Chapter 2. Methodology of the Survey and Analysis
Methodology of the Survey
This report presents the findings of a national survey of attitudes towards firearms and gun control conducted by Canadian Facts on behalf of Professor Gary A. Mauser, Simon Fraser University, and Professor Taylor Buckner, then of Concordia University (and now retired). It was funded by the Langley Symposium of Responsible Firearms Owners.
A total of 1,505 interviews were conducted with Canadians 18 years of age or older in all provinces between January 18 and 26, 1995. Interviews were conducted over the telephone by trained and experienced interviewers from 9 facilities across Canada. Interviewing was continually monitored and supervised by Canadian Facts using a computer aided telephone interviewing system [CATI].
A plusdigit dialing method was utilized to select a random sample of telephone numbers in order to provide coverage of all households with telephones in all ten Canadian provinces. Households in the territories were excluded. A disproportionate sampling design was chosen in order to obtain a sufficient number of responses from less populated regions. This enabled more reliable statistical comparisons to be made between regions. Interviews were conducted in five regions across Canada: 275 in BC, 326 in the Prairies, 350 in Ontario, 351 in Quebec, and 203 in the Maritimes. The questionnaire and interviewer instructions were translated into French for interviewing in Quebec. Within each region, interviews were proportionate to the population in order to provide adequate representation for rural and urban residents. In order to represent the Canadian population as a whole, the results were weighted to reflect the actual proportions of households in Canada.
A sample this size has a sampling error of ± 2.5 % points. This means that means that the results are accurate within 2.5 percentage points at the 95 percent confidence level. The sampling errors of the regional estimates are correspondingly larger due to the smaller sample sizes.
|Sample Size||Sampling Error|
|BC||275||+ 6.0% points|
|Prairies||326||+ 5.5% points|
|Ontario||350||+ 5% points|
|Quebec||351||+ 5% points|
|Maritimes||203||+ 7% points|
|Canada||1,505||+ 2.5% points|
Within each household, an adult respondent was randomly selected to be interviewed using the nextbirthday method to ensure proportionate representation of all ages of both male and female residents. All calls were placed in evenings between 4:30 p.m. and 9:30 p.m. on weekdays and between 10:00 am and 4:00 p.m. on Saturday to ensure the inclusion of working individuals.
Up to five calls were made to each residence in an attempt to obtain a completed interview with a qualified respondent. No substitutions were permitted at the household or individual selection stages. The completion rate for this study was relatively high in comparison to similar studies. The highest completion rates achieved were in the Maritimes and Prairies at 41%, the lowest completion rate was in Toronto with just 26%.
The questionnaire used in this survey was developed by the authors in conjunction with Canadian Facts in order to accommodate the questionnaire into the CATI format. The final draft of the questionnaire was pretested the day before interviewing began in Vancouver. Minor revisions were made in question wording in order to ensure that the interviews flowed smoothly. A copy of the final version of the questionnaire is appended to this report.
Methodology of the Analysis
For this analysis, Weighted Representative Samples were used in almost all cases. The purpose of the analysis is to give an idea of the national views and attitudes, so the weighted sample is appropriate. For some specific tables involving small numbers of cases, such as in the self-defense chapter, the unweighted sample was used because of an interest in individual instances rather than a national average. These exceptions are always noted in the tables.
Each Chapter provides the weighted frequencies for each question, and examines each question in terms of the possible influences of the respondents' backgrounds. These include:
Underlying this progression is an assumption about the nature of causality - that basic demographic factors, such as region of the country, rural or urban residence, gender, age, education and income exert their primary influence on the basic values of the respondents. These values, plus knowledge and salience of gun control, then influence perceptions of problems. Background, knowledge, and perceptions of problems then influence perceptions of the utility of gun control.
Perceptions of the utility of gun control, along with basic values and knowledge, then influence attitudes towards issues such as universal registration, the confiscation of handguns, and support or opposition to politicians who favour these measures.
In addition, somewhat out of this causal sequence, the backgrounds of firearms owners, and those who have used firearms for self-defence are analysed.
Throughout the analysis the fundamental influence of values, and of gun ownership will be discussed. In general the results are presented in percentage tables.
Note: Those who might require cross tabulation
of all the questions by background and analytic variables may
find them in Annex B. Annex B provides further details, such as Chi-square
values, probability significance levels and measures of
association (such as gamma and phi) and other relationships of
relevance to professional readers of statistical data .
| Forward to Chapter 3 | Back to Top of Chapter 2 | Back to Table of Contents | Back to Home Page | E-mail Taylor Buckner |