Hi @yabwon Thankyou for your reply. Yes its perfect and that's what i was looking for. Really appreciate. ... View more

So explicitly test for ABC. create table want as select id , group , max( group='ABC' ) as has_ABC , count(distinct group) as count_of_group , count(*) as count_of_account , sum(sum_var) as total from have group by id order by id, group ; quit; Now it looks like you want these two groups:  where has_abc and count_of_group=2 ; where has_abc and count_of_group>2; ... View more

@GAL1986 personally I don't use PROC COMPARE, I would do a custom SQL or data step since PROC COMPARE output is difficult to wrangle. If you really want PROC COMPARE you likely need to take the full data out and restrict the data set in a second data step anyways.   PS. Please post your own question, not as a response to a thread that's 9 months old.   @GAL1986 wrote: Hi,    I would like to create a dataset that shows the values produced by OUTBASE and OUTCOMP but only for people with unequal values. Is there a way to do this? The following code, even with OUTNOEQUAL, shows the values for all observations.    proc compare base=HLA_internal_test comp=HLA_external_test out=dif2 outbase outcomp outnoequal; ID PID; run;   Thank you in advance!    ... View more

Its a good idea to start a new thread for new questions http://documentation.sas.com/?docsetId=proc&docsetTarget=p1l5iwaf47ma83n1euxxp10g8xh5.htm&docsetVersion=9.4&locale=en I don think there is an option for that.  Dates and datetimes are actually numbers, number of days or seconds since a certain point in time, hence they are very different.  You can only compare like for like. ... View more