RE: st: RE: RE: RE: RE: Combining multiple observations by an ID variable

[Clyde B Schechter <clyde.schechter@einstein.yu.edu>]

A consensus appears to have emerged, now including Claude, that -reshape wide- isn't possible for his data set.  He seems to remain daunted at the task of calculating what he needs from his long-form data set.

One more command that may prove useful to Claude is -collapse-.

For example, if the presence of a certain condition is coded 0 = no, 1 = yes, and varies within patient over visits, a simple way to get a one-record-per-patient data set that says whether the patient ever had it is:

collapse [other interesting variables] (max) condition [other interesting variables], by(patient_id)

The resulting data set will have one record per patient, and in this data set condition will be 1 if the patient ever had the condition, and 0 otherwise.  And, if desired, he can rename the variable either in the -collapse- command or afterward. See -help collapse-.