Interval-censored Cox model
Difference in differences (DID)
Panel-data multinomial logit
Zero-inflated ordered logit
Bayesian IRF and FEVD analysis
Bayesian dynamic forecasting
Do-file Editor enhancements
Intel Math Kernel Library (MKL)
Stata on Apple Silicon
Customize your tables of
Stata is fast, and keeps getting faster.
You want to model time to an event.
But you don't know the exact event times—only the intervals in which events happen.
And you don't want to make parametric assumptions.
Try an interval-censored Cox model.
Do you have multiple effect sizes?
Do they share a common control group?
Do they share the same group of subjects?
Multivariate meta-analysis can help.
You fit your VAR models with var.
You fit your Bayesian regression models with bayes:.
Now fit your Bayesian VAR models with bayes: var.
Nonlinear, joint, SEM-like, and more.
More multilevel models.
Easier to use.
When you want:
Causal inference, average treatment effects, potential-outcome means, double-robust estimation
And you have:
Many (maybe hundreds or thousands of) potential covariates
Use treatment-effects estimation with lasso variable selection.
Are there influential studies in your data?
Use leave-one-out meta-analysis to find out.
Graphically summarize meta-analysis results
Detect potential outliers
You can model categorical outcomes with mlogit.
You can model panel data with xt.
Now you can do both!
Stata's new xtmlogit command models categorical outcomes that change over time.
Bayesian analysis lets you answer probabilistic questions with panel-data models.
Incorporate prior knowledge, see posterior distributions of random effects, compute Bayesian predictions, and more.
Need to model an ordinal outcome?
Have excess zeros (or responses in the lowest category)?
ziologit is the answer.
Do responses have an increasing or decreasing trend? Find out using one of four nonparametric tests for trend:
What is the effect of a shock over time?
What is the mean or median of the effect for a distribution of probable scenarios?
Bayesian IRF analysis answers these and more.
After VAR, you want a dynamic forecast.
After Bayesian estimation, you want statistics of posterior distributions.
Estimate both. Visualize both.
Your data have ...
Your data have ...
clusters of observations.
Your lasso for prediction, model selection, or inference can now select variables while accounting for clustering.
Which variables should lasso include?
BIC for lasso penalty selection can tell you.
Forming rational expectations
of the future is hard.
DSGE models include
Prior information helps.
Mata functions and operators use heavily optimized LAPACK routines underpinned by the Intel Math Kernel Library.
Use your favorite Stata commands like always; underlying functions are faster, so you get results faster.
Connecting Stata to databases is now easier.
Want to access data from Oracle, MySQL, Amazon Redshift, Snowflake, Microsoft SQL Server, and others?
Want one driver that works on Windows, Mac, and Linux?