1. Choose two ads, one that is significant and one that is not. Verify significance based on the average, standard error, and n, to make sure that they are correct. Is it appropriate to use one-sided tests here?
2. If the type I error is supposed to be controlled at 5%, how is it that in the computer simulation model, type I errors occurred 70% of the time?
3. Could it reasonably be that no ads are worthwhile, in a study for which 2 of 22 are significant?
4. What is your interpretation of the effectiveness of the ads in this study? What would you recommend in this situation?