Rdrobust: Software for Regression-discontinuity Designs

Calónico, Sebastián; Cattaneo, Matias D.; Farrell, Max H.; Titiunik, Rocío

doi:10.1177/1536867x1701700208

Cited by 700 publications

(545 citation statements)

References 21 publications

Supporting

Mentioning

529

Contrasting

Unclassified

Order By: Relevance

“…Table presents the results from the nonparametric local polynomial analysis in the Head Start case study, estimated using the software packages described in Calonico et al. (). The table presents results from two analyses, one based on a local constant (

p = 0

) and the other on a local linear (

p = 1

) polynomial approximation.…”

Section: Rd Based On Continuity At the Cutoffmentioning

confidence: 99%

“…Adopting a local randomization framework leads to similar conclusions: we estimate treatment effects of −2.3 and −2.5 deaths per 100,000, we reject the (sharp) null hypothesis that the treatment has no effect for any unit with randomization‐based p ‐values below 0.01, and we show that these findings are robust to window selection, parametric misspecification, local interference, and the presence of unobserved confounders assessed with our newly developed sensitivity analysis methods. All the methods are implemented in publicly available R and Stata software packages (see Calonico et al., ; Calonico, Cattaneo, & Titiunik, , ; Cattaneo, Titiunik, & Vazquez‐Bare, ). Accompanying this article, we also provide data and complete replication codes in both R and Stata.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Comparing Inference Approaches for RD Designs: A Reexamination of the Effect of Head Start on Child Mortality

Cattaneo¹,

Titiunik²,

Vázquez-Baré³

2017

J. Pol. Anal. Manage.

Self Cite

137

145

View full text Add to dashboard Cite

The regression discontinuity (RD) design is a popular quasi‐experimental design for causal inference and policy evaluation. The most common inference approaches in RD designs employ “flexible” parametric and nonparametric local polynomial methods, which rely on extrapolation and large‐sample approximations of conditional expectations using observations somewhat near the cutoff that determines treatment assignment. An alternative inference approach employs the idea of local randomization, where the very few units closest to the cutoff are regarded as randomly assigned to treatment and finite‐sample exact inference methods are used. In this paper, we contrast these approaches empirically by re‐analyzing the influential findings of Ludwig and Miller (), who studied the effect of Head Start assistance on child mortality employing parametric RD methods. We first review methods based on approximations of conditional expectations, which are relatively well developed in the literature, and then present new methods based on randomization inference. In particular, we extend the local randomization framework to allow for parametric adjustments of the potential outcomes; our extended framework substantially relaxes strong assumptions in prior literature and better resembles other RD inference methods. We compare all these methods formally, focusing on both estimands and inference properties. In addition, we develop new approaches for randomization‐based sensitivity analysis specifically tailored to RD designs. Applying all these methods to the Head Start data, we find that the original RD treatment effect reported in the literature is quite stable and robust, an empirical finding that enhances the credibility of the original result. All the empirical methods we discuss are readily available in general purpose software in R and Stata; we also provide the dataset and software code needed to replicate all our results.

show abstract

p = 0

) and the other on a local linear (

p = 1

) polynomial approximation.…”

Section: Rd Based On Continuity At the Cutoffmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Comparing Inference Approaches for RD Designs: A Reexamination of the Effect of Head Start on Child Mortality

Cattaneo¹,

Titiunik²,

Vázquez-Baré³

2017

J. Pol. Anal. Manage.

Self Cite

137

145

View full text Add to dashboard Cite

show abstract

“…The estimation is implemented by the Stata package rdrobust (Calonico et al ). It offers a whole array of options for bandwidth selection as well as the order of the polynomial approximation.…”

mentioning

confidence: 99%

Life After Debt: Postgraduation Consequences of Federal Student Loans

Gervais¹,

Ziebarth

2019

Economic Inquiry

View full text Add to dashboard Cite

We estimate the causal effect of student loans on postgraduation labor market outcomes exploiting a kink in the formula determining eligibility for need‐based student loans. Using a representative sample of students graduating with a bachelor's degree in 1993, we find that student debt has nonnegative effects on earnings. This result holds with differing levels of statistical significance across a battery of different empirical designs: (1) ordinary least squares, (2) partially linear, and (3) regression kink. We find similar results for the 2008 graduating cohort. (JEL I22, I26, J32, J33)

show abstract

“…The treatment effect is estimated as the difference between the intercepts in these two regressions,

\overset{τ}{true} = {\overset{truê}{α}}_{1} - {\overset{truê}{α}}_{0}

, as this corresponds to the difference in predicted values of the outcome variable for treated and nontreated cases either side of the cut‐off at the cut‐off boundary. The Stata program rdrobust (Calonico, Cattaneo, Farrell, & Titiunik, ) is used to produce bias‐corrected point estimates with accompanying robust standard errors. The bandwidths for each local linear regression are selected using the optimal data‐driven method as per Calonico, Cattaneo, and Titiunik ().…”

Section: Methodsmentioning

confidence: 99%

The effect of personalised weight feedback on weight loss and health behaviours: Evidence from a regression discontinuity design

Cook

2018

Health Economics

View full text Add to dashboard Cite

Using a regression‐discontinuity approach on a U.K. longitudinal dataset, this research analyses whether personalised weight feedback resulted in individuals losing weight over a period of between 2 and 7 years. The analysis presented here finds that being told one was “overweight” had, on average, no effect on subsequent weight loss; however, being told one was “very overweight” resulted in individuals losing, on average, approximately 1% of their bodyweight. The effect of feedback was found to be strongly moderated by household income, with those in higher income households accounting for seemingly all of the estimated effect due, in part, to increased physical activity. These findings suggest that the provision of weight feedback may be a cost‐effective way to reduce obesity in adults. They do however also highlight that the differential response to the provision of health information may be a driver of health inequalities and that the provision of feedback may bias longitudinal health studies.

show abstract

Rdrobust: Software for Regression-discontinuity Designs

Cited by 700 publications

References 21 publications

Comparing Inference Approaches for RD Designs: A Reexamination of the Effect of Head Start on Child Mortality

Comparing Inference Approaches for RD Designs: A Reexamination of the Effect of Head Start on Child Mortality

Life After Debt: Postgraduation Consequences of Federal Student Loans

The effect of personalised weight feedback on weight loss and health behaviours: Evidence from a regression discontinuity design

Contact Info

Product

Resources

About