Abstract
External validity is questionable whenever generalizations are made over any facet sampled selectively in a design, including experimenters. Sampling experimenters selectively is particularly common in research on outcome of therapy in which typically only volunteer therapists are included. Using archival data capturing a 6-mo. period, the present study yielded differences between therapists who did (n = 5) and did not (n = 9) volunteer to serve as experimenters in a study of outcome on three variables, number of “no show” clients, average total number of sessions clients were seen, and age of experimenter. These findings suggest the importance of considering hidden interactions between therapist volunteer status and experimental treatments, interactions which would limit the external validity of results of study of outcome.