You will need to do the summary in two steps to get both types of information. If your data is really large and you do not want to read it twice try creating a view that creates new variables that you can use to count the different missing values.

data x / view=x; set have ; nmiss = x=. ; nnd = x=.n ; run; proc summary data=x ; var x nmiss nnd; output out=want sum(nmiss)=nmiss sum(nnd)=nnd ..... ; run;

On Fri, 24 Feb 2012 07:00:57 -0500, anon <rhian.pilling@GMAIL.COM> wrote:

>Dear all, > >I am still struggling with my previous post. > >I am trying to output summary statistics for a CONTINUOUS variable - >specifically the median, IQR, total observations, complete observations, >those not done (.n) and missing (.) > >However, my issue is is that not done (.n) and missing (.) are continually >being combined and being output as a combined missing (.n + .) whatever >procedure I use.I have used proc means looked and tried proc freq, proc >univariate and proc tabulate. > >Does anyone know how I can differentiate between these missing values in SAS >without categorising the variable (if I categorise I lose the correct median >/IQR values). > >Ultimately I am putting all summary statistics into proc report. > >BW