STATS 2B03 Lecture Notes - Analysis Of Variance, Null Hypothesis
Document Summary
October 30th, 2012 in general, we are first going to test h0= 1 = 2 = = k, ha: i j for at least one pair (i, j) Reject h0 if fk-1,n-k>fk-1,n-k,1- in our example: 4. 4>3. 03 calculations for the above f statistic summarized in an anova table. 3. 891/0. 8828 if h0 is not rejected, then we conclude that there are not significant differences between any of the pairs of means and analysis would stop. If h0 is rejected, it means that at least one of the hypotheses h0: 1= 2, 1= 3, 1= 4 is rejected. Calculate: x-. i x-. j (testing how equal they are), if difference is large, population sample averages are different. Reject h0 if |x-. i x-. j| >hsd*, where hsd* is . A 100(1 )% turkey ci for 1- 2 is x-. i x-. j hsd*