long day at the election polls. Yes the census long form may not change votes directly, but I do know that it will do more to unite an opposition than any phone calls by Chretien or Broadbent combined.

    This is just inexcusible. A privacy crisis that did not exist and blatantly fabrication to the public, and then Clement totally being dishonest about Statscan’s deciion. When are the tories going to drop the notion that a voluntary form cannot replace a mandatory one, and hence this is purely an attack on Statistics Canada.

    I surely hope the emdia can start reporting how backward and so wide and deep the inefficiencies of this decision could be for the standard of living in Canada. For all the talk about competitive edges and such, Mr. Harper needs to have a real long hard look in the mirror.


  • Patricia Elliott

    I’m astounded that the question that delivered the most ‘privacy’ complaints was household activities. This is the question series that measures all the unpaid work we women do. I wonder what the gender breakdown on those complaints was!

  • As I watch the standing committee, if I had a cell phone number to Fellegi or Dr. Sheikh, I would say the following- we have got to have an understanding as to the cost and benefit of the decision to go from voluntary to mandatory. We are about to lose a whole lot of benefit from but a small cost to privacy. Why did they not mention that at all. It is such a basic level of understanding that the public must be made aware of.


    It is a very basic, but an essential goal that must be met with this committee. For some reason this is not coming out.

    Somebody give the committee a call.

  • Insert Real Name

    Can somebody explain what’s so wrong with question 33 on the 2006 Census? It asks a) how many hours of unpaid housework and b) how many of hours of unpaid childcare were performed by the respondent during the past week. Surely those are important statistics which measure aspects of the division of household labour, and collecting them consistently across the entire country would be useful–what other survey is in a position to do that, country-wide?

    I smell a small anti-social-science concession being thrown in the direction of the Conservatives by this advisory body, but that will only embolden these ideologues.

  • Okay, I know I have been going on and on about the census, however, being that I had worked at the place for 15 years as an analyst/survey designer/ labour economist, I feel it my duty to help with a progressive perspective on this change.

    However, I do want to make one thing clear today. I have said a couple of things about Dr. Munir Sheikh on this blog, that I realize I was wrong about. I think I called him a right ring lap dog at one point, so with all due respect I retract that statement.

    Instead I realized today watching his testimony, that he understood very well what was happening to the barn under the regime in place. He bent but finally he did not break, and to uphold the integrity of the place and for the employees, he resigned. And that took a whole lot more backbone than most many people I know.

    I also wanted to make a point about Don Drummond today. He suggested that somehow one could eventually make do with a voluntary survey, given enough cycles of a pilot to assess and measure the incoming bias.

    Okay, I am not sure he was making this statement to bring people together or what, but he is wrong.

    First off, the bias may not actually be a asymmetrically frozen from one cycle to the next. That would take a few cycles to be sure.

    Second, he forgets that change over time will eventually even erode some of these estimated distributions of under and over coverage. So one would have a hard time measuring change, especially if it occurred within the under-coverage.

    Lastly, he forgets about the functionality aspect of the census. It is used to benchmark and calibrate many other survey programs so by the time one could actually get around to producing some of these estimates the function of the census would be compromised.

    I also would like to say that Mr. Bricker, with all due respect is merely one of those private sector, I have a sample n=500 and I can tell you anything you want. So he might have a bit of experience within the voluntary survey world, but he has little experience from an institutional data collection, survey design, for the long haul. So again I am not sure why this guy was called in as a witness.

    They should have brought in more methods people from Statcan.

  • what was missing today was a well presented fact based assessment of the costs and the benefits of the census long form. We have a notion of privacy and the time it will take 20% to fill out the long form. The benefits or should I say lost benefits will be in the billions. From less reliable information feeding into countless planing …systems and models to a lost sense of belief in data. We need a clear message on the costs and benefits of the long form census data. We need to gather together a quite serious estimate and present it to Mr. Harper. It will take some work but I do know we need to have a figure to present to Canadians- we are rational animals after all. Hence the civil ness of our nation.

    Cost versus benefits. We need numbers to give to the media.

  • I suggest that we all need to look much more closely at what is going on here:

    The removal of Q 33 re unpaid care work from the census has been in the works for a long time. It is part and parcel of the same process that led to the removal of hundreds of stats-based feminist studies on women’s economic status from Status of Women Canada’s webpage, cancellation of funding for virtually all pro-equality groups, etc.

