Hostname: page-component-cd9895bd7-gvvz8 Total loading time: 0 Render date: 2024-12-28T21:01:05.505Z Has data issue: false hasContentIssue false
  • English
  • Français

A National Sample Design

Published online by Cambridge University Press:  10 November 2009

Lawrence Leduc ,
Harold Clarke ,
Jane Jenson  and
Jon Pammett
Lawrence Leduc
Affiliation:
University of Windsor
Harold Clarke
Affiliation:
University of Windsor
Jane Jenson
Affiliation:
Carleton University
Jon Pammett
Affiliation:
Carleton University
Get access Rights & Permissions [Opens in a new window]

Extract

The construction of national survey samples in Canada involves a number of theoretical and practical problems that range well beyond purely statistical considerations. Sample sizes for instance will tend to be large in comparison with those employed in many other nations not for greater accuracy but primarily because of the sheer diversity and geographical dispersion of the Canadian population. Likewise, the high cost of survey research at the national level in Canada virtually dictates that substantial efforts be made to achieve an “optimum” design for any single study. The purpose of this note therefore is to outline in brief the several considerations which underlay the design of a national sample for the post-election survey which will be carried out under our direction following the federal election of July 1974. It is offered as a commentary on the steps that were taken to provide a sample design consistent with the research focus of the project. The design has a number of important implications for both primary and secondary analysis of what will be the third national election study to be carried out in Canada. Each of these studies has followed particular theoretical interests, in all cases having important implications for potential use of the data set by other analysts.

Type
Notes
Information
Canadian Journal of Political Science/Revue canadienne de science politique , Volume 7 , Issue 4 , December 1974 , pp. 701 - 708
Copyright
Copyright © Canadian Political Science Association (l'Association canadienne de science politique) and/et la Société québécoise de science politique 1974

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 This study is supported by Canada Council Grant S73-0633.

2 The most important single text on sampling is Kish, Leslie, Survey Sampling (New York, 1965).Google Scholar A more general discussion may be found in Kish, L., “Selection of the Sample” in Research Methods in the Behavioral Sciences, ed. Festinger, L. and Katz, D. (New York, 1953), 175–240.Google Scholar Other useful sources are Hyman, H., Survey Design and Analysis (Glencoe, 1955)Google Scholar; and Moser, C., Survey Methods in Social Investigation (London, 1958).Google Scholar Moser describes in chapter 8 a British sample design similar in many respects to the design reported here. The sample design employed by Butler and Stokes in British electoral studies also provides a useful comparison. See Political Change in Britain (London, 1969), 449–62. A useful comparison is also afforded by the studies conducted for the Royal Commission on Bilingualism and Biculturalism. See the discussion of sample design for these studies in Johnstone, J.C., Young People's Images of Canadian Society (Ottawa, 1969), 105–11.Google Scholar

3 See Gray, P. et al., The Register of Electors as a Sampling Frame (London, 1950).Google Scholar There also exists the option of treating electoral lists (particularly ones which may be out of date) as a frame for drawing a sample of households. See Moser, Survey Methods, 125. A procedure developed by Kish may then be followed for random selection of individuals within sampled households. Kish, L., “A Procedure for Objective Respondent Selection Within the Household,” Journal of the American Statistical Association, XLIV (1949), 380–7.CrossRefGoogle Scholar

4 Excluding those areas of Canada which are inaccessible to interviewing, for example, Yukon and Northwest Territories, Labrador, and the far northern portions of Ontario, Quebec, and the western provinces. Depending on the organization conducting the field work, it is possible to achieve 90 to 95 per cent coverage of the total population.

5 But not anticipating the creation of new substrata. For example, the use of second party vote does not imply the creation of additional substrata but rather assures that the over-all proportion of second party vote in each stratum sample will match that in the stratum population. See Yates, F., Sampling Methods for Censuses and Surveys (London, 1953)Google Scholar for a general discussion of multiple stratification without control of substrata. The use of second party vote was intended to assure the adequate representation of political minorities, the original scheme being to use ndp vote in Ontario, Social Credit vote in Quebec, and second party vote in all other provinces. Since Social Credit finished second in total vote in Quebec in 1972, the distinction is redundant for that province. In Ontario however, ndp vote is used for the stratification. The variable employed in each case is British Columbia, Progressive Conservative: Quebec, Social Credit; Ontario and Saskatchewan, ndp; all other provinces, Liberal.

6 The stratification and selection of the first two sample stages is achieved by computerizing the relevant electoral data for each unit, and selecting units in the prescribed manner by means of a computer program written for the purpose. A random start is obtained from a bloc of random numbers inputted with the data.

7 Adjustment of this estimate may be required to allow for a lower completion rate resulting from the timing of the election. The fact that our field work will now coincide with a peak vacation period may result in a lower completion rate. However, we have adjusted the call-back procedure to attempt to minimize this effect, and may draw a slightly larger initial sample to permit more flexibility.

8 See Butler and Stokes, Political Change in Britain, 450 for a parallel example. The major alternative to this design would be that of a variable cluster size, with every nth name being selected from poll lists. Because poll sizes can vary quite widely, this would tend to result in an overly wide scattering of interviews in some parts of the country, and an over clustering effect in urban polls where it is least desirable. It would be possible to vary the cluster sizes within narrower limits than those provided by systematic sampling, but this would seem a needless complication of the design.

9 This ratio is discussed in some detail in Kish, L. and Hess, I., The Survey Research Center's National Sample of Dwellings (Ann Arbor, 1969), 47–52.Google Scholar

10 See Keyfitz, Nathan, “Sampling Probabilities Proportional to Size,” Journal of the American Statistical Association, XLVI (1951), 105–9.CrossRefGoogle Scholar