How Natural are Naturalistic Observations?

Abstract

Using a modification of Barker's (1965) operator-transducer system for conceptualizing data collection models, we raised question about whether the context in which behavioral assessors typically conduct naturalistic observations is truly natural. It was suggested that under most circumstances this context is not free of the influence and interference of the investigator. Consequently, subject reactivity is possible and there is potential for distortion or unrepresentativeness of the behavior under study. Three types of subject reactivity were described: (1) reactivity to the presence of the observer, which has been extensively investigated and demonstrated, (2) reactivity produced by changes in the conditions of observation brought about by the investigator when subject behaviors have undesirable effects on the observer or investigator, (3) and reactivity to the conditions of observation. The second and third types have not received empirical attention. Data from a parent training outcome study were used to demonstrate the three types of subject reactivity noted by observers while collecting naturalistic observation data in family homes. Finally, some suggestions were made for future research, including the systematic collection of reactivity data during the course of observations, and the examination of covariations between subject reactivity behaviors and dependent variables.