Data peeking: a quantitative and qualitative exploration of the use of interim analysis

Abstract

Data peeking, quitting data collection early or adding more participants at the end, offers the advantage of saving time and money. However, performing an interim analysis without correction leads to a Type-I error inflation. Using alpha spending function could be used to solve this problem. In this paper, we simulated the effects of interim analysis with and without an alpha spending function on type-I error, power and expected sample size. We also offer a Bayesian perspective to interim analysis. In the last part, we discuss the use of interim analysis in psychological research using a qualitative approach.