More new features are available in StataNow. Update your copy of StataNow to try out the latest additions.

Difference in differences with switching treatments for panel data. With the new xtswitchdid command, you can estimate event-study treatment effects for panel data using the method of de Chaisemartin and D’Haultfœuille (2026). This heterogeneous difference-in-differences estimator is useful when subjects switch in and out of treatment or to different levels of treatment. The treatment can be binary or multivalued. You can also test for parallel trends and no anticipation of the treatment, plot the event-study treatment effects, estimate effects for specific paths, and estimate the average total effect.

Spatial error covariance structures for multilevel models. The mixed command now allows you to model the correlation between errors as a function of geographic or physical distance. You can choose from spatial exponential, spatial Gaussian, spatial spherical, and spatial power covariance structures. These structures are useful for data collected at specific spatial locations, such as agricultural field trials and ecological surveys. They can also be used for irregularly spaced longitudinal data, sometimes referred to as intensive longitudinal data, by treating time as a one-dimensional spatial domain. You can incorporate a nugget effect to account for microscale variation and measurement error.

Frames in the Data Editor. When you have multiple datasets in memory, the Data Editor now allows you to view and edit data in each frame. In addition, you can copy and paste from one frame to another, save a snapshot in one frame and restore it in another, manage existing frames, and create new frames.

New priors for Bayesian analysis. The bayes and bayesmh commands now support six new priors.

Two new half-distribution priors, the half-normal and location-scale half-t distribution, are useful for positive-valued parameters such as the scale parameters of the Laplace and Cauchy distributions or the standard deviation of the normal distribution.

Four new square-root parameter prior distributions are also available. The uniform, half-normal, location-scale half-t, and half-Cauchy distributions for the square root of a parameter are useful for positive-valued model parameters such as variances.

Append PDFs. The putpdf command has a new append suboption that allows you to append multiple PDFs into one file.

We’ve also recently added new features that are available to everyone with Stata 19 and StataNow 19.

Customizable tables improvements. The collect suite of commands now allows you to create and customize tables in new ways.

The new collect fromframe command allows you to collect results from a data frame rather than from the stored results after running a Stata command. You can then create and customize tables using the collected data.

The collect style cell command has a new missing() option that allows you to control how missing values are displayed. You can specify a custom label, or you can hide the missing values.

The collect style row split command has the new titles() and extraspace() options. With titles(), you can specify where dimension titles and factor-variable labels should be placed in the row header. With extraspace(), you can specify extra space be placed between row header columns.

collect now supports a new dimension, header_cell_contents, that allows you to apply style edits to dimension titles, factor-variable titles, and levels directly. For example, you can underline dimension titles or make factor-variable titles bold.

Effect size improvement. The esize twosample command now supports the reverse option to reverse the order of the groups when computing differences in means. When specified with the pbcorr option, this provides the same estimate of point-biserial correlation that you obtain by using correlate.

For more features of StataNow 19 and Stata 19, see https://www.stata.com/new-in-stata/features/.

You can type update all in your copy of StataNow or Stata 19 to get access to these version-specific updates.

Reference

de Chaisemartin, C., and X. D’Haultfœuille. 2026. Difference-in-differences estimators of intertemporal treatment effects. Review of Economics and Statistics. 108: 863-880. https://doi.org/10.1162/rest_a_01414.