No but there is in R but you need more than one command I am attaching a paper with a protocol that may be helpful. I can supply the program if you like that idea. DO NOT USE STEPWISE METHODS THEY ARE UNSTABLE. See these references:: https://b-ok.cc/book/5825760/ed313f. Paper with protocol is attached. Read:https://www4.stat.ncsu.edu/~boos/var.select/lasso.adaptive.html This shows how to run it and where to find the program. Read the references first please. Contact me if you have questions or need software. Best of luck, David Booth.

#David Eugene Booth thanks for your immediate response and sharing your experience.

In a single package sometimes, you can’t find a command to analysis your data in the way you want. Hence, you are obligated to use different packages concurrently.

For instance; You can’t find AIC in SPSS for logistic regression and you can’t find variable selection approaches in STATA for building logistic regression model. if I need to have AIC and automated variable selection while I’m fitting Logistic regression, I can’t use a single package.

So, is it recommendable to use different statistical software to analysis a single data in a single research?

Of course, otherwise the research doesn't get done. One of the reasons that I prefer R is that that rarely happens there because new methods tend to be coded in R. Best wishes and Good luck in your work, David Booth

My two bits.

1. It is preferred, in most situations, to have you and your knowledge about the variables/theory, guide variable selection rather than letting any algorithm blind to the theory solely determine which variables to include.

2. lasso, as usually used does variable selection since some values drop to zero, but it also does shrinkage/regularization. It might be conceptualized better as a method to shrink the betas, and as a nice consequence of minimizing the absolute values of the sum of these some go to zero.

Lami Abebe Gebrewold, this may be of interest, assuming your version of Stata is recent enough.

• https://www.stata.com/new-in-stata/lasso-inferential-methods/

Lami Abebe Gebrewold wrote: "You can’t find AIC in SPSS for logistic regression".

Perhaps not when using the LOGISTIC REGRESSION command. However, you can estimate the same model via NOMREG or GENLIN, both of which have options to display information criteria including AIC.

• https://www.ibm.com/support/pages/can-spss-produce-aic-or-bic-logistic-regression-models
• https://www.ibm.com/support/knowledgecenter/en/SSLVMB_26.0.0/statistics_reference_project_ddita/spss/advanced/syn_genlin_print.html

HTH.

I believe you absolutely can get AIC in Stata as well as BIC. You can type In estat ic after running a logistic model. I also recommed you download the program spost13 by Long and Freese. This includes a command called ‘fitstat’ that will print out a ton for fot statistics including AIC and BIC. You can also use it to compare model fit. if you save the models as you go along, also using fitstat.

You can run stepwise in Stata as well as nested regressions. Here is a reference for the stepwise command (https://www.stata.com/manuals/rstepwise.pdf)... However, I support what another reviewer stated that it is better to use conceptual rather than brute force for variable selection. I just reviewed a paper that wqs rejected for using stepwise. It’s not viewed favorably these days.

Regarding the original question, I would like to point out that Stata stepwise prefix command does in fact include all of the following options:

--------------------------------------------------------------------------------------------

pr(#) ........................ backward selection

pr(#) hierarchical ... backward hierarchical selection

pr(#) pe(#) ............. backward stepwise

pe(#) ........................ forward selection

pe(#) hierarchical .... forward hierarchical selection

pr(#) pe(#) forward .. forward stepwise

--------------------------------------------------------------------------------------------

Source: https://www.stata.com/manuals/rstepwise.pdf

However, I must also point out this Stata FAQ that lists many of the problems associated with such algorithmic selection methods:

• https://www.stata.com/support/faqs/statistics/stepwise-regression-problems/

And for completeness, here is the link I posted previously showing info about Stata's lasso command:

• https://www.stata.com/new-in-stata/lasso-inferential-methods/

I hope this clarifies what Stata can and cannot do.

Thanks all for sharing your experience.