estimatr is part of the DeclareDesign suite and provides fast linear and instrumental variables estimators with robust standard errors designed for experimental and observational research. It covers OLS with lm_robust(), two-stage least squares with iv_robust(), fixed-effects regression, and two-group estimators, with ggplot2 and texreg integrations.

What’s covered#

  • OLS with lm_robust() – HC2 default, SE type options
  • Clustered standard errors – CR2 default
  • Fixed effects – dummies approach and fixed_effects absorbing approach
  • 2SLS with iv_robust() – clustered IV regression
  • Two-group estimators – difference_in_means(), horvitz_thompson()
  • Post-estimation commands – tidy(), vcov(), confint(), predict()
  • ggplot2 integration – stat_smooth(method = "lm_robust")
  • Multiple models – same outcome different subsets, different outcomes same subset
  • Extras – Lin covariate adjustment (lm_lin()), texreg tables
  • estimatr-to-Stata dictionary