Quasi‐Experiments

Abstract

Abstract When randomized controlled trials are not feasible, quasi‐experiments are the second‐best alternative for estimating causal treatment effects. Quasi‐experiments are characterized by an active intervention, but in contrast to randomized experiments, they lack random assignment of treatments to subjects. Instead, subjects deliberately select or are systematically assigned to treatment conditions. According to the type of the assignment or selection mechanism, four basic quasi‐experimental designs can be distinguished: regression discontinuity design, interrupted time series design, nonequivalent control group design, and instrumental variable design. This entry reviews the basic designs, analytic strategies, and the assumptions required for estimating causal treatment effects from quasi‐experiments.