Hi, You are getting the error because of this (this is what the compiler sees after macro decoded): if index(upcase(GENERIC),('drug_a' 'drug_b')) >0 ; As you can see from the above this is not how the index function works so you get the errors. You could update the macro to loop over the macro using %scan to get each one out.  Its a bit of a faff though.  Might I ask what you are trying to achieve as I would assume there are better solutions than using a macro variable - for instance you are going to hit the maximum length quickly. One suggestion is: data drugs;      drug="Drug A"; output;      drug="Drug B"; output;      drug="Drug C"; output; run; data _null_;      set drugs;      call execute('data '||strip(drug)||';                                 set have;                                 if index(upcase(GENERIC),'||strip(drug)||') >0 ;                            run;'); run; This will generate a datastep for each row in drugs with the condition to output if generic in have has the word from the dataset drug. Other alternatives can be had depending on scenario. Just to add on the above data _null_ method, the drugs dataset can be separate from the data, updated as and when, and also be any size without changing the main program.  For instance if you hardcode your drug list, then you need to update the program if any new drugs are added etc. With this you could for instance use ATC Coding datasets. ... View more

Ah, it is Oracle Clinical.  I am leaving for the day now, but can pick up further question tomorrow. ... View more

So you could try: proc sort data=have out=tmp;      by set; run; data have;      set have;      if _n_ <= 300 then output; run; This will give the first 300 observations e.g all the 1's, then all the 2's, ... ... View more

hi, Do some sort of directory listing: filename dirlist pipe "dir 'somewhere' /b"; data dirlist;      length buffer $2000;      infile dirlist;      input buffer $; run; Next use that dataset to generate your code: /* open package here */ data _null_;      set dirlist;      call execute('ods package(ProdOutput) add file="'||strip(buffer)||'";'); run; /* close package here */ Note I am assuming of course that all files in the specified directory are to be included. ... View more

Hi, Just want to chime in here and say that writing to a temporary file is not necessary.  You can utilize call execute to do exactly the same thing, using your rules dataset as the loop, and a proc sql insert statement to insert fallouts into a master table.  (Pinching your test data Tom, hope you don't mind): data rules ; infile cards dsd dlm='|';   length exceptioncode $10 exceptiongrade $1 exceptionage $20 ;   length client $20 rule $1000 ;   input exceptioncode -- rule ; cards; 3535asdf|B|12 Months|Brandon|balance > 100 and loantype="purchase" and age<50 3535asdf|B|12 Months|Enron|balance > 580 and loantype="purchase" and age<50 and producttype='test type' run; proc sql;   create table EXCEPTIONS   (     EXCEPTION_CODE char(200),     NAME char(200)   ); quit; data _null_;   set rules;   call execute('proc sql;                   insert into EXCEPTIONS                   select  "'||strip(exceptioncode)||'",                           NAME                   from    YOUR_DB_DATA                   where   '||strip(exceptionrule)';                 quit;'); run; Its pretty much how OC does things, except there it is a cursor, here its a dataset.  So the rules are passed into the dataset.  The call execute generates the sql code including the rule, that is run after the datastep completes. ... View more

Hi, So something like: proc sql;   create table WANT (EMAIL_TYPE char(200),CONTROL num,TEST num,P-VALUE num); quit; proc sql;      create table LOOP as      select     distinct EMAIL      from       HAVE; quit; data _null_;      set LOOP;     call execute('proc ttest...; where email='||strip(EMAIL)||';...;');     call execute('proc sql;                     insert into WANT                     select  "'||strip(EMAIL)||'",                             RESULT                     from    OUTPUT_FROM_TTEST;                   quit;');   end; run; ... View more

Agree with Reeza.  Just to add, if its giving a blank then just add a datastep before your print: data new;      set old;      attrib new_var format=$200.;      if old_var=. then new_var='';      else new_var=put(old_var,best.); run; So you end up with a char field to print. ... View more

You should be able to do: options missing=''; ... View more

I think you may want to look at: options missing=; SAS(R) 9.2 Language Reference: Dictionary, Fourth Edition ... View more

Could you not just wrap your code up in a call execute block, e.g (very pseudo): proc sql;   create table WANT (EMAIL_TYPE char(200),CONTROL num,TEST num,P-VALUE num); quit; data _null_;   do i="Spotlight Email","Movers","Event"...;     call execute('proc ttest...; where email='||strip(i)||';...;');     call execute('proc sql;                     insert into WANT                     select  "'||strip(i)||'",                             RESULT                     from    OUTPUT_FROM_TTEST;                   quit;');   end; run; Note that if you don't have a fixed list (from the do part) then you can sql distinct list and then set that dataset in the data null. ... View more

Hi, Well, you wouldn't need A as it doesn't add anything.  Try: proc sql;      create table C as      select     distinct CITY,                     count(ID) as N      from          B      group by CITY; quit; ... View more