differenceindifference

Difference-in-differences, commonly abbreviated as DiD, is a quasi-experimental research design used to estimate the causal effect of a treatment or policy. It compares how an outcome changes over time in a group exposed to the intervention (the treated) to how it changes in a similar group that is not exposed (the control).

The core idea is that, in the absence of the treatment, the treated and control groups would have followed parallel trends. By examining the differential change after the intervention, researchers can attribute the difference to the treatment rather than to other factors affecting both groups. A typical implementation uses panel data or repeated cross-sections and regresses the outcome on group and time indicators, along with an interaction term between treatment status and a post-treatment period. The coefficient on this interaction estimates the average treatment effect on the treated (ATT).

Key assumptions include parallel trends in the absence of treatment, no spillovers between units, and no anticipation of the policy. These are often assessed with pre-treatment trends and robustness checks such as placebo tests, sensitivity analyses, or alternative control groups. It is important to ensure the control group would have mirrored the treated group’s trajectory if the intervention had not occurred.

Extensions cover generalized DiD with multiple periods and staggered adoption, as well as event-study designs that visualize dynamic effects. Standard two-way fixed effects estimators can be biased when treatment effects vary over time or across groups, leading to alternative estimators and robust inference methods. Limitations include sensitivity to violations of assumptions, potential spillovers, and challenges in selecting an appropriate control group. Despite these caveats, DiD remains a widely used tool for causal inference in economics and policy analysis.