Question: An experimental psychologist at a Veterans Administration hospital obtained approval from the Institutional Review Board to conduct a study of the efficacy of Cymbalta for treating depression. The experiment lasted 60 days, during which one group of depressed patients was given placebos, one was given a low dose of Cymbalta, and one was given a high dose. At the end of the experiment, the patients were observed by two outside psychologists who rated several behaviors such as eye contact, posture, verbal output, and activity for degree of depression. The ratings went into a composite score for each patient, with high scores indicating depression. The means were: placebo, 18.86; low dose, 10.34; and high dose, 16.21. An ANOVA summary table and two Tukey HSD tests are shown in the accompanying table

Source                                df              F

Between treatments              2             21.60

Error                                  18

HSD (placebo vs. low)    =   13.14

HSD (placebo vs. high)   =    0.65