Not all the questions I set for myself could be appropriately answered by the data collected by participant observation alone. Among the questions I was interested in was the following: How consequential are negative sanctions to an individual's career path? To get at this question, I had to ask: How do attending physicians construct an evaluation of a house officer's competence from observations of the house officer's day-to-day task performance? I used two sources of data to supplement my field data: (1) I perused the written evaluations of house staff by attending surgeons as contained in personnel files and (2) I attended the faculty meetings of the Department of Surgery at which the decision to retain or terminate junior house staff in the training programme is made.
Once fieldwork was completed, I interviewed those attending physicians and house staff with whom I worked most closely. The interviews served two purposes. First, they served as a check on the validity of my data. Many of my interpretations were re-phrased as questions for comment. In this way, I could match my assessments against those of actors in the scene; and I could fill in those spots where my observational material was thin. Second, the interviews allowed me to see my observations in a larger context. For example, house staff were asked to compare their performance on different services. This provided me with a fuller picture of their activity and their reaction to it than I could gather from my field observations. Two different interview schedules were constructed: one for house staff and one for attending physicians. These schedules serve as guides rather than formal instruments. All interviews were kept as conversational as possible. Interviews lasted between forty-five minutes and one and a half hours. For attending physicians the focus of the interview was on how they decided if a subordinate [house staff] was good or bad and what they considered to be unforgivable errors. Of house staff, I asked what they thought their superordinate's [attending physicians] performance expectations were, what they considered to be unforgivable errors, and what they considered the major difficulties in becoming a surgeon.
This triangulated approach to data gathering gives me confidence in the validity of the inferences which I have drawn. Often conclusions reached about performance from one data source are confirmed by another.