Que ferait perdre au Canada l’abandon définitif du questionnaire long obligatoire du recensement, annoncé par Ottawa en juin dernier ?
In June, the federal government announced the discontinuation of the mandatory long-form census. Over the ensuing weeks, the government cited privacy concerns among others as the reasons for the decision, and explained that the voluntary National Household Survey would replace the previously mandatory long-form census, while the short form of the census would remain mandatory.
Barring some reversal, all Canadians will be required to answer the short-form questions, which probe basic household composition information (number of residents, marital status, family relations among residents) and agricultural production. In response to a Charter challenge and a suit regarding the census’s role in enabling the government to fulfill its responsibilities under the Official Languages Act, two language questions about language ability and usage were shifted from the (now voluntary) long form to the (still mandatory) short form.
The new long form of the census will be distributed to a third of Canadian households, to be completed at the discretion of the recipients. It will probe demography, activity limitations, citizenship and migration status, language, ethnic origin, Aboriginal identity, religion, mobility, place of birth of parents, education, labour market activities, place of work, work activity, child care and support payments, income and housing. (The 2011 form includes all the questions included in the 2006 mandatory long form, with the exception of those on unpaid/household work.)
In this article I will attempt to answer the question “What will Canada lose if the mandatory long-form census is permanently discontinued?” I will not revisit debates about the methodological validity of the voluntary long form. I will take as given, on the grounds that there is broad agreement about this point among statisticians and social scientists, that the voluntary survey will not provide data of the same quality and reliability as the data gathered through the mandatory form of years past. Even Industry Minister Tony Clement has acknowledged in correspondence with critics that “the government understands that this change could have an impact on data quality.”
Of the many organizations that have expressed opposition to the census decision, many have issued statements or made submissions to the hearings of the Standing Committee on Industry, Science and Technology, describing the consequences of the census change in their own fields of work. Having sifted through as much material as I could locate, I will attempt to summarize the lost data and diminished analytical possibilities in as much detail and with as little redundancy as possible. This list will not, of course, be exhaustive; organizations that made statements tended to provide just an overview of their uses of long-form data in open letters that stated their case in just a page or two.
Numerous commentators have remarked that the geomake census data especially valuable here. For instance, the Canadian Nurses Association reports that nurses use census data to make projections about the health care needs of people in remote communities. One nurse planning to deliver cancer screening services to a community in northwestern Ontario examined, for instance, the number of women between the ages of 50 and 70 in a given community, along with important factors relating to determinants of health (income level, adequacy of housing and clean water). Census data helped to determine how long health professionals would likely be needed in the community to deliver the necessary screening services. In smaller, less diverse countries, for example Denmark, the potential inefficiencies resulting from an inaccurate portrait of a particular community would be less.
A second point noted widely pertains to the quality and longevity of the Canadian census to date. Canada’s strong census-taking tradition has resulted in a store of data that tells the story of Canadians — their work, their families, their tools, their homes, their lifestyles — for a century and a half. As Lisa Dillon points out in Canadian Public Policy, considerable investments have been made over time not only in collecting census data but in making them accessible to researchers and the public. “With the exception of the 1861 and 1961 census years,” Dillon writes, “sample databases and in some cases 100 percent databases exist for every census year from 1852 to 2006. These efforts have been supported by significant investments funded by the Canadian government in the form of grants from the Social Sciences and Humanities Research Council as well as the Canadian Foundation for Innovation.” These investments increase the stakes of contemporary decisions about the census, since a major methodological change that eliminates the continuity across census waves diminishes the value of all past investments. Dillon later observes that although historical data are valuable in themselves, “Unfortunately, the extent of unmeasurable differential nonresponse in the [voluntary 2011] NHS will render it non-comparable to all previous census data. This development will put an end to the usefulness of all previous censuses for the purposes of studying our contemporary population and to the national series as a whole.”
The new long form of the census will be distributed to a third of Canadian households, to be completed at the discretion of the recipients. It will probe demography, activity limitations, citizenship and migration status, language, ethnic origin, Aboriginal identity, religion, mobility, place of birth of parents, education, labour market activities, place of work, work activity, child care and support payments, income and housing.
