Quasi-experimental designs for causal inference: Addressing threats to validity from a graphical models perspective

In the developmental sciences, and social sciences in general, we are often interested in determining the effect of a specific intervention on a specific outcome. While randomized control trials allow for the investigation of such causal questions, the randomization of participants into particular conditions is often infeasible or unethical. Quasi-experimental designs provide an alternative, in which participants select their preferred condition but researchers retain some control over the implementation of a treatment. In this chapter, we use causal graphs to introduce the rationale of specific quasi-experimental designs and to discuss their causal assumptions. Causal graphs are a formal language for explicitly stating subject matter theory about a causal process and its implied causal assumptions. Additionally, we discuss potential validity threats in these designs that can reduce the credibility of causal claims.