Counting number of variables with certain name

[Lindsey Buck]

Hi all,

I have data with variables called "hosp_date1 hosp_date2 hospdate_3..." and so on. I would like to count how many hosp_date variables are not missing by a person's id. Can you help with the code for this?
I basically want a variable that will=3 if you have a nonmissing hosp_date1, hosp_date2, hosp_date3, and will=2 if one is missing, and so on. Thanks!

[Rich Goldstein]

my guess (please supply dataex data examples in CODE blocks - see the FAQ) is that you want the -egen- command with the "rownonmiss" function; see

help egen

[Lindsey Buck]

Thank you! This gets me close except one problem. There are variables between my hosp dates and it seems to count those too. Is there a way to stop this? I have 437 hospital variables so I don't think I can just order them all

[Nick Cox]

You are not showing any code to comment on but if you have three such hosp_date1, hosp_date2, hosp_date3 then any of the following should work

egen wanted1 = rownonmiss(hosp_date1 hosp_date2 hosp_date3)

egen wanted2 = rownonmiss(hosp_date?)

egen wanted3 = rownonmiss(hosp_date*)

egen wanted4 = !missing(hosp_date1) + !missing(hosp_date2) + !missing(hosp_date3)

But if your variables are in some order like

hosp_date1 lindsey hosp_date2 buck hosp_date3

then indeed the syntax hosp_date1-hosp_date3 will catch lindsey and buck too.

See

help varlist

for the basics here.

[Nick Cox]

Sorry,

 
 egen wanted4 = !missing(hosp_date1) + !missing(hosp_date2) + !missing(hosp_date3)

should be

 
 gen wanted4 = !missing(hosp_date1) + !missing(hosp_date2) + !missing(hosp_date3)