3. Sampling, study size, and power

Publisher Website
Google Scholar
Abstract
Precise definitions of study populations and samples are key to the interpretation and generalizability of findings. This chapter describes types of sampling, how to deal with non-response, and validity of measures (including problems relating to bias and variance). Details are then given of how sample size relates to the testing of the null hypothesis, and the meaning and definition of significance level and power. This is followed by detailed techniques for the determination of sample size for different types of epidemiological studies (ecological, cross-sectional, case-control studies, cohort studies, and experimental studies). It defines sample size and power in relation to measures of difference between matched and unmatched samples, correlation, odds ratio, and relative risk.