quasiexperiment

Quasi-experiment, or quasi-experimental design, is a research approach used to evaluate causal effects when random assignment to treatment and control groups is not feasible. In a quasi-experiment, the intervention is observed in real-world conditions, and treatment exposure is determined by policy decisions, natural variation, or institutional processes rather than by randomization. Although these designs aim to infer causality, they are more vulnerable to confounding than true randomized experiments.

Common designs include non-equivalent control group designs, interrupted time series, regression discontinuity designs, and natural experiments. Examples include evaluating a new curriculum introduced in some schools but not others, or assessing the effect of a policy enacted in one region with a comparable region serving as a comparison.

Quasi-experiments enable causal inference when randomized trials are not possible and can offer reasonable external validity, but they face threats to internal validity. Selection bias, history, maturation, instrumentation, and regression to the mean can distort results if not properly addressed.

Analytic strategies to bolster validity include propensity score matching, difference-in-differences, interrupted time series analyses, regression discontinuity analyses, synthetic control methods, and instrumental variable approaches.

Quasi-experiments are widely used in education, public policy, health, and economics to evaluate interventions under real-world conditions.