|
At 09:38 27/05/99 -0400, Douglas Dame wrote:
>(2) I'm not sure how the included code shows that procs MEANS and SUMMARY
>differ in their need for an explicit VAR statement ... but you did get 25
>obs out of MEANS but only 5 out of SUMMARY ... which does illustrate one
>sometimes significant difference between the two procs, namely their
>defaults on how they handle missing values.
>> 9 proc summary data=sasuser.houses;
>> 10 class style;
>> 11 output out=b;
>> 12 run;
No, Douglas, that not the reason - if you run Peter's code, and look at the
output, you'll see what's happening. Even if there were a million
observations (still with just 5 levels of his variable 'style', with no
missing data, that SUMMARY step would still only output 5 observations - one
for each level of 'style' - and with no information about any other variables.
Regards,
John
----------------------------------------------------------------
Dr John Whittington, Voice: +44 (0) 1296 730225
Mediscience Services Fax: +44 (0) 1296 738893
Twyford Manor, Twyford, E-mail: medisci@powernet.com
Buckingham MK18 4EL, UK mediscience@compuserve.com
----------------------------------------------------------------
|
