The Just-About-Right Pilot Sample Size to Control the Error Margin

Abstract

In practice, the required sample size for a two-arm randomised controlled trial cannot always be determined pre-study with great accuracy. This lack of accuracy has economic, ethical and scientific implications. The sample size for a pilot study is an important consideration in helping the decision making for the sample size of a follow-on trial. Consideration of under- and over-estimation of the sample size results in the idea of a Just-About-Right (JAR) sample size. For studies involving a minimally clinical important difference (MCID) we present the pilot sample sizes to meet investigator desired JAR considerations.