Propensity Score Matching

Definition

A quasi-experimental technique that estimates the causal effect of a treatment by matching treated and untreated units with similar propensity scores — the estimated probability of receiving treatment given observed covariates. By balancing the distribution of covariates between groups, propensity-score methods mimic aspects of a randomised experiment. The approach is widely used in policy evaluation and health economics.

Why It Matters

Randomised controlled trials are not always feasible or ethical. Propensity score matching provides a principled way to approximate experimental conditions with observational data, reducing selection bias and enabling more credible causal inferences. However, it can only balance on observed covariates — unmeasured confounders remain a threat.

Example

A policy analyst evaluates a job-training programme by matching each programme participant with a non-participant who has a similar propensity score (based on age, education, prior earnings, and region). After matching, the difference in average earnings between the two groups approximates the programme's causal effect.

Related Terms

• Difference-in-Differences (DiD)
• Instrumental Variables (IV)
• Regression Discontinuity Design (RDD)

Software Notes

• SPSS: Not built-in; use the R plugin or the PSMATCHING extension
• R: MatchIt package; e.g., matchit(treat ~ x1 + x2, data = df, method = "nearest")
• Stata: teffects psmatch command; e.g., teffects psmatch (y) (treat x1 x2)