What commands would be needed to create this table

[Luis Mijares]

I want to create a mean randomization check table for a RCT. Basically I want to generate a table like the one shown below but, I don't know what commands to run to describe the difference in the mean of my categorical variables.

[Leonardo Guizzetti]

I can't give you specific commands to help you, I'm afraid. For a table like that I would be more apt to create the table in Excel using putexcel.

I would question the motive of having the difference column with testing at all. Randomization ensures, on the long-run average, that factors known and unknown are balanced. So one expects similar but never exactly equal characteristics between groups, except by chance. One would also expect any test of imbalance to fail, except for a number of false positives at the nominal significance level of the test. In other words, it doesn't serve a useful purpose that can't be served by looking at each group individually.

[Luis Mijares]

Well I guess what I really wanted was to generate a table with the means for the categorical variables before treatment, after treatment, and the difference between those means. I try using ttest but I can't seem to generate the table

[Attaullah Shah]

You may be interested in asdoc (available on SSC). One example of creating a table like given in #1 is available here
https://fintechprofessor.com/2019/09...f-asdoc-stata/

[Adrien Bouguen]

I have been working on a command to generate tables for RCT, rctable. It is available on the SSC. ssc install rctable.

[Brad Anderson]

You may be interested in asdoc (available on SSC). One example of creating a table like given in #1 is available here
https://fintechprofessor.com/2019/09...f-asdoc-stata/

ASDOC (I got ASDOCX) is really helpful for table generation. If the OP has Stata 17 the collect, table, etable, and related commands are very powerful for generating tables. There are some tutorials but there is a distinct learning curve. In medical research the CONSORT guidelines explicitly recommend against statistically testing baseline differences between treatment groups in RCTs. If the randomization has been done correctly and if the work in the field is faithful to the protocol, then differences are attributable to chance. And typically there are numerous comparisons in these kinds of tables so the likelihood of finding a "significant" difference on some comparison variable is considerable. Doesn't stop reviewers from requesting those tests. And conventions in other fields might be different.