Suppose you are interested in the behaviors of physicians that have high ratings of patient
satisfaction. The research goal is to identify the behaviors in the natural clinical settings of these
successful physicians so that these behaviors can be built into the curricula of medical preparation
programs. The main data were collected by the video recording of five randomly selected physicianpatient
encounters from 10 physicians who have been rated highly by patients in a reliable
In analyzing the recordings, what would you define as the unit for analysis? Why? How many data
units (in rough estimates) are you likely to get based on this decision? Does the estimated number of
data units seem adequate? Why or why not?