If this message is double sent I apologize, but i had yet to receive the email I sent and was worried it didn't get out. I am working with several sets of code. Each produces statistics I then place into a perm dataset later in the code. In one of the runs in one condition I am working with the following data set: Obs ER COUNT 1 .26-.30 2 2 .21-.25 32 3 .16-.20 31 4 .11-.15 30 5 .06-.10 38 6 .01-.05 17

I am using the following code: data pcntna; set errfreq; pcntna=0; if ER ne 0 then delete; if ER=0 then pcntna=count/∋ drop count ER; run;

In the other conditions this worked because I always had some items that were not administered (ER=0). In this instance I have no er=0 so I do not end up with the correct statistics of percent not administered. Can anyone offer suggestions to incorporate into this so it can capture and deal with data sets that have no non-administered items and therefore no ER=0 condition? As is i just work with that condition and that has been my mistake. More to the point my problem is that I could just get rid of all observations and just set pcntna=0. Is that the best way to do this? Thanks

Edgar Sanchez Suppression@mail.utexas.edu