Power analysis and sample size estimation

Determining appropriate sample size is often a difficult decision in the process of developing quantitative research proposals. The novice researcher may well understand that the need for an adequate sample size is an important issue, but lack the knowledge to make an informed decision. Often the sample size will be based on the constraints of practical considerations, such as time or cost, but with little confidence that the sample is adequate in any statistical sense. This paper explores power analysis as an approach to sample size estimation that can be used even by novice researchers to provide a more rational basis for such decisions. The principles underpinning power analysis and the factors that contribute to statistical power are discussed, with an example of power analysis applied to a simple experimental design. Some arguments against a perceived over-emphasis on power analysis are raised. Finally, relevant literature, computer software and World Wide Web resources are included (see Bibliography page 139).