A first step to fix the data structure a bit (it is still very fragile, in my opinion) makes things easier:   data have; input ID visit nrs01_1 nrs01_2 nrs01_3 nrs03_1 nrs03_2 nrs04_1 nrs04_2 (q1a q3b q4a) (&:$20.); datalines; 1 0 1 . 1 1 . 1 1 Joe H Samantha B Jodi R 1 1 . 1 1 . 1 1 . Joe H Samantha B Jodi R ; data step1; set have; array nrs_ nrs:; array q_ q:; length code_name $32; nrs_Id = 1; do q_Id = 1 to dim(q_); real_name = q_{q_Id}; do until(scan(code_name,1,"_") ne scan(vname(nrs_{nrs_Id}),1,"_")); code_name = vname(nrs_{nrs_Id}); value = nrs_{nrs_Id}; output; nrs_Id = nrs_Id + 1; if nrs_Id > dim(nrs_) then leave; end; end; keep id visit real_name code_name value; run; proc sort data=step1; by id visit real_name code_name; run; proc transpose data=step1 out=step2(drop=_name_); by id visit real_name; var value; id code_name; run; and so on...   ... View more