Regression Discontinuity Design (RDD)
Also known as: RDD, regression discontinuity, sharp regression discontinuity, Regresyon Süreksizliği Tasarımı (RDD)
Regression Discontinuity Design is a quasi-experimental method that estimates a local causal effect around a threshold (cutoff) value, comparing units just below and just above the cutoff as if they were almost randomly assigned. It is the design developed for applied practice by Imbens and Lemieux (2008) and by Lee and Lemieux (2010).
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use RDD when assignment to treatment is governed by a genuine, non-manipulable threshold on a continuous running variable, and you have enough observations clustered around that cutoff (typically at least about 100, with adequate density near the threshold). It fits cross-sectional and panel data with continuous or binary outcomes. The design is credible only when units cannot precisely control which side of the cutoff they fall on, when unobserved characteristics evolve smoothly through the cutoff, and when the bandwidth is chosen by a principled rule rather than by hand.
Strengths & limitations
- Delivers a credible causal effect from observational data by exploiting a natural, rule-based cutoff, with weaker assumptions than most matching designs.
- Comparing units just on either side of the threshold mimics random assignment locally, so the local effect is transparent and easy to defend.
- The discontinuity can be inspected visually, making the identifying assumption unusually intuitive to communicate.
- Estimates a local effect at the cutoff only; it does not generalise to units far from the threshold.
- Requires enough observations densely packed around the cutoff, so the effective sample for estimation can be small even in a large dataset.
- If individuals can manipulate the running variable to cross the cutoff, the design breaks down and the effect is biased.
- Results are sensitive to the chosen bandwidth and to the order of the local polynomial.
Frequently asked
What is the running variable in RDD?
It is the continuous variable that determines treatment assignment, such as a test score, income, or age. Treatment switches on the moment this variable crosses the cutoff, and the design compares outcomes just below and just above that point.
Why do I need the McCrary density test?
RDD assumes individuals cannot precisely manipulate the running variable to land on the favourable side of the cutoff. The McCrary test checks for bunching just above or below the threshold; a sharp jump in density signals manipulation and threatens the design's validity.
How is the bandwidth chosen?
The bandwidth sets how wide a window around the cutoff is used for estimation. It should be selected with a data-driven rule such as the IK or CCT method rather than by hand, because the estimate is sensitive to this choice.
Does RDD give an effect for everyone?
No. RDD identifies a local effect at the cutoff for units near the threshold. It does not tell you what the treatment would do for units far from the cutoff, so the estimate should not be extrapolated to the whole population.
Sources
- Imbens, G. W., & Lemieux, T. (2008). Regression Discontinuity Designs: A Guide to Practice. Journal of Econometrics, 142(2), 615-635. DOI: 10.1016/j.jeconom.2007.05.001 ↗
- Cattaneo, M. D., Idrobo, N., & Titiunik, R. (2020). A Practical Introduction to Regression Discontinuity Designs: Foundations. Cambridge University Press. DOI: 10.1017/9781108684606 ↗
- Lee, D. S., & Lemieux, T. (2010). Regression Discontinuity Designs in Economics. Journal of Economic Literature, 48(2), 281-355. DOI: 10.1257/jel.48.2.281 ↗
How to cite this page
ScholarGate. (2026, June 1). Regression Discontinuity Design (RDD). ScholarGate. https://scholargate.app/en/econometrics/rdd
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Difference-in-DifferencesEconometrics↔ compare
- Instrumental Variables in Health ResearchHealth Economics↔ compare
- OLS RegressionEconometrics↔ compare
- Panel Fixed EffectsEconometrics↔ compare
- Propensity Score MatchingResearch Statistics↔ compare