1. Many people (including myself) will NOT download Microsoft files. Please provide your data as part of your post: How to convert datasets to data steps How to post code   2. Please also provide the expected output.   ... View more

Post test data in the form of a datastep: https://communities.sas.com/t5/SAS-Communities-Library/How-to-create-a-data-step-version-of-your-data-AKA-generate/ta-p/258712   Then show what the output should look like.  It is impossible for us to suggest something on things we cannot see.  From your text it looks like you have some variables for Last Observation Carried Forward (LOCF), could I ask why your not utilising CDISC standards on the format of your data, as these would then be identified as parameters?  That would be a very different structure to the one you describe. ... View more

Not sure what your trying to do. As far as I understood here is my tip:   If your scale is in continuous sequence then  data test; input num; datalines; 1 2 3 4 5 ; run; proc sql; select *,(max(num)+1)-num as new_num from test; quit; If not  continuous then,  data test; input num; datalines; 1 2 4 5 ; run; proc sort data=test out=reverse_test; by descending num; run; data reverse; merge test reverse_test(rename=(num=new_num)); run;   ... View more

data have; input Var1 Var2 Var3 Var4; datalines; 1 3 . 1 2 8 2 . 3 . 7 8 ; data temp; set have; array v{*} var:; do n=1 to dim(v); value=v{n};if not missing(value) then output; end; keep n value; run; proc sort data=temp out=want; by n; run; ... View more

Something like this?   data have; input A1 A2 A3 B1 B2 B3 C1 C2 C3; datalines; 1 2 3 4 5 6 7 8 9 2 3 4 5 6 7 8 9 6 5 7 2 0 8 6 5 8 6 4 6 3 9 7 1 6 9 7 4 8 3 5 8 3 1 6 9 ; data want(keep= A B C); set have; array As{*} A:; array Bs{*} B:; array Cs{*} C:; do i=1 to dim(As); A=As[i];B=Bs[i];C=Cs[i]; output; end; run; ... View more

Your recoding is unrelated to Proc Freq as far as I can tell and is pretty easy using arrays. data want; set have; array g group_a1- group_a4 group_b1- group_b3 group_c1- group_c2; array t treat_a1- treat_a4 treat_b1- treat_b3 treat_c1- treat_c2; array f a1- a4 b1- b3 c1- c2; do i= 1 to dim(g); f[i] = (g[i]=1 and t[i]=1); end; drop i; run; array is way to reference variables using an index number. The variables are added to the array definition above using one of the List methods: group_a1-group_a4 assigns all the variables in the range indicated in order for example.   The short hand to use group_a1 is g[1], group_a2 is g[2]. The variable in the "do I=" is used as the index.   The proc freq might be done with something like this: proc freq data= have; tables Group_A1*Treat_A1 Group_A2*Treat_A2 Group_A3*Treat_A3 Group_A4*Treat_A4 ; run; and use the editor to copy the A parts and paste below and edit to have the needed B or C.   However for another, and often better is to not have data in the variable names. The following restructures the data so that each records is from A1, A2.. and then uses the classification variable I call class to use BY group processing to do the many freq requests.   data want; set have; array g group_a1- group_a4 group_b1- group_b3 group_c1- group_c2; array t treat_a1- treat_a4 treat_b1- treat_b3 treat_c1- treat_c2; do i= 1 to dim(g); class= scan(vname(g[i]),2,'_'); group= g[i]; treat= t[i]; val = (g[i]=1 and t[i]=1); end; keep class group treat val; run; proc sort data=want; by class; run; proc freq data=want; by class; tables group*treat; run; This is one way of transforming data. Proc transpose can do similar for other data structures.   ... View more

data sss; input a_1 a_2 a_3 a_b1 a_b2 a_b3; cards; 1 2 3 3 2 1 2 1 3 3 2 1 1 2 3 3 3 3 run; proc format; value likert 1 = 'good' 2 = 'better' 3 = 'best'; run; proc print data=sss; format a_: likert.; run; ... View more

But if you want all tests to be based on the same error variance estimate, you can do   data have; input Pre Post1 Post2; datalines; 12 21 26 15 15 16 18 18 18 25 30 28 25 25 27 ; data test; set have; length test $5; subj + 1; test = "Pre"; value = Pre; output; test = "Post1"; value = Post1; output; test = "Post2"; value = Post2; output; keep subj test value; run; proc glm data=test; class subj test; model value = subj test; lsmeans test / pdiff; run; ... View more

You can do this. data ds1 (keep=ID_No Type); input ID_No $ Type; datalines; A1 1 A1 1 B2 3 C3 4 C3 4 C3 4 D4 4 ; run; proc print data=ds1; run; data ds2 (keep=ID_No Branch); input ID_No $ Branch; datalines; A1 1 B2 2 C3 3 D4 4 ; run; proc print data=ds2; run; proc sort nodupkey data=ds1; by ID_No Type; run; proc sort data=ds2; by ID_No; run; data outdata; merge ds1 ds2; by ID_No; run; proc print data=outdata; run; Your output will look like this   Obs ID_No Type 1 A1 1 2 A1 1 3 B2 3 4 C3 4 5 C3 4 6 C3 4 7 D4 4 Obs ID_No Branch 1 A1 1 2 B2 2 3 C3 3 4 D4 4 Obs ID_No Type Branch 1 A1 1 1 2 B2 3 2 3 C3 4 3 4 D4 4 4 ... View more