Robust data-driven inference in the regression-discontinuity design

Sebastian Calonico, Matias D. Cattaneo, Rocío Titiunik

Research output: Contribution to journalArticle

109 Citations (Scopus)

Abstract

In this article, we introduce three commands to conduct robust data-driven statistical inference in regression-discontinuity (RD) designs. First, we present rdrobust, a command that implements the robust bias-corrected confidence intervals proposed in Calonico, Cattaneo, and Titiunik (2014d, Econometrica 82: 2295–2326) for average treatment effects at the cutoff in sharp RD, sharp kink RD, fuzzy RD, and fuzzy kink RD designs. This command also implements other conventional nonparametric RD treatment-effect point estimators and confidence intervals. Second, we describe the companion command rdbwselect, which implements several bandwidth selectors proposed in the RD literature. Following the results in Calonico, Cattaneo, and Titiunik (2014a, Working paper, University of Michigan), we also introduce rdplot, a command that implements several data-driven choices of the number of bins in evenly spaced and quantile-spaced partitions that are used to construct the RD plots usually encountered in empirical applications. A companion R package is described in Calonico, Cattaneo, and Titiunik (2014b, Working paper, University of Michigan).

Original languageEnglish (US)
Article numberst0366
Pages (from-to)909-946
Number of pages38
JournalStata Journal
Volume14
Issue number4
StatePublished - Jan 1 2014

Fingerprint

Data-driven
Discontinuity
Regression
Kink
Confidence interval
Average Treatment Effect
Fuzzy Regression
Selector
Design
Treatment Effects
Nonparametric Regression
Quantile
Statistical Inference
Bandwidth
Partition
Estimator

Keywords

  • Bandwidth selection
  • Bias correction
  • Fuzzy kink RD
  • Fuzzy RD
  • Local polynomials
  • RD plots
  • Rdbwselect
  • Rdplot
  • Rdrobust
  • Regression discontinuity (RD)
  • Sharp kink RD
  • Sharp RD
  • st0366
  • Treatment effects

ASJC Scopus subject areas

  • Mathematics (miscellaneous)

Cite this

Calonico, S., Cattaneo, M. D., & Titiunik, R. (2014). Robust data-driven inference in the regression-discontinuity design. Stata Journal, 14(4), 909-946. [st0366].

Robust data-driven inference in the regression-discontinuity design. / Calonico, Sebastian; Cattaneo, Matias D.; Titiunik, Rocío.

In: Stata Journal, Vol. 14, No. 4, st0366, 01.01.2014, p. 909-946.

Research output: Contribution to journalArticle

Calonico, S, Cattaneo, MD & Titiunik, R 2014, 'Robust data-driven inference in the regression-discontinuity design', Stata Journal, vol. 14, no. 4, st0366, pp. 909-946.
Calonico S, Cattaneo MD, Titiunik R. Robust data-driven inference in the regression-discontinuity design. Stata Journal. 2014 Jan 1;14(4):909-946. st0366.
Calonico, Sebastian ; Cattaneo, Matias D. ; Titiunik, Rocío. / Robust data-driven inference in the regression-discontinuity design. In: Stata Journal. 2014 ; Vol. 14, No. 4. pp. 909-946.
@article{dae5e8f1b26d4d1e83769f5fd71ea928,
title = "Robust data-driven inference in the regression-discontinuity design",
abstract = "In this article, we introduce three commands to conduct robust data-driven statistical inference in regression-discontinuity (RD) designs. First, we present rdrobust, a command that implements the robust bias-corrected confidence intervals proposed in Calonico, Cattaneo, and Titiunik (2014d, Econometrica 82: 2295–2326) for average treatment effects at the cutoff in sharp RD, sharp kink RD, fuzzy RD, and fuzzy kink RD designs. This command also implements other conventional nonparametric RD treatment-effect point estimators and confidence intervals. Second, we describe the companion command rdbwselect, which implements several bandwidth selectors proposed in the RD literature. Following the results in Calonico, Cattaneo, and Titiunik (2014a, Working paper, University of Michigan), we also introduce rdplot, a command that implements several data-driven choices of the number of bins in evenly spaced and quantile-spaced partitions that are used to construct the RD plots usually encountered in empirical applications. A companion R package is described in Calonico, Cattaneo, and Titiunik (2014b, Working paper, University of Michigan).",
keywords = "Bandwidth selection, Bias correction, Fuzzy kink RD, Fuzzy RD, Local polynomials, RD plots, Rdbwselect, Rdplot, Rdrobust, Regression discontinuity (RD), Sharp kink RD, Sharp RD, st0366, Treatment effects",
author = "Sebastian Calonico and Cattaneo, {Matias D.} and Roc{\'i}o Titiunik",
year = "2014",
month = "1",
day = "1",
language = "English (US)",
volume = "14",
pages = "909--946",
journal = "Stata Journal",
issn = "1536-867X",
publisher = "DPC Nederland",
number = "4",

