My chi code looks like this: proc freq data = chi4 noprint; by who svcd svdesc; tables source*admission / chisq; weight count; output out = chifinal; run; My output is this: who svcd svdesc numsubstrata chisquare more columns but not the problem 1111 1 minor 31698 1111 2 moderate 42019 That above is not the correct output. The table chi4 from above has this information in it: source who svcd sbvdesc admission count national 1111 1 minor no 31645 national 1111 1 minor yes 52 prov_tin 1111 1 minor no 1 prov_tin 1111 1 minor yes 0 national 1111 2 moderate no 41864 national 1111 2 moderate yes 154 prov_tin 1111 2 moderate no 1 prov_tin 1111 2 moderate yes 0 So the code I have above is actually combining all the items under the who and the svcd and summing the no and yes. That is not what I want I want the output to be who svcd svcdesc admission count national 1 minor no 31645 1111 1 minor no 1 national 1 minor yes 52 1111 1 minor yes 0 national 2 moderate no 41864 1111 2 moderate no 1 national 2 moderate yes 154 1111 2 moderate yes 0 Then it will calculate the rest of the chi information. It just has to do with something I am not doing with the source portion and the who. If it helps here is my code to combine the who and source and get the national data into the information with the tin data. proc sql; create table chi4 as select * from (select distinct "prov_tin" as source, who, svcd, svcdesc, admission, count from chi) union all (select distinct "national", a.who, a.svcd, a.svcdesc, a.admission, a.count from chi as inner join chi3 as b on a.svcd = b.svcd and a.svcdesc = b.svcdesc and a.admission = b.admission) order by who, svcd, source, admission; run;
... View more
