Stata 18 offers several new features for the analyses of linear models. With the new hdidregress and xthdidregress commands, you can estimate heterogeneous average treatment effects on the treated for cross-sectional and panel data. The existing commands regress, areg, and xtreg, fe now provide more precise standard errors and confidence intervals. The new wildbootstrap command uses a wild cluster bootstrap to provide valid inference with a small number of clusters or an uneven number of observations per cluster. In StataNow™, you can specify multiple high-dimensional categorical variables in the absorb() option with commands areg and xtreg, fe to control for these variables in your linear and fixed-effects linear models. You will experience remarkable speed gains over the traditional approach that includes these variables as indicators in your models. In StataNow, you can also fit Bayesian quantile regression with the new bayes: qreg command. You can fit linear models that include continuous, binary, fractional, or count endogenous variables by control-function regression using the new cfregress command. And you can fit correlated random-effects models by using xtreg, cre and perform a Mundlak specification test by using the new postestimation command estat mundlak.
Read more about these features and see them in action:
• Heterogeneous difference in differences (DID)
• High-dimensional fixed effects (HDFE)
StataNow• Robust inference for linear models
• Wild cluster bootstrap for linear regression
• Correlated random-effects (CRE) model StataNow
• Control-function linear regression
StataNow• Bayesian quantile regression
StataNow• Mundlak specification test
StataNow• And more
View all the new features in Stata 18 and, in particular, New in instrumental-variables analysis.
