Using Stata to present difference-in-differences results in a visually appealing and intuitive way

[Ronald Biefinger]

Hello,

after running several difference-in-differences (DID) regressions lately, I was wondering if there are any interesting ways to present my findings using Stata. Of course my results will be presented in the standard table format, but I would really like to add a visually appealing and intuitive option (i.e. some sort of graph). After reading several research papers which also included DID models I did not yet find a "nice" solution. I was hoping to find some suggestions here.

The DID estimates are the only results of interest to me with regard to the graphically presentation (in my case an interaction term of two dummy variables). The control variables will be displayed in the table.

Any suggestions of doing that in Stata? It would be great if you could post some links or ideas here. Maybe we can get a collection togethere which can benefit all members.

Thank you very much in advance!

Best

[Stephen Jenkins]

Ben Jann has written a really good program called coefplot which will perhaps help you. For presentations by Ben, see the archives of recent Stata User Group meetings in the UK (and, I think, Germany) -- on the Stata website. There is a version of the program on SSC, and the very latest Stata Journal (14-4) has an article by Ben about it.

[Ronald Biefinger]

Thank you very much.

Any particular representation/graph you would personally choose for displaying (DID) estimates?

[Stephen Jenkins]

If there is one "treatment effect" estimate from the DiD, you can show that? [But a whole graph for one number plus CI seems like a waste of space to me.] If effects differ across subgroups, you can show the various estimates? Of you can show the One Estimate derived from several plausible model specifications on one graph?
All in all, what's the "best" representation is your call as researcher, I think!