@Patrick ,
With normalized data, the structure of the flags might change a little. But it should be simple, something along these lines:
proc sort data=normalized;
by id organism visit;
run;
data check_these;
set normalized;
by id organism visit;
if last.organism and resolution="No";
run;
So in this case, I didn't even bother to create flags. I output only the cases that remained unresolved.
There's a more complex situation that I didn't bother to investigate. It might be possible that an organism has "No" at visit 1, doesn't appear at visit 2, and has "Yes" at visit 3. That would take a little more programming ... think DIF(VISIT) ... but shouldn't be that difficult to identify.
... View more
