Consider a crossover trial design in which subjects are randomized to either 5 weeks of the PUFA diet followed by 5 weeks of the SFA diet, or 5 weeks SFA followed by 5 weeks PUFA. The mean within-subject differences between insulin sensitivity at the end of week 5 of the PUFA and SFA diet periods will be estimated. 14. Using the sample size formula in terms of within-subject variance, compute the necessary total sample size required to have 80% power at the 5% significance level to detect a difference of the size seen in Summers et al. How does this compare to the sample sizes for the parallel arm designs