Hello, I would do this way:   /* Copy your test dataset */ data test; infile datalines delimiter="|"; length v1 $10 v2 $20 v3 $50; input v1 $ v2 $ v3 $; datalines; survey_1|age|Under 21 survey_1|age|21 - 25 survey_1|age|26 - 30 survey_1|age|31 - 40 survey_1|age|41 - 50 survey_1|age|50+ survey_2|Education|Some High School survey_2|Education|High School Graduate survey_2|Education|Some College ; run; /* Adding colomn v4 */ %let letters = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"; data test1; set test; retain v4_tmp; by v1; if first.v1 then v4_tmp = 0; v4_tmp = v4_tmp + 1; v4 = substr(&letters,v4_tmp,1); drop v4_tmp; run; I hope this helps! ... View more

Hello,   You can try this code:   /* test data */ data a;     infile datalines delimiter="|";     length var1 $10 var2 $10 var3 $10;     input var1 $ var2 $ var3 $;     datalines;     ab1|text1|text2     a2|text3|text4     ab3|text5|text6     ab4|text5|text8     ; run; data b;     infile datalines delimiter="|";     length var1 $10 var2 $10 var4 $10;     input var1 $ var2 $ var4 $;     datalines;     ab1|text1|text11     b2|text9|text10     ab3|text5|text12     ab4|text5|text13     ; run; /* we sort datasets */ proc sort data=a;     by var1 var2; run; proc sort data=b;     by var1 var2; run; /* then we merge them */ data aorb;     merge a b;     by var1 var2; run; data aandb;     merge a (in=a) b (in=b);     by var1 var2;     if a and b; run; data aonly;     merge a (in=a) b (in=b);     by var1 var2;     if a and not b; run; data bonly;     merge a (in=a) b (in=b);     by var1 var2;     if b and not a; run;   ... View more