Estimation of the two-group pilot sample size with a cautionary note on Browne’s formula
sample size, estimation, pilot study
Using data obtained from a pilot study, Browne (1995) proposed a procedure for estimating the sample size needed for a definitive two-arm randomised controlled trial when the minimal important difference is specified. Simulations confirm these findings. The results attributable to Browne are extended to consider the degree of error attached to sample size estimation using this procedure. A consideration of the error provides a simple mechanism to estimate the sample size needed for a pilot study so as to control the degree of error in the follow-on substantive trial.