1. Use the sample_size program to determine sample size needed to estimate a mean from normal data for
desired alpha=0.05 and precision=0.10, observed CV=0.8, and from an infinite population (make N=10E06).
2. Allocate the sample you just estimated to 5 strata under optimal allocation, where per-stratum areas are c(100,200,300,150,50), per-stratum standard deviations are c(50,10,10,10,5), and costs per sample are equal among strata.
3. Simulate data for Poisson counts from a population mean of 5. Find (by trial and error) sample size to achieve a CV<0.05 for the mean.
4. Repeat the above but with a random effect sigma =0.8. What does this do to sample size needed?