iebaltab: suggestion for new options

[MRuzzante]

About command syntax

• [ ] I noticed that iebaltab and ieddtab do not allow to abbreviate the variance estimator option. I would add it, consistently with what is indicated in vce_types options in the official Stata manual.

About results display

• [ ] Display the table in the Stata results window as you do with ieddtab (so the user does not have to open the Excel/Tex exported file or use the browse option to check the results)
• [ ] Use label, instead of variable name, for cluster, fixed-effects, and weights variables, in the table footnote

About table design

• [ ] Add option to show confidence interval instead of standard error below the treatment/control mean
• [ ] Show both t-test and p-value in the difference-in-means column

[kbjarkefur]

@MRuzzante , thanks for this!

Here are my reactions to your suggestions:

Things I agree that we should implement

• I like the vct_type abbreviation, that should be possible to implement. Thanks for the suggestion.
• The Stata window is something we have planned for the iebaltab-rewrite, so that feature is already coming

Things that if we do we only do LaTeX The other output formats (Stata's result window and Excel) are not as dynamic as LaTeX and too much modification is not worth the effort to implement. But LaTeX is better at this, so we might make this an option only when using that output.

• confidence interval instead of standard error
• Show both t-test and p-value

Things I am not sure we will implement

• Value labels instead of name in note. We never meant for the note we write to be the final note that goes in the report, at that time we expect the two options tblnote() tblnonote to be used to create a note that goes well with the language and tone used in the report. The automatically generated note can be a good starting point, but it is mainly a tool to help you remember what you did in that table when writing the report. I am afraid that labels will be confusing if that is the default. Does that makes sense? Please let me know if you think users think differently about this!

@luizaandrade , let us know if you agree with this or not.

[luizaandrade]

Reading this with some delay, but I mostly agree with Kris. Just not sure if it is not more confusing to display for t-test and p-value at the same time. My worry is that the output becomes too polluted without adding much information that wasn't already there. I can still try it as I re-write the command and see what it looks like.