Charities use long-form census data to target their investments geographically and demographically. Frances Lankin, the former president and CEO of United Way Toronto, has written that “the long-form census is the only reliable source of data for understanding who is affected by issues like poverty — their age and gender, their ethno-cultural background and other details…Census data not only tells us who is affected by poverty, but where poverty is located in terms of geography. Perhaps most importantly, census data allows us to track this information over time, in order to assess trends and identify possible solutions.” The United Way’s influential 2004 report Poverty by Postal Code relied on long-form census data to identify neighbourhoods where poverty was not just concentrated, but also growing over time. The report mobilized millions of dollars of philanthropic and public money to target 13 “priority neighbourhoods” in the Greater Toronto Area, based on strong evidence that these were the areas that needed investment the most.
Pointing to the value of the longform census in establishing the weighting and methodology for other studies, Imagine Canada President and CEO Marcel Lauzière has indicated that “we have serious concerns about the impact [the long-form decision] will have on, for example, the Canada Survey of Giving, Volunteering and Participating — a world-leading initiative that informs the nonprofit sector’s thinking and actions.”
The organization Sexual Assault and Violence Intervention Services points out that women and children who belong to marginalized social groups, especially those with low incomes, are at highest risk for sexual violence. Long-form data enable social agencies that address sexual violence to make their services as accessible as possible to those who need them. Many experts, including former chief statistician Ivan Fellegi, have argued that marginalized people are those most likely to be underrepresented in the voluntary National Household Survey.
The census provides data on household income and enables analysis of the relationship between educational attainment and household income across the country and in particular regions. Past long-form questionnaires have asked questions about paid and unpaid work, enabling analysis of labour force participation according to demographic traits, as well as analysis of household labour over time in the face of changing family composition and gender norms. The Labour Force Survey, which tracks trends such as unemployment and labour force participation, relies on long-form census data to establish sampling frames.
Businesses rely on long-form census data to learn where prospective workers and customers are located, so they can determine advantageous sites for both plants and retail facilities.
Marketers use long-form census data to determine how to reach their target audiences through direct mail and other campaigns. The Canadian Marketing Association has suggested that irrelevant and redundant marketing appeals are likely to increase in the absence of good data.
And as Lisa Dillon notes, longform census data have enabled researchers to better understand the relationship (not a simple one) between pensions and Canadians’ retirement choices.
As Paul Jacobson has observed on his website (jacobsonconsulting.com), businesses use longform census and Labour Force Survey data to understand local and regional workforce composition, and thereby develop their recruitment and retention strategies.
Micheline Dionne, President of the Canadian Institute of Actuaries, writes in an open letter, “The retirement savings sector needs accurate, reliable data on labour markets and on sources of income to back solid pension reforms, not only for the current proposals but also the future reforms that will, no doubt, be needed as the Canadian pension landscape evolves.”
Canadian health workers use longform census data to understand the health of populations down to the neighbourhood level and to understand how factors such as employment status, gender and income are connected to health outcomes and access to care. (An example in one open letter was a study that found infant mortality rates in low-income areas of Saskatoon to be 448 percent higher than in the rest of the city.) The Canadian Medical Association Journal editorial on the census issue stated that 7,060 articles in the National Library of Medicine’s database referred to the “Canadian census.”
The Urban Public Health Network, whose statement on the census was endorsed by the Canadian Public Health Association, states that long-form census data enable it to assess the efficacy of public health programs, measuring whether health outcomes in particular areas are better, worse or the same after an investment or intervention.
Census data enable public health workers to deliver services and information in appropriate languages — whether this means Cantonese or Tagalog in a Vancouver neighbourhood or French in a small northern Ontario community.
The Canadian Nurses Association notes that in Alberta, 7.2 percent of registered nurses speak French but just 1.0 percent report using their French at work on a regular basis. Managers can use census data to locate neighbourhoods with high concentrations of francophones (the fastest-growing language group in the province) and redistribute their personnel to make better use of nurses’ existing language capabilities.
Census data enable public health workers to target particular migrant populations for initiatives that would disproportionately benefit them. For instance, Toronto Public Health offers tuberculosis prevention classes in neighbourhoods with high concentrations of people who have migrated from areas where TB is endemic.
Alberta Health Services President and CEO Dr. Stephen Duckett sums up the need for census data thus: “If we are to plan for health services we need good information about both what the population is like now (to respond to current needs) and how the population is likely to evolve over time. We also know that health service use varies by factors such as income, country of birth, employment status, and other social and economic realities. So putting these two issues together, we need to have accurate and longitudinal [time series] data about these factors which influence health status and/or health system use.”
