Let me state my question clearly. I currently have a dataset which has so many variables. (since this is from a large survey) Thus, when I do data analysis A, I need relevant questions as my variables. So I use -keep var- command. But when I do data analysis B, I probably need other variables.

In this case, as a current student, I want to inquire you what your practice is. I remember there is codes that I can save and load dataset at any time.

But just use different do files?