Thank you so much everyone for your help!   I adapted the code to account for every single race value in my "race variable" and "missing" so that my 95% CI would account for missing values as well.   proc freq data=data; tables race: / missing binomial(level='1' cl=exact); tables race: / missing binomial(level='2' cl=exact); tables race: / missing binomial(level='3' cl=exact); tables race: / missing binomial(level='4' cl=exact); tables race: / missing binomial(level='5' cl=exact); tables race: / missing binomial(level='6' cl=exact); run; ... View more

Thank you so much to everyone for your help and for the different ways to solve this problem. Thank you again! ... View more

Hello to both of you, Thank you so much for your help. Is there a way to do it without having to transpose? This is a play set, but my actual dataset is over 300,000 with over 21 columns per entry to transpose. Thank you so much! ... View more

Thank you so  much, PG! It was exactly what I wanted. Thank you so much for your help, Louise ... View more

Hi PG, Thank you so much. The last code was wonderful and very helpful. I have one last question. For another part of my code, the whichn function is the best fit. Is the whichn function account for a record with all missing values? Is there a way to allow for missing values so that the code does not come out with an error using the whichn function? For example just changing your first code with an entry  (ID B) to all missing values, I get an error " Array subscript out of range at ID=B Age1=. Age2=. Age3=. Check1=. Check2=. Check3=. bestAge=. _ERROR_=1 _N_=2". Is there anyway to work around all missing values for the function whichn? Otherwise, it does exactly what I want for another part of my code. Thank you so much, Louise Code below with ID B changed to missing values: data have; input ID $ Age Age1 Age2 Age3 Check1 Check2 Check3; datalines; A 65 65 66 . 1 0 . A 66 65 66 . 1 0 . B . . . . . . . B . . . . . . . C 68 68 52 54 0 1 1 C 52 68 52 54 0 1 1 C 54 68 52 54 0 1 1 ; proc means data=have(drop=age) noprint; by ID age: notsorted; output out=checks(drop=_:) sum(check:)=; run; data want; set checks; array age{*} age:; array check{*} check:; bestAge = age{whichn(max(of check{*}), of check{*})}; keep ID bestAge; run; ... View more

Hi PG, Thank you so much for your help. I ran this, and it works. However, is it possible for the "whichn" function to account for any ties? Such as if the summed  value for Check1 and Check 2 are equal-- is there a way to show any ties and the age1 and age2s (if there are ties)? Thank you so much for this, Louise ... View more

Hi everyone! This was my first time posting a question, and thank you to everyone for your help. I have received so many different ways of getting what I wanted. Part of the reason that I wanted all the "DaysToEvents" in columns was because I need to find the difference between all combinations of the difference of DaystoEvents by ID, and I thought that the best way to take differences was to use columns. But Astounding is right, I will have very many columns of new DaystoEvents (233 to be exact!). I will think about how to move forward with the analysis that I want using this data structure, given the large columns. Thank you to everyone! Louise ... View more

Hi Arthur, Thank you so much! Is there a way to make the array number for DaystoEvent a variable as opposed to just 3? This is because in my sample data, I know that the largest number of records per ID is 3. However because I have such a large dataset (300,000 IDs) with varied #'s of records per ID, I'm not sure sure how many variables of DaystoEvent I will need for the array in the actual dataset. Thank you so much, Louise ... View more

Hi Fareeza, Thank you so much--really appreciate it. Unfortunately, I need to keep the second variable (DaysToEvent) because I need to take the difference between DaytoEvent_1- DaystoEvent, DaystoEvent_2-DaystoEvent, etc. Is there a way to keep the DaystoEvent column? Thank you so much, Louise ... View more