As the C.D. Howe Institute’s president and CEO, William Robson, put it in the Globe and Mail, arguing that Canadians who want smaller government should care about robust, longitudinal census data: “The state plays a huge role in Canadian health care: Good information on personal and neighbourhood characteristics can help us know if we are healthier or sicker as a result. It redistributes income on a colossal scale: The long-form census can reveal much about the successes and failures of these programs.”
The Urban Public Health Network notes that much of the independent research conducted in the field of public health relies on census data to establish sample frames.
Long-form data keep policy-makers, educational institutions and business groups apprised of how educated the Canadian population is and where the educated are living. They track post-secondary school attendance and completion rates, as well as awarded certificates, diplomas and degrees. Census data also help us to understand flows of skilled and unskilled workers across provinces and territories, and into and out of Canada. (Think “brain drain” and quantitative data against which to test the common anecdotes about foreign-trained brain surgeons driving taxi cabs.)
Education ministers in Ontario and Quebec argue in an open letter to Industry Minister Tony Clement that long-form census data help us to understand the relationship between post-secondary education — including fields of study and types of degrees pursued — and labour market outcomes. Understanding this relationship helps us to understand whether we are investing in the right aspects of our education system. In a knowledge-driven economy, the ministers describe this knowledge as increasingly valuable.
Long-form census data have been used to evaluate the effectiveness of high school programs in French and English schools in New Brunswick, Canada’s only officially bilingual province.
Those working on local issues across a range of fields emphasize that the long-form census is in many cases the only source of reliable data at the neighbourhood level. Representative samples from voluntary surveys may be able to generalize reliably about a city, but they are not able to differentiate populations block by block, which is the level of granularity local workers often need in making considerations such as where to locate, for instance, a transit route, a child care or seniors’ centre or a vaccination clinic.
Urban planners use long-form data to:
- locate public facilities in such a way as to maximize their usefulness to local populations;
- target investments in transit expansion where and when transit is most needed;
- situate affordable housing in areas with high demand;
- determine where children’s services should be located;
- determine where recreational and cultural opportunities are most needed; and
- understand how people’s homes and workplaces are related geographically. Long commutes consume time and fuel and often cause stress and those nasty “externalities”: smog and greenhouse gas emissions. Sound journey-to-work data can help communities reduce commute times by co-locating homes and workplaces, and improving transportation infrastructure where long commutes remain common.
According to the Canadian Institute of Planners and the Transportation Association of Canada, voluntary surveys about lifestyle issues related to the built environment (commuting to work, shopping, use of public facilities, transportation use) often use the long-form census to establish their sample frames.
Canada welcomes a million immigrants every four years. One in five of us is foreign born, and that proportion is increasing over time. The country’s economic and social success is strongly linked to how these newcomers fare. In the past few years, long-form census data have revealed that recent immigrants’ fortunes in the labour market have been deteriorating since the 1980s. Social service agencies use these data to target their efforts to support newcomer groups (in the appropriate languages and in the right neighbourhoods). Census data will enable us to track the effectiveness of their interventions in the years to come — and to do so in detail, according to language group, settlement area, country of origin, educational attainment and other dimensions. (Recent immigrants are among the populations that tend to be underrepresented in voluntary surveys.)
Environics used the 2006 long-form data to construct our sampling frame based on identity group (First Nations, Métis, Inuit), age, gender and education. Without a credible and rigorous sampling methodology, our study could easily have been dismissed; instead it was taken seriously by a wide range of major organizations focused on Aboriginal issues.
Outgoing Toronto Mayor David Miller wrote to the Industry Minister to state that the city uses long-form census data to determine where to locate newcomer settlement services.
The Ontario Council of Agencies Serving Immigrants (OCASI) expresses the concern that “the elimination of the mandatory long census questionnaire, and the resulting diminished ability to account for our population’s profile might end up working against one of the central tenets of Canada’s Immigration and Refugee Protection Act (IRPA), which states in section 3.1.e that immigrant settlement and integration is a two-way endeavour involving immigrants’ efforts to adapt to this new country as well as Canada’s investments to respond to immigrants’ integration requirements.”
The Canadian Bar Association indicates that Canadians involved in serious personal injury cases benefit from long-form census data because lawyers and judges rely on those data to evaluate the lost income and projected economic needs of people who are removed from the paid labour force through injury. Those who have not been closely connected to the paid labour force, such as children and people with disabilities, rely especially heavily on census data.
