The `R`

package `METests`

implements tests for various hypotheses about measurement error as proposed in Wilhelm (2019). The R package contains help files describing the various commands, their syntax and gives examples.

Install the package `devtools`

if it isn’t already:

install.packages("devtools")

Load the package `devtools`

:

library("devtools")

Install the package `METests`

:

install_github("danielwilhelm/R-ME-test")

Here is an example of how to test the null hypothesis that there is no measurement error in an explanatory variable X, using a second measurement Z and an outcome Y that depends on the true explanatory variable:

rm(list = ls(all = TRUE)) set.seed(1090) library(METests) # generate true explanatory variable Xstar <- runif(100) # generate repeated measurements X and Z X <- Xstar + rnorm(100, 0, 0.3) Z <- Xstar + rnorm(100, 0, 0.3) # generate outcome Y <- Xstar^2+Xstar/2 + rnorm(100, 0, 0.5) # perform the test for measurement error using cross-validated bandwidth DMTest(Y, X, Z)