Economics
Saved & recent

Causal identification

Identification states what data and assumptions allow a causal effect to be learned. More precise associations cannot repair a design whose comparison group fails to represent the counterfactual.

Read first
Counterfactual outcomes, Correlation and causation
Related
Randomized experiments, Natural experiments, Regression and control variables
Further reading
World Bank: Research design for impact evaluation
U prior need T Y training later earnings target effect comparison within a need stratum training no training same measured prior need
Conceptual adjustment design for applicants. Identification additionally requires consistent treatment and outcome definitions and appropriate weighting for the target effect. The arrows encode assumptions; sample size cannot establish that no relevant confounder was omitted.

View in the full Economics collection