Researchers have used long-form census data to understand the performance of Aboriginal people — especially the young — in the school system and in the economy, knowledge that will help us to understand which approaches and environments help to promote the success of First Nations, Métis and Inuit people. A small number of national surveys ask respondents whether they claim Aboriginal identity, and these generally use the long-form census to establish their sample frames.
Researchers use long-form census data to:
- understand the distribution of the Aboriginal population between urban, rural and on-reserve communities;
- gauge the current and projected proportion of school-age cohorts in various provinces, knowledge that might inform approaches to education and curricula;
- compare the average incomes and earnings of Métis versus First Nations/Indian people over time, knowledge that might inform public policy or other interventions to help understand and spread the drivers of success;
- understand how educational attainment affects employment rates in Aboriginal communities;
- examine intergenerational trends in educational attainment among on- and off-reserve Indian/First Nations, Métis and non-Aboriginal populations;
- understand the high school noncompletion rates among Indian/First Nations, Métis and non-Aboriginal young adults across the various provinces;
- study trends and patterns in Aboriginal home ownership;
- learn whether high school and post-secondary education completion rates are improving or deteriorating over time — both overall and in particular locations; and
- study the socioeconomic status of Aboriginal peoples broken out by school district for 360 schools in British Columbia. (This was in support of an assessment of the performance of BC schools with large Aboriginal populations.)
On a personal note, my Environics Institute has recently completed a landmark study of the experiences, identities, values and aspirations of Canada’s urban Aboriginal peoples. Environics used the 2006 longform data to construct our sampling frame based on identity group (First Nations, Métis, Inuit), age, gender and education. Without a credible and rigorous sampling methodology, our study could easily have been dismissed; instead it was taken seriously by a wide range of major organizations focused on Aboriginal issues.
Long-form census data have enabled studies of trends in family and household composition (for instance, the rise of single-person dwellings and the decline and resurgence of multigeneration households over the past century).
Long-form census data have told the major stories of urbanization, industrialization and post-industrialization of Canadian society over the past century, tracking not just people from the country into the cities (this would be possible with the short-form census) but also their jobs from farms and resource extraction sites into a range of urban and suburban jobs.
Long-range trends, such as declining fertility rates, come into focus in longform census data. Once these trends have come to light, long-form data enable us to examine the correlates of diminished fertility, such as urbanization and high rates of education for women.
Religious groups use long-form census data to understand where they should locate facilities, services and messaging efforts. The Canadian Jewish Congress called long-form data “a critical source of information for planning, fundraising and implementing programs and services that support…the cultural, social, health care, educational, housing, recreational and spiritual needs of our communities.” The president of the Canadian Conference of Catholic Bishops, Bishop Pierre Morrisette, points out that the charitable arms of religious groups use census data — presumably as other charities do — to target their efforts geographically and demographically, and to aid their fundraising efforts. The Evangelical Fellowship of Canada emphasizes the value of long-form census data in helping it reach out to evangelical newcomer communities, for whom belonging to a Canadian religious congregation may prove an important path to social integration.
Lisa Dillon notes that Canada’s participation in International Integrated Public-Use Microdata Series (IPUMSi) has made Canada a candidate for international comparison studies. The absence of census data that are comparable over time will diminish the likelihood that international researchers will examine Canada’s performance relative to other societies, a change that may deprive Canada of valuable insights about our competitiveness or the effectiveness of our policies. (Imagine those bar charts showing the rankings of our comparator countries like the United States, Britain, France, Australia and New Zealand on dimensions related to post-secondary education or the success of immigrants and their children — with question marks where the Canadian bars should be.)
Perhaps the biggest thing Canadians will lose with the mandatory long form is our confidence in our shared understanding of what is happening in our society. “What is happening” is something we come to know by examining how things have changed over time and measuring how we compare with other countries. It brings to mind Daniel Patrick Moynihan’s often-quoted remark that we are all entitled to our own opinions, but not our own facts. Put another way, we should strive to privilege facts over anecdotes and assumptions; we should test our hypotheses — to learn what is true to the extent we can, and undertake actions that have been demonstrated to achieve the results we desire. I would contend that agreeing on the facts enables us to compare and debate our diverse opinions not only with greater clarity and mutual respect, but also with more fruitful outcomes.