The paper describes an unacknowledged artifact that may confound experimental evaluations of innovations. The paper hypothesizes that control group members (teachers, pupils, etc.) perceiving the consequences of an innovation as threatening to their job, salary, status, or traditional patterns of working, may perform atypically and confound the evaluation outcomes. Demand characteristics within the social psychology of the experiment provide the theoretical framework. The paper compares similarities and differences among the John Henry Effect and other research-biasing factors. Four evaluation studies in which the John Henry Effect was manifested are described. Alternative evaluation designs for the artifact's control are discussed. (Author)