rddapp
Regression Discontinuity Design Application
Estimation of both single- and multiple-assignment Regression Discontinuity Designs
(RDDs). Provides both parametric (global) and non-parametric (local) estimation choices for
both sharp and fuzzy designs, along with power analysis and assumption checks.
Introductions to the underlying logic and analysis of RDDs are in
Thistlethwaite, D. L., Campbell, D. T. (1960)
rddapp
Overview
rddapp provides a set of functions for the analysis of the regression-discontinuity design (RDD).
The three main parts are:
- Estimation of effects of interest
- Power analysis
- Assumption checks
Estimation
The package estimates treatment effects from RDDs for the following designs and approaches:
- Parametric RDD with single assignment variables (both sharp and fuzzy designs)
- Non-parametric RDD with single assignment variables (both sharp and fuzzy designs)
- Parametric RDDs with two assignment variables (both sharp and fuzzy designs), using univariate, centering, and frontier approaches
Power analysis
Given input from the user about desired Type I error rate and assumptions about the population, the package allows estimation of power for the following designs:
- Single assignment RDDs (both sharp and fuzzy) using both parametric and non-parametric estimation
- Multiple-assignment RDDs (both sharp and fuzzy) using various parametric models
Assumption checks
The package allows the user to perform a variety of assumption and sensitivity checks, including:
- McCrary's sorting test on the assignment variable
- Sensitivity to the chosen bandwidth in non-parametric estimation
- Placebo tests to examine treatment effects at values away from the cut-off
- Discontinuities in the treatment probability at cut-off
- Discontinuities for baseline covariates
Installation
# Install the released version from CRAN
install.packages("rddapp")
# Or the development version from GitHub:
# install.packages("devtools")
devtools::install_github("felixthoemmes/rddapp")