Indeed, or why not just use a TYPE statement (which I prefer to the NWAY for one-class means, even if it is slightly more typing - more clear what you're doing)? If you're just concerned about not getting the overall row...

proc means data=psy noprint; class student_id; type student_id; var hours gpa; id ACT; output out=need3(drop=_type_ _freq_) mean=hours_mean gpa_mean; run;

Perhaps toby is explaining how to do the other things for a more general purpose however (Teaching how to do those operations without another data step) :)

-Joe

On Thu, Oct 21, 2010 at 9:28 AM, Data _null_; <iebupdte@gmail.com> wrote:

> On Thu, Oct 21, 2010 at 8:52 AM, toby dunn <tobydunn@hotmail.com> wrote: > > The proper way to do the Mean Procedure, I stuck with the Class statement > because one doesn't have to have it sorted like in the previous Data Step > Statement. > > However, that means I had to use the Where data step option on the > resulting data set.: > > > > Proc Means > > Data = Psy NoPrint ; > > Class Student_Id ; > > Var Hours GPA ; > > ID ACT ; > > OutPut Out = Need3 ( Drop = _Type_ _Freq_ > > Where = ( Not Missing( Student_ID ) ) ) > > Mean = Hours_mean GPA_mean ; > > Run ; > > Can you explain this a bit more for us commoners. What is different > about MEANS that a WHERE data set option is needed? Couldn't you use > virtually then same syntax for either MEANS or SUMMARY, e.g. NWAY. >