17. Soap experiment, continued In Example 3.6.2, page 53, a sample size calculation was made for the number of observations needed to detect, with probability * (0.25) = 0.90, a difference in weight loss of at least A = 0.25 grams in v = 3 difference types of soap, using an analysis of variance with a probability of o = 0.05 of a Type I error. The calculation used an estimate of 0.007 grams for o and showed that r = 4 observations were needed on each type of soap. The experiment was run with r = 4, and the least squares estimate for o' was 0.0772. If the true value for o was, in fact, 0.08, what power did the test actually have for detecting a difference of A = 0.25 grams in the weight loss of the three soaps