I have used the function power.t.test() (power calculations for one and two sample t tests) but have noted that it handles the samples as being the same size. How does one handle sample sizes that are different. In other words how does one handle unbalanced designs. Farrel Buchinsky