This chapter discusses the use of replications in order to enhance the ability of a researcher to make causal inferences from data gathered in field (non-experimental) settings. Replications are common in experimental research and are used both to study interactions between treatments and other variables (e.g., gender or aptitude in two-way designs) and to increase the power of the design. In many of these designs, each condition of a blocking variable often constitutes a self-contained true experiment (Campbell & Stanley, 1963), with internal randomization. In such research, the data are processed statistically to study generalizability of treatment effects across replications.
I argue that replicating even weak research designs can greatly expand the usefulness of research in settings, such as field research, where experimental control is difficult or even impossible to achieve. Some examples of this approach, called replicated field study design, are given and some suggestions for data collection and analysis are given.