This chapter argues that sample size and variance of the outcome variable play the main roles in defining statistical efficiency. It discusses the crossover design for an illustration of the trade-offs in efficiency. The crossover design is also useful as a way to introduce the analysis of variance cell-means model. A characteristic of a crossover design is that each patient receives the test compound as well as the control agent. Drugs that elicit an immune response are ideal for crossover designs. More discussion on this topic, including drawbacks and limitations of crossover designs, can be found in Louis et al. Consider a randomized study to compare treatment groups, again with a continuous outcome variable. In practice, the design is implemented by assuming some initial values, then estimating the parameters using the available data up to the first cohort of patients, finally plugging them into Equation 3A.4 to find the optimal allocation proportion to treatment X for the cohort of patients.