Dropping individual cases

[ACarroll]

I've looked in the manuals and can't find any information on how to drop individual cases. There is a small number of non-consecutive, non-missing cases that I need to drop. For example, I need to drop observation number 689, 759, 2131. I could just go in the data editor and delete the handful of cases individually, but I need to know how the command for this because I have to send my do-file to a colleague. Any suggestions?

Thanks in advance.

[Nick Cox]

drop in ... is documented. You could go

 
drop if inlist(_n, 689, 759, 2131)

But it is much better style to spell out a reason for dropping using if. Otherwise do files are difficult to understand.

Note that e.g.

 
drop in 689

means that observations numbers are shifted so that 690 on are now 689 on, and so forth.

[ACarroll]

Okay, thanks for the help. I guess I was looking for the wrong thing. I'll give this a shot.

[Nick Cox]

The clear ideal here is, in a standard phrase, reproducible research and, in a not so standard phrase, clear and rational reproducible research, so that the various steps are not only documented, but comprehensibly so. If you don't know how to code up a reason for dropping individual observations, you could always include in the do-file

  
list if inlist(_n, 689, 759, 2131)
* these observations should be dropped because ...

Here the asterisk introduces a comment.