In kota7/rddsigma: Sharp Regression Discontinuity Design with Running Variable Measured with Errors

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "README-"
)

rddsigma

Sharp Regression Discontinuity Design with Running Variable Measured with Error

This package provides estimators for the standard deviation of measurement errors in running variables for sharp regression discontinuity designs. The estimation requires the data for observed (mismeasured) running variables and assigment variables.

Installation

Install from GitHub using devtools::install_github function.

devtools::install_github("kota7/rddsigma")

Quick Start

Use estimate_sigma function.

library(rddsigma)
set.seed(875)

# generate simulation data
dat <- gen_data(500, 0.3, 0)

estimate_sigma(dat$d, dat$w, cutoff=0)

Estimation Methods

The package support two estimators for estimating sigma. The first estimator assumes that both the true runnning variable and the measurement error follow the Gaussian distribution. This is the default estimator of estimate_sigma function, or specified explicitly by setting method="tsgauss"

estimate_sigma(dat$d, dat$w, cutoff=0, method="tsgauss")

The second estimator relaxes the Gaussian assumption and allows the true runnning variable and the measurement error to follow some parametric distribution. Currently, the package supports the Gaussian distribution for the running variable and the Gaussian and Laplace distributions for the measurement error. This estimator is called with method="emparam"

estimate_sigma(dat$d, dat$w, cutoff=0, method="emparam", x_dist="gauss", u_dist="gauss")
estimate_sigma(dat$d, dat$w, cutoff=0, method="emparam", x_dist="gauss", u_dist="lap")

kota7/rddsigma documentation built on May 20, 2019, 1:11 p.m.