Power and sample size for the S:T repeated measures design combined with a linear mixed-effects model allowing for missing data