Power-analyse, of de bepaling van de steekproefgrootte

Summary

Power analysis

Although standard statistical texts pay considerable attention to the problem of alpha-errors (concluding that there is a difference, while in fact there is not) little attention has been paid to beta-errors (concluding there is no difference, while in fact there is). The probability of committing the first kind of error is established by the investigator as the level of significance, or alpha (a). The complement of ß (1-ß) is the probability of obtaining a significant result, and is referred to as the power of a statisticaI test. Power analysis represents a method for determining power and/or establishing the needed sample size. Due to small sample sizes, many studies have insufficient power. In this article we discuss poweranalysis in the context of several research examples.