}

TY - JOUR

T1 - Robust data-driven inference in the regression-discontinuity design

AU - Calonico, Sebastian

AU - Cattaneo, Matias D.

AU - Titiunik, Rocío

PY - 2014/1/1

Y1 - 2014/1/1

N2 - In this article, we introduce three commands to conduct robust data-driven statistical inference in regression-discontinuity (RD) designs. First, we present rdrobust, a command that implements the robust bias-corrected confidence intervals proposed in Calonico, Cattaneo, and Titiunik (2014d, Econometrica 82: 2295–2326) for average treatment effects at the cutoff in sharp RD, sharp kink RD, fuzzy RD, and fuzzy kink RD designs. This command also implements other conventional nonparametric RD treatment-effect point estimators and confidence intervals. Second, we describe the companion command rdbwselect, which implements several bandwidth selectors proposed in the RD literature. Following the results in Calonico, Cattaneo, and Titiunik (2014a, Working paper, University of Michigan), we also introduce rdplot, a command that implements several data-driven choices of the number of bins in evenly spaced and quantile-spaced partitions that are used to construct the RD plots usually encountered in empirical applications. A companion R package is described in Calonico, Cattaneo, and Titiunik (2014b, Working paper, University of Michigan).

AB - In this article, we introduce three commands to conduct robust data-driven statistical inference in regression-discontinuity (RD) designs. First, we present rdrobust, a command that implements the robust bias-corrected confidence intervals proposed in Calonico, Cattaneo, and Titiunik (2014d, Econometrica 82: 2295–2326) for average treatment effects at the cutoff in sharp RD, sharp kink RD, fuzzy RD, and fuzzy kink RD designs. This command also implements other conventional nonparametric RD treatment-effect point estimators and confidence intervals. Second, we describe the companion command rdbwselect, which implements several bandwidth selectors proposed in the RD literature. Following the results in Calonico, Cattaneo, and Titiunik (2014a, Working paper, University of Michigan), we also introduce rdplot, a command that implements several data-driven choices of the number of bins in evenly spaced and quantile-spaced partitions that are used to construct the RD plots usually encountered in empirical applications. A companion R package is described in Calonico, Cattaneo, and Titiunik (2014b, Working paper, University of Michigan).

KW - Bandwidth selection

KW - Bias correction

KW - Fuzzy kink RD

KW - Fuzzy RD

KW - Local polynomials

KW - RD plots

KW - Rdbwselect

KW - Rdplot

KW - Rdrobust

KW - Regression discontinuity (RD)

KW - Sharp kink RD

KW - Sharp RD

KW - st0366

KW - Treatment effects

UR - http://www.scopus.com/inward/record.url?scp=85001085842&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85001085842&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:85001085842

VL - 14

SP - 909

EP - 946

JO - Stata Journal

JF - Stata Journal

SN - 1536-867X

IS - 4

M1 - st0366

ER -