Project Canada 1975
DOI
10.17605/OSF.IO/GKADCCitation
Bibby, R. W. (2021, March 16). Project Canada 1975.Summary
Begun in 1975, Project Canada has generated extensive information on life in Canada, with specific attention given to social issues, intergroup relations, and religion. The project has taken representative samples of Canadians every five years, creating panel studies through which social change and stability can be monitored.The ARDA has added six additional variables to the original data set to enhance the users' experience on our site.
Data File
Cases: 1917Variables: 305
Weight Variable: WT
Data Collection
May-October 1975Funded By
The United Church of Canada, the Canadian Broadcasting Corporation, the Solicitor General of Canada, the University of LethbridgeCollection Procedures
The survey was carried out by mail between mid-May and mid-October of 1975 -approximately a five-month period - from York University in Toronto. Post-August work was concluded from the University of Lethbridge. French versions of the questionnaire were returned to the Centre du Sondage at the University of Montreal.The questionnaire was constructed to provide comprehensive data pertaining to three main subject areas being investigated: social issues, intergroup relations and religion, with the latter including many of the items used by Charles Glock and Rodney Stark in their definitive 1963 study of religion in the United States. Many other items were drawn from the annual National Opinion Research Center (NORC) survey in the U.S., facilitating Canadian-American comparisons. The length of the survey was 11 pages, with approximately 300 variables and a completion time of one-two hours. Approximately two to three weeks after the initial mailing, a follow-up letter was sent to non-respondents, while a second follow-up - a returnable postcard - was sent still another two to three weeks later. In mid-September, non-respondents in communities under 50,000 were mailed a second copy of the questionnaire. Data collection was ended prematurely in early October by a national postal strike. In April 1976, respondents were sent a thank-you post card, informing them how the results were being disseminated, and welcoming further inquiries.
Sampling Procedures
The initial intent of the project was to draw a sample that would facilitate provincial comparisons; therefore, 500 adults from each of the 10 Canadian provinces were selected. They were stratified according to province size, 400 from the largest city in each province, 50 from a randomly selected community of 10,000 to 15,000, and an additional 50 people from a randomly selected community of under 10,000. The total urban sample was 4,000; the smaller community sample, 1,000. Sampling was carried out using telephone directories for each of these 30 communities.Some 300 (6%) of the sample were "legitimate losses" - people who could not be located or were unable to fill out the questionnaire because of health and age factors. Telephone interviews with a random sample of 50 of the remaining non-respondents in four cities - St. John's, Montreal, Toronto and Edmonton -revealed an average of an additional 20% of the sample could not respond because of factors that included not only health and age, but also language and the fact that the sample member was deceased. This percentage did not include people who maintained they never received the questionnaire (a very high number), since clearly there are problems in assuming such a report was accurate. It appears no more than 3,700 people (i.e., 74% of 5,000) could return the questionnaire. Of these, 52% did so.
The initial decision to focus on interprovincial comparisons obviously produced a sample that required an adjustment for provincial size, even though the unweighted sample bears a very close resemblance to the Canadian population with respect to age, income, marital status and - allowing for the anticipated under-representation of the Francophones - even ethnicity and religion. Differences between the sample and population by province, community size and gender have been adjusted by weighting; such a procedure produces a sample that very closely resembles the Canadian population in a demographic and social sense.
The weighted sample has been reduced in size to 1,200 cases to minimize the use of large weight factors. So weighted, the sample results are accurate within about plus or minus 4 percentage points of the population figures, 19 times in 20.
