FAQ tweaked for version 19

[Nick Cox]

The Statalist FAQ https://www.statalist.org/forums/help has been tweaked given the release of Stata 19.

As is longstanding convention, posters are asked to flag if they are not using the latest version of Stata, now 19 (or 19.50 if you have StataNow).

[Richard Williams]

Thanks Nick. A suggestion for pt 12.2: People often post sample data using dataex, but the data don’t always include the variables that are needed to replicate the user’s problem. Before submitting data and code, users should make sure that dataex actually includes all the variables needed to replicate their problem.

[Richard Williams]

More succinctly, “Be sure that dataex includes all the variables needed to replicate your problem.”

[Nick Cox]

I don't know. Many, many details in the FAQ could be flagged with an injunction "Make sure you really do this!".

[Hemanshu Kumar]

It’s not just lack of variables, but lack of variability in variables. The admonition should be for them to check they are able to reproduce their issue with their data extract.

[Nick Cox]

You've persuaded me, Richard Williams and Hemanshu Kumar We will tweak the wording again. Thanks for your suggestions.

[Richard Williams]

Thanks. People will still screw it up, but at least we've tried. I suspect this is one of those things where you can read the docs but won't actually appreciate them until you've encountered the problem yourself.

The lack of variability in variables can be tough to deal with. It might be that, because of the way the data are sorted, the first 100 cases are all women. It might be nice to add an option to dataex to draw a random sample of cases. Even then, I bet an xt problem with 500 panels and 20 time periods is going to be hard to replicate with a 100 record extract.

[Nick Cox]

Richard Williams The help for dataex does spell out the scope to use randomtag to get random samples that give the flavour of a dataset, so I don't think any new option is needed: Robert Picard already provided the functionality in 2014!

StataCorp treat dataex unusually in declaring it a community-contributed command, which it is, but one that is always now distributed with Stata's official versions. In contrast randomtag is on SSC only, but easily downloaded.

Problems with very large or very complicated datasets are always going to be challenging.

[Richard Williams]

Incidentally I wish everybody followed this point:

16.1 Close by giving a summary and thanks

Trying to wrap up a thread you started is helpful, especially if you report what solved your problem. You can then thank those who tried to help. Conversely, ignoring answers is less sociable, even if those answers did not solve your problem. "Thanks in advance" does not absolve you from either expectation.

It is very annoying when you take the time to try to help somebody and never hear back from them. Plus anybody who encounters the thread in the future will benefit from knowing what worked and what didn't.