12. For g groups with large sample sizes, we plan to compare sinultaneously all pairs of popu- lation means. We want the probability to equal at least 80 that the entire set of confidence intervals contain the true differences.

a) If g 10, which tabled t-score should we use for each interval?

b) Which 1-score should we use if g = 5 TABLE 12.31 Analysis of Variance Procedure Dependent variable: TIME Sum of Source DF Squares Model 2 72.000 Mean Square 36.000 F value 0.79 Pr > F 0.4813 Error 9 408.000 45.333 Total 11 480.000 Bonferroni T tests for variable: TIME Alpha 0.05 df= 9 MSE= 45.33333 Critical value of T- 2.93 Minimum Significant Difference= 13.965 Means with the same letter are not significantly different. Bon Grouping Mean N GROUP A 18.000 4 3 A A 15.000 4 2 A A 12.000 4 1

c) Describe how the t- (or 2-) score depends on the number of groups. and explain the implication regarding width of the intervals.