    Here is what is going on:

    First, this never was and is not now really about privacy, ‘invasiveness,’ or optimal sampling techniques. It is about trying to destroy knowledges crucial to full and accurate analysis of anti-poverty, anti-racism, feminist, pro-equality, and other progressive issues — the measurement of the total work effort in the production of monetized and unmonetized value, by sex, by marital/relationship/family status, by race, immigration, disability status, by class, by education…. This total work effort can only be estimated when time spent providing child/self/home/spouse-partner/elder care is no longer measured.

    This change has been in the works since at least 2006-7, as StatsCan began planning the 2011 census.
    As part of the planning process, it carried out its usual survey of user feedback on existing and potential questions. See this report for how StatsCan developed the ‘evidence’ on which it based its decision to drop the unpaid care work issues in July 2008:

    This 2008 report is worth reading. The analysis of how and why it was decided at the management level to drop this question does not meet even the most basic standards of statistical analysis. This vague sort of ‘sampling’ is typical of how the current government decides what policies are best — its evidence is little better than an impression gleaned from statements obtained from a very small number of nonrepresentative respondents. This is not a proper basis for a decision that itself has such huge statistical and policy implications.

    This position is even more irresponsible when it is remembered that Canada has been at the forefront of the development of highly-accurate data on the entire range of use of time, in paid and unpaid activities. And when it is remembered that Canada is under active obligations it helped formulate under the Convention for the Elimination of All Forms of Discrimination Against Women, the Platform for Action adopted at the UN’s Fourth World Conference on Women in Beijing in 1995, and its own 1995 Federal Action Plan — in all those commitments, consistent collection of detailed sex-disaggregated data is a core and essential requirement.

    Second, the National Stats Council is obviously driven by ideology to the extent that it is willing to trade any standard questions in exchange for keeping the longform mandatory. Giving up the ‘jail time’ clause is the only proposed ‘compromise’ change that is rationally related to the alleged complaint of ‘privacy.’ Removing Q 33 re unpaid care work is not rationally connected to that alleged problem; thus the removal of Q 33 must be motivated by some other consideration.

    Given this government’s track record on issues re women, vulnerable groups, income disparities, and the needs of the rich, this look ideological.

    Third, the Minister, the National Stats Council, and StatsCan are all misleading the Canadian public in claiming that the new ‘National Household Survey’ will provide adequate data on Q 33 issues. As you can see from the newly released NHS form, there is NO equivalent to Q 33 on this new NHS form:
    > t- eng.pdf I am shocked at how misleading this is. And in a political debate that is obviously being taken very seriously by a wide range of responsible indioviduals and groups.

    Fourth, if you look closely at what will remain in the 2011 census and what is also going to be in the NHS, you will see that unpaid monetized work will stay in the census — and will also be included in the NHS.

    This calls the whole reasoning behind this process even further into question: If StatsCan originally decided to remove unpaid work questions from the census, why are those questions most directly related to care issues being dropped but those relating to family businesses, farms, etc., being kept in? And if some unpaid work questions are valid why are not all of them valid and still of importance?

    What is this? I think the real point of this whole game has been to remove unpaid care from stats instruments while keeping unpaid ‘productive’ work (in GDP terms) in all statistical instruments.

    Why bother to do this? Because if benchmarked census data on unpaid care work is not available, it will be all the more difficult to integrate data on the nature of growth and change in the care economy, etc., into national accounts — but it will be all the easier keep measuring Person #2’s contributions to monetized production. And I suspect that purely ideological policy proposals will be justified on the basis of that skewed data.

    In particular, I hypothesize that this skew in the collection of different types of time use data will result in the proposal of policies detrimental to women and anyone else vulnerable to low incomes. It is of particular concern as the care economy comes under the microscope re demographic issues, and of course being able the chart the impact of the 2008 economic crisis on the allocation of various kinds of work will also be more difficult.

    If anyone can suggest further points that help illuminate the damage that is being done here, and what its implications are, please post them asap!

    Kathy Lahey

  • Insert Real Name

    Thanks Kathy for your detailed discussion which help me add some more meat to my (non-economist/non-statistician) layman’s concerns and suspicions about this whole stupid decision.

  • Yes, thanks Kathy for illuminating this discussion. I have been trying to find out who sits on the National Statistical Council to see what the gender representation is, but names are not listed on the Statscan website. Although they do not specifically mention removal of the household questions, the national compaign to reverse the government’s decision on the long form census seems to endorse the National Statistical Council’s recommendations. Those of us who are involved with this national campaign need to raise this issue.

  • Insert Real Name

    Kathy Lahey has just posted a longer discussion of Question 33’s removal at

Leave a Reply

Your email address will not be published. Required fields are marked *