In the study of Example 15.1, independent random samples of six advertisements from True Confessions, People Weekly, and Newsweek were taken. The fog indices for these advertisements are given in the accompanying table. Test the null hypothesis that the population mean fog indices are the same for advertisements in these three magazines and compute the minimum significant difference and indicate which subgroups have different means.

True Confessions 12.89 12.69 11.15 9.52 9.12 7.04 People Weekly 9.50 8.60 8.59 6.50 4.79 4.29 Newsweelk 10.21 9.66 7.67 5.12 4.88 3.12