1. (StataNow™) New priors for Bayesian analysis. bayesmh and bayes have the following new priors:
   1. bayesmh and bayes now support the half-Cauchy prior. halfcauchy(loc,beta) specifies a half-Cauchy distribution with location loc and scale beta. This heavy-tailed prior is useful for modeling nonnegative parameters such as variances and standard deviations that tend to have large values.
   2. bayesmh and bayes now support the Rayleigh prior. rayleigh(beta) specifies a Rayleigh distribution with scale beta. This prior distribution is related to chi-squared and exponential distributions and thus can be used to model nonnegative parameters with skewed distributions. It is often used in physics and engineering to model parameters — for instance, wind speed — that correspond to a norm of a bivariate standard normal random vector.
2. Existing commands cnsreg, eivreg, eregress, hetregress, intreg, rreg, tobit, truncreg, xteregress, and xtreg now support options eform and eform(string) to report exponentiated coefficients. The column title is exp(b) or the specified string.
3. Online help and the search index have been brought up to date for Stata Journal 24(2).
4. The Java Development Kit (OpenJDK) that is redistributed with Stata is now updated to version 17.0.12+7-LTS acquired from Azul Systems.

© Copyright 1996–2026 StataCorp LLC. All rights reserved.

5. dtable, when specified with a string variable in option by(), failed to show results for the levels of the by() variable containing spaces. This has been fixed.

6. estat ic after mixed with option nostderr or emonly used degrees of freedom based on the posted VCE rank. In these cases, the posted rank does not include the random-effect parameters. The number of random-effect parameters is now included in the AIC and BIC degrees of freedom.
7. frames use, when specifying only some of the smaller frames to be loaded into memory, took longer than necessary to load the subset of frames when the frameset on disk contained one or more large frames. In this situation, the loading time is now significantly quicker.

8. predict with option csnell or mgale after stintcox, when specified with multiple-record-per-subject data where some observations were excluded from estimation for containing missing covariate values or being recorded after the event of interest, exited with a Mata conformability error. This has been fixed.