psacalc with sureg - sharing code

[Jane Fry]

Hi,

I have been working on bounds analysis using psacalc for my regress and xtreg, fe models. I needed to do the analysis on an sureg and adapted the code. I am happy to share the code but wondered about etiquette. Should I just share the code here or should I contact the authors of the psacalc ado file?

Thanks,
Jane.

[Nick Cox]

If authors have made code public and you have used it, you don't need their permission to post what you did.

A public or private signal that you found their work useful, or even invaluable, never goes amiss. Many journals discourage, or don't encourage, citing software if there isn't a paper associated with it.

Sometimes authors ask that their work be cited in a particular way.

Does that answer the question?

[Nick Cox]

Just as a generalized comment: as an occasional programmer myself I note that in many fields there is a big gap between citation of papers and citation of programs used, and the last is very often shockingly if unsurprisingly deficient.

Something that usually goes back to undergraduate years, and even secondary [high] school, is the importance of knowing and citing the literature either in general or in particular the literature likely to be familiar to reviewers -- especially that written by the reviewers themselves if only they can be guessed at.

With software the scene seems more variable. I've seen people cite Stata as a whole, and fine.

Otherwise people seem more likely to cite a command that is tied to a paper to do a particular kind of statistical analysis, and less likely to cite commands for data management or graphics or tabulation.

That is perhaps tied up with different biases.

One is that data management itself is too mundane to be written up at length or even at all, which leads to all sort of puzzlement from those trying to replicate published papers that say little about it. Very many projects here on Statalist depend on a great deal of data management before the interesting analysis is possible at all.

Another is that once a graph or table is created it usually looks fairly obvious and the hours or days of work that produced it, or the small or large trickery needed, are undersold or regarded as not worth documenting in detail.

The pressure to post full code as a sign of reproducible research may push in the opposite direction, except that often only a very few people will study the code to see where credit is due.

[Jane Fry]

Thanks Nick. I will post my code here when I have finalised a few last details. Thanks so much for the sage advice (as always). I agree about the (often huge) amount of work that goes into manipulation, tabulation and graphics! The more we can encourage people to cite where their work comes from the better in my opinion.

[Jane Fry]

Please find attached the do file for implementing the delta calculations for Oster's bounds analysis modified to run with SUR estimation. delta calcs.do