A program to automatically quasi-demean regressors following a FGLS-RE or MLE-RE regression. Available in both R and Stata
qdmean is a program to automatically quasi-demean regressors following the estimation of a random effects (either using FGLS or maximum likelihood) model. This program requires you to first estimate a model using xtreg, with options , re or , mle. The program will automatically obtain theta_i and generate quasi-demeaned regressors, which are useful for post-estimation analysis. See the Stata help file for more details.
To install qdmean in Stata direct from Github, type the following:
cap ado uninstall qdmean
net install qdmean, from(https://github.com/andyphilips/qdmean/raw/main/)
qdmean() in R requires the estimation of a random effects model using either plm or lmer. Once estimated, pass the model, predictor variable (in quotes), and grouping variable (in quotes). If using lmer, additionally pass the dataset used to estimate the model and the dependent variable (in quotes). For help and examples, reference ?qdmean
To install qdmean in R direct from Github, use the devtools package:
library(devtools)
install_github(''andyphilips/qdmean'')
library(qdmean)
Soren Jordan, Department of Political Science, Auburn University
Andrew Q. Philips, Department of Political Science, University of Colorado Boulder
If you use qdmean, please cite:
Jordan, Soren and Andrew Q. Philips. 2023. "Improving the interpretation of random effects regression results." Political Studies Review: 21(1): 210-220.