Question: Problem 7.3: On the basis of a review of large amounts of data, it is well established that the variance of population of ratings of the quality of a particular wine is 12.64. A new method for training raters is established in hopes of reducing the variance. In a sample of 10 judges trained under the new method, s² is 3.51.

(a) Would you conclude that the new method has effectively reduced the variance of ratings?

(b) Suppose that the population of scores was not normally distributed. Why is this a problem for the approach in (a)? Would it be less of a problem in the sample size were larger?