Читать книгу Pediatric Epidemiology - Группа авторов - Страница 29

WHO Europe [12] suggests a 2-stage sampling process where first a subset of schools is selected via cluster probability sampling and thereafter a randomised sampling of school classes for the appropriate ages. This is actually the way it has been performed in Sweden since 1995.

When calculating the sample size, a number of factors have to be considered:

•Estimated prevalence of the problem

•An acceptable error margin (normally 5%)

•Precision level

•Clustering of data

•Estimated non-response rate

Prevalence estimates should ideally be based on findings from previous studies with similar populations.