Thank you for giving me the opportunity to share some of my comments on the proposal to abolish the long-form questionnaire.
My name is Jean-Pierre Beaud. I am a political science professor, the dean of a faculty of political science and law and a researcher in an interuniversity research centre on science and technology. I will discuss the long-form census questionnaire based on my expertise not only in political science, but also in methodology, history and statistical sociology.
The first point is more specifically related to the analyses I have conducted as a political scientist. My colleague talked briefly about this issue earlier, but every citizen, even in Canada, must at least accept a few small constraints that enable us, as we say in political science, to live together. We pay our taxes, and if we fail to do so, we are penalized. We show our passport when entering the country. We provide proof of identification in order to access certain areas, such as Parliament. We answer questions that are sometimes very indiscreet in order to be able to perform certain occupations or, as in the case of Quebec, to rent an apartment. We must answer census questions once every five years. Once every 25 years on average, or at most two or three times in our lifetime, we will answer the long-form questionnaire. That's very little to ask, especially since the information we provide remains confidential and is basically aggregated with other information. When it comes to confidentiality, statistics bureaus, in Canada and abroad, have developed very elaborate protection mechanisms. They view confidentiality as a major issue, one which I believe they have resolved in a satisfactory manner.
The second point, which I will not discuss at length, touches on a common criticism from advocates for abolishing the mandatory long-form census. They claim that since Canadians are forced to answer questions even when they do not necessarily have a clear answer, some of them might say just about anything. In such cases, the reliability of the data would be compromised, and a volunteer sample would provide more accurate information about the phenomena being gauged. As a political scientist and a methodologist, I believe that this is a serious issue. When measuring a phenomenon, a statistics bureau cannot just take into consideration firm, unambiguous answers, such as “Yes, I often have difficulty hearing.” What is more, it must not base its analyses only on answers of people who respond because they have an interest in doing so. This is usually the case with questionnaires in volunteer samples.
I will not discuss well-known analyses, such as the results of the Kinsey Report in the United States, a volunteer survey, which, in my opinion and that of many others, projected a false reality of how Americans behave sexually. A statistics bureau must also collect and analyze answers that are less firm, more ambiguous, such as “Should I answer this question sometimes or often?” The reality of a phenomenon is made up of all those elements, and that is what a statistical system must measure.
My third point is based on methodology. Abolishing the mandatory long-form census would create a major problem because it would replace a sound or almost sound methodology—there are always issues, of course—involving a random sample where answering is mandatory, with a much less sound methodology, one involving a voluntary sample. The latter sampling strategy is often considered as the worst one possible. It is used for lack of a better option, especially in medical surveys, and it is riddled with serious problems.
The other major issue, which has already been mentioned, is the facts that abolishing the long-form questionnaire would break the historical chain of data. Longitudinal studies conducted by statistics bureaus and research teams require consistent methodology: the same question, the same method of collection. However—and the paradox is a well known one—at some point, it will be necessary to change the questions, because they are no longer relevant, or the possible answers. Societal structure can change enough to warrant a break in the chain, but in such cases, scientific considerations should take precedence. Outside of those periods, as rarely as possible for researchers and political decision makers, consistency in the data collection process should be required.
The last point I wanted to raise is about the reputation of our statistical agency. I travel abroad very often. My work enables me to compare various bureaus and systems of statistics. I can tell you that the best calling card when I am travelling abroad consists in saying that I am from Canada.
Generally speaking, people have nothing but praise for Statistics Canada procedures. During an interview, the person appointed chief statistician of the Spanish statistics bureau called Ivan P. Fellegi a genius. That might have been an exaggeration, but it is nevertheless indicative of the kind of reputation our organization has built over the decades.
Having studied the facts, I know that that reputation is mainly owing to elaborate survey technologies and methodology. First and foremost, Statistics Canada is recognized for its methodology. I fear that if Statistics Canada had to collect very important data using highly criticized technology, it would not only make it impossible to conduct numerous longitudinal studies and would weaken the political decision-making process, but it would also—and this would have serious implications in the long term—jeopardize one of our prestigious institutions in Statistics Canada.
Thank you.