nesslasas.blogg.se

STATA 15 SERIES

First, you might have an endogenous variable because lots of models have omitted variables that are correlated with the variables in the model. These new ERM commands- eregress, eintreg, eprobit, and eoprobit-are destined to become popular because they address so many of the problems researchers have. They can even be interacted with themselves to form squared or cubic terms! And they can be interacted with other covariates, whether exogenous or endogenous. And endogenous covariates are not limited to being continuous. These new commands are just short of amazing because you can put endogenous covariates in any of the equations, and that includes the treatment-assignment and probit-selection equations. endogenous (Heckman-style) sample selection.

We call them ERMs-extended regression models. Add your own power and sample-size methods.Panel-data interval regression with random coefficients.Panel-data nonlinear models with random coefficients.Poisson models with Heckman-style sample selection.

STATA 15 SERIES

Tests for multiple breaks in time series.

Multilevel tobit regression for censored outcomes.

Multilevel regression for interval-measured outcomes.

Panel-data tobit with random coefficients.

The sixteen features listed above certainly important ones, but there are others worthy of mention.

Federal Reserve Economic Data (FRED) support.

Power analysis for cluster randomized designs and regression models.

Interval-censored parametric survival-time models.

Linearized dynamic stochastic general equilibrium (DSGE) models.

I go into depth below on the sixteen highlighted features. Here on the blog I will say more about them, and you can learn even more by visiting our website and seeing the Stata 15 features page. I posted to Statalist this morning and listed sixteen of the most important new features. It’s a big deal because this is Stata’s biggest release ever.