Census-Taking More Frequent, Less Controversial in Canada

(April 2001) As Americans wait for the detailed results of the 2000 U.S. census, Canadians are preparing for their twice-a-decade head count.

Statistics Canada, the country’s major statistical agency, has designated May 15 as Census Day. The agency expects to count 31 million people in nearly 12 million households.

Quinquennial, not Decennial

Unlike the United States, which conducts a census once every 10 years, Canada takes its census of population every five years, in years ending in “1” and “6.” According to demographer Roderic Beaujot of the University of Western Ontario, the mid-decade census started in 1906 with the census of the prairie provinces, which were growing quickly and required more frequent head counts. It then became a national tradition in 1956. Dale Sewell of Statistics Canada noted that, because the census is viewed as an economical means of collecting a great variety of data, the mid-decade census was incorporated in the Statistics Act when it was updated in 1971.

Canadian statisticians and demographers appear happy with their system. “Before 1986,” said Beaujot, “the mid-decade census was basically a head count. Now it allows us to sometimes have different questions on the two censuses. A big advantage is that the whole apparatus does not go extinct over a shorter interval.”

The Long and Short of It

Like its U.S. counterpart, the Canadian census has a short form and a long form. The short form, which four-fifths of Canadian households receive, contains seven questions related to population — for example, sex, age, and household and family relationship. (By contrast, U.S. census short form contains a housing question on whether or not householders own or rent their dwellings.) Also included on the Canadian short form are questions about marital status, including common-law status, and mother tongue. (Canada’s 2001 census questionnaire redefines a common-law couple to include two people of the same sex.)

Canada’s long form, sent to the remaining one-fifth of the population, includes the short form questions and 52 additional questions. As in the United States, the long form items are mostly on detailed demographic, social, and economic characteristics of the population-including items on activity status (disability), language spoken at home, citizenship status, migration, level of education, labor force and employment status, and income received in the previous year. Questions about race and ethnicity, which the U.S. census places on its short form, are on Canada’s long form, as are questions that the U.S. census doesn’t include: religion, parents’ place of birth, and language spoken at work.

Like the United States, Canada uses census information to redraw election districts, to establish a basis for transfer payments from the federal government to the provincial governments, to administer national programs, and to facilitate planning by local governments and businesses.

Those Who Refuse to Answer Are Taken to Court

The return rates for the short and long forms have historically been about the same. There is roughly an 85 percent mail-back rate, said Sewell. He noted that eventually Statistics Canada gets about 99.9 percent of the forms back. “We make every effort to convert refusals, and give them many chances to change their minds. As a final step, however, we take hard-line refusals to court.” After the 1996 census, he said, 37 people received summons from the Department of Justice. He stressed that: “This is a criminal charge and results in criminal convictions.”

Net Undercount in 1996 of 2 Percent

One of the major — and most contentious — issues surrounding the U.S. census was the “undercount” and how to handle it. Coverage is also a concern to Statistics Canada officials. The net undercount in the 1996 census was 2.45 percent, which translates into about 723,000 persons. (The net undercount in the 1990 U.S. census, by comparison, was 1.8 percent.) Just as in the United States, the extent of Canada’s net undercount is not evenly spread out across the country. In Canada, the net undercount in 1996 was higher in western Canada and the territories (3.8 percent in British Columbia, 5.2 percent in the Northwest Territories), in large metropolitan areas (3.9 percent in greater Vancouver, 3.2 percent in the Toronto area), and among young adults (5.6 percent for persons ages 20 to 24) and those whose mother tongue was neither French nor English (5.2 percent).

Yet compared with public reaction in the United States, the political and fiscal repercussions of the undercount are less important in Canada, according to Barry Edmonston, director of the Population Research Center at Portland State University. Edmonston said that the strict one person-one vote rules are less pressing and that, although the census is used to develop boundaries for parliamentary seats, there is less legal or political wrangling over them than in the United States. Therefore, he believes, “adjustment and sampling for completing the count have not been an important issue for Canada.”

Availability of Data

Data from Canada’s 2001 census will be released according to the following schedule:

Data Release Date
Population and dwelling counts Spring 2002
Age and sex July 2002
Marital status October 2002
Language, mobility, and migration December 2002
Citizenship, immigration, origin January 2003
Employment questions February 2003
Education questions March 2003
Aboriginal April 2003
Socioeconomic questions May 2003

Kelvin Pollard is a research demographer with the Population Reference Bureau.

For More Information

More information about the 2001 Canadian census is available through Statistics Canada’s website: French speakers may access this information at: