When a Control is not in Control: The Effect of Indirect Treatment Exposure in Experimental Research

• In Event: Police Use of Cameras

Abstract

Experimental research often assumes that after randomization occurs, equivalence established between treatment and control groups is stable and endures throughout the trial period. Observations of the control group are assumed to be unaffected by treatment administered to the treatment group. However, this assumption may be violated due to contamination as a result of control subjects receiving indirect exposure to treatment over time. Using data from a field experiment conducted with a large police department in the Southwestern United States (Mesa,AZ), we measure the indirect treatment exposure of control officers not wearing on-officer video cameras from contact with treatment officers assigned to wear the cameras. We hypothesize that over time the control group becomes increasing like the treatment group because of contamination. Contamination effects would highlight a significant threat to the internal validity of randomized controlled trials (RCTs) conducted among police officers within the same patrol precinct. This paper serves as one of the first studies in experimental criminology to try to study indirect treatment effects by measuring control subjects’ exposure to treatment officers, excluding treatment subjects from analysis, and testing for significance in outcomes based on variability in the control subjects’ indirect exposure coefficient (IEC).

Authors

• Jacob Young, Arizona State University

• Justin Ready, Arizona State University

• William Sousa, University of Nevada, Las Vegas