Thank you.
Madame Chairperson and members of the committee, it is an honour to be invited to speak to you and to make an input to your proceedings as an individual.
To begin, I'll make a few remarks about my professional background. In the mid-1990s Canada established an international lead in formally incorporating measures of unpaid work into the assessment of gender equality. I was the principal statistician behind that effort, and that effort was based on a system called the “total work accounts system”, which is based upon time-use data.
Later on, after three questions concerning unpaid work were inserted into the census, I was the co-author of a small book entitled The 1996 Census: Unpaid Work Data Evaluation Study, published by Status of Women Canada.
Finally, it is relevant to note that I have served brief terms as the president of the Canadian Population Society and as a member of the board of directors of the Population Association of America. These are organizations of social scientists specializing in demography and related aspects of non-partisan data development and data analysis.
You have already heard some excellent testimony, and your members have raised interesting questions. Your past hearings form part of the context for what I will now say.
In trying to provide a helpful contribution to your proceedings, I have kept two general questions in mind. First, to what extent does the change in the design for gathering data in the census put at risk the quality and/or quantity of support provided by statistical data or by data analysis for programs and policies that bear upon gender equality? Second, what are the consequences of having no questions on unpaid work in the new national household survey?
As you're all aware, these questions lead to complex discussions. There is consensus on some points of knowledge and there is also a great deal of speculation around other issues, because the information required to resolve those issues will come in the future. I hope what I say will be a small contribution to the points of generally accepted knowledge and to those speculations that are fruitful from a scientific discussion context.
With regard to the impact of the design change on questions that have been approved for inclusion in the census or in the new national household survey, there appears to be a consensus on two important points.
First, there is almost certain to be an impact, for reasons already explained to the committee by experts in assessing the quality of census data.
Secondly, we have little idea at this time concerning the extent of the impact. On this point there needs to be some concentrated thinking about the particular census or national survey variables, which are especially important in the assessment of data on gender equality.
Census variables that have been repeatedly used among indicators of gender equality include income, education, labour force participation, class of worker, and occupation. In order to interpret variations in gender equality values based on those variables, it is essential, within our national context, to also take into account variables such as language, various dimensions of ethnic origin, place of birth, aspects of family and household status, and age.
When I was a graduate student at the University of Pennsylvania, which at that time was one of the leading population studies centres, and later when I was a professor in the University of Michigan, we had many discussions about census data. During that time it was understood that there are some variables within the census that are measured with very good quality, others with consistently poor quality, and others with quality in between. Now, I would imagine that in 2011 that kind of knowledge will be much more advanced than it was when I was a student.
So we already have good grounds for sensing where we ought to look for strange data patterns in measuring changes over time between earlier censuses and that of 2011.
Unfortunately, this knowledge will not take us very far. As you have already heard from the experts on data quality, if we are in the presence of strange data patterns, without access to independent sources of information we probably would not be able to assess the extent to which those patterns are affected by changes in the design for gathering the data; hence those leaders in any sector with projects based on data that are thought to be reliable and who learn that the available data patterns are of highly doubtful validity will be required to bring new financial and professional resources to the table in the search for independent data. I am speaking about new resources that would need to be brought on stream after Statistics Canada has delivered the data.
Within any of the relevant sectors of our society, the harnessing of professional and financial resources to search for independent data could be a major new product that has to be put on their books at the expense of other uses of those same resources. The projects could be so large that a consortium of sponsors might be required to get it started.
Now, I wish to emphasize that what I have just said deals only with the context in which an important stakeholder learns that the available data patterns are of highly doubtful quality. There is a strong consensus that for many uses, strange data patterns will not be found. Only the future will tell the relative frequency of good and bad situations concerning these data patterns.
In this connection, we should note that many users troubled by what they feel are serious defects in data delivered to them by Statistics Canada will turn to Statistics Canada for help. Experts in data analysis who are well grounded in such subject matters as income are the key Statistics Canada resources to provide that help to those users. Are sufficient professional resources to respond to users who need help going to be available from Statistics Canada as we go forward?
What are the consequences of a new national survey in which the former census questions about unpaid work are absent? I wish to make three points in connection with this debate.
Please study the central conclusions of the book I cited earlier. There you will learn that when unpaid work questions were inserted into the census, the authors did not think it possible or even advisable to try to estimate the total volume of unpaid work using census questions, and this for two reasons.
First, significant elements of unpaid work are not covered, such as volunteering for organizations or unpaid inputs to a family business. Second, the kinds of questions used in the census risk double counting of time between at least two of the activities in question 33. So these activities need to be treated separately when you are using the census data.
In inserting those items into the census, the idea was that communities concerned with services pertaining to child care or pertaining to elder care could turn to Statistics Canada for assistance in connection with background data. When I speak of concerned communities, I would like you to think of the mayor of Kamloops in British Columbia. As Dr. Fellegi has already pointed out, the role of the census in these areas was to provide complementary data. I repeat for emphasis that what we are thinking about are such specific services as child care and elder care, and not the overall volume of unpaid work.
The second point I would like to make in connection with this debate concerning the time-use data is that where these data count only the primary activity undertaken within a given time slot by a respondent, the census question on child care provides better data about the amount of time being spent on child care. You will find evidence for that in my book. However, on listening to the hearings of the 16th, I learned that the latest time-use survey has arrangements to better record activities in situations in which people are doing more than one thing at the same time. This will be an important improvement in the time-use data.
My last point in connection with the debate is to plead with you to reflect upon a process in which the debate is resolved by thinking only about the needs of the federal government departments and what Statistics Canada may or may not be able to do to meet those particular needs.
Traditionally, work done at Statistics Canada has had to be sensitive to at least the following three classes of stakeholders: one, the designer of a marketing or product innovation program in the business sector; two, the non-profit organization leader developing a program of assistance to specific population groups; and three, a government policy or program leader needing to fine-tune a municipal, provincial, or federal policy program to address the situations of particular subgroups of men and women in our society.
Sensitivity to the widely varying data needs of these three classes of stakeholders was explained on the grounds that Statistics Canada was a national resource when viewed from the perspective of society as a whole. At the same time, we all need to be conscious of the fact that this agency is now subject to constraints that are affecting its ability to be adequately helpful across the range of sectors that I've just identified. Some of these constraints are demographic; I'm referring, of course, to the wave of baby boomer retirements.
Thank you for your attention. I would be happy to try to respond to your questions as an individual.