Methodology
General Social Survey on Victimization
In 2004, Statistics Canada conducted the victimization cycle of the General Social Survey for the fourth time. Previous cycles were conducted in 1988, 1993 and 1999. The objectives of the survey are to provide estimates of the extent to which people experience incidences of eight offence types, examining risk factors associated with victimization, reporting rates to police, and measures fear of crime and public perceptions of crime and the criminal justice system.
Sampling
The 2004 GSS on victimization had a sample size of 24,000 households in the provinces that were selected using Random Digit Dialling (RDD). Once a household was chosen an individual 15 years or older was selected randomly to respond to the survey. The use of telephones for sample selection and data collection means that the 2004 GSS sample in the provinces only covers the 96% of the population that had telephone service. Households without telephones, households with only cellular phone service, and individuals living in institutions were excluded. These groups combined represented 4% of the target population. This figure is not large enough to significantly change the estimates. The response rate for the GSS Cycle 18 sample was 75%.
Data limitations
As with any household survey, there are some data limitations. The results are based on a sample and are therefore subject to sampling error. Somewhat different results might have been obtained if the entire population had been surveyed. The difference between the estimate obtained from the sample and the one resulting from a complete count is called the sampling error of the estimate. This profileuses the coefficient of variation (CV) as a measure of the sampling error. Any estimate that has a high CV (over 33.3%) has not been published because the estimate is too unreliable. An estimate that has a CV between 16.6 and 33.3 should be used with caution and the symbol ‘E’ is used.
When comparing estimates for significant differences, we test the hypothesis that the difference between two estimates is zero. We construct a 95% confidence interval around this difference and if this interval contains zero, then we conclude that the difference is not significant. If, however, this confidence interval does not contain zero, then we conclude that there is a significant difference between the two estimates.
|