The four types of observer bias identified in the study are "reactivity," which involves research subjects reacting to the presence of an observer, especially if the subject knows that his/her behavior is under observation; "culture shock," which refers to feelings of disorientation and anxiety that observers feel during the early stages of field research; "going native," which involves a process of conversion or resocialization of the observer during fieldwork; and "burnout, which refers to inaccurate documentation of observational data at the later stages of fieldwork due to the mentally and physically demanding nature of data collection. An in-depth review of qualitative data on reactivity for patrol officers involved in the Project on Policing Neighborhoods revealed that reactivity was often embedded within social exchanges between officers and observers; these often involved social cues from patrol officers or explicit changes in officer behavior. One key finding from the qualitative analysis was that the level of reactivity within observational data (in the form of social cues or explicit changes in patrol officer behavior) depended on the specificity of the data. Multivariate analyses were conducted to determine whether the qualitative coding or other potential sources of reactivity derived from the field research literature (i.e., observer sex, status congruency, and time in the field) were isolated instances or involved a systematic effect on patrol officer behavior in the form of significant effects within multivariate equations. Multivariate analyses were conducted at three units of analysis and examined multiple aspects of patrol officer behavior. Encounter-level multivariate analyses addressed the patrol officer's decision to arrest and the use of force. Ride-level multivariate analyses examined the level of aggressive patrol and the amount of "goofing off" in which an officer engaged in per shift. Ride segment multivariate analyses focused on variation in patrol officer behavior in the context of a ride. Selected findings from the multivariate analyses indicate that patrol officers who expressed concerns about safety were less likely to arrest suspects; patrol officers were more likely to use force against citizens if the observer helped the police officer in some capacity over the course of the shift; and patrol officers were less likely to use force against citizens if they were less familiar with the observer. Further, patrol officer behavior during the first hour of the shift was significantly different from his/her behavior during the rest of the shift. Overall, the study indicates that reactivity could act as a systematic bias and mask or alter the true relationships between independent and dependent variables, leading to mistaken inferences being drawn from all studies that use observational data. Suggestions are offered for how studies might reduce the bias associated with officer reactivity. 98 tables, 4 figures, and 143 references