With reference to Problem 25, our psychologist now wishes to change the nature of the reaction task and to compare mean reaction time on the original task with that on the altered task. He assumes that the standard deviation will be the same on both tasks and wants to estimate at the 95% level of confidence the difference between the two population means. The psychologist wants this difference to be within 5 msec of the difference between the sample means.

(a) If the psychologist selects two random samples, what should be the size of each?

(b) If he uses subjects matched on a pretest of reaction time and is estimated to be .40 between pairs of measurements, what should be the size of each sample?