An organizational psychologist wants to examine if the new training program significantly lowers the number of errors made during data entry tasks. In the past, there have been a mean of 12 errors per 500 words, with standard deviation of 10. It is believed that the program will decrease errors to an average of 6 which would create a considerable time savings (as less proofreading and correcting would then be required). State null and research hypotheses. If tested with a significance level of .05 (and assuming a normal distribution), what is the estimated power of the test to detect the hypothesized decrease of 6 defects if 20 subjects were used? What if 40 subjects were used?