Date