A group sequential t-test with updating of sample size

If the maximum sample size for a group sequential t-test is calculated on the basis of a prior estimate of the variance, power at a fixed distance from the null hypothesis is not robust to misspecification of the variance. We propose a new group sequential t-test, based on Stein's (1945) two-stage test, in which we use internal estimates of the variance to update the maximum sample size, and allow early stopping to reject the null hypothesis through a Lan & DeMets (1983) error-spending function. We evaluate our procedure's performance and show that it closely controls the type I and II error rates.