Your Advisor Conducted a Study on Decision Making Using a 2

Your advisor conducted a study on decision making using a 2 (mood prime: negative, positive) x 2 (physiological arousal: adrenaline injection, saline solution injection) factorial design, and had a final sample of 80 participants. However, upon further investigation, you realized that half of participants, randomly chosen, were administered a questionnaire about their attitudes toward authority figures before the primary dependent measure, and half completed the attitudinal questionnaire after the primary dependent measure. What sort of design would this most likely be? In this example, is it possible to obtain a significant three-way interaction, even if there are no significant main effects or two-way interactions? Do you have sufficient power to obtain a significant three-way interaction? How reliable would you expect the results of a three-way interaction test to be? Is it ever appropriate to divide participants into additional groups or categories after all data have been collected - why or why not?