Subjects were exposed to substance X on day 2 of 3 days. 6 measurements were taken on each of the 3 days at the same times each day. Some of the measurements were taken after exercise (at hours 2 and 4). Some subjects are smokers and others not. I am interested in the main effect of Day (as a surrogate for exposure to X), the main effect of hour (as a surrogate for how the outcome variable naturally varies by time-of-day and in response to exercise) and the Day x Hour interaction. But I am also interested in effect measure modification by smoking status. Should I do 2 repeated measures ANOVAs (one for smokers, one for non-smokers) or is there a way to incorporate smoking status from the outset in a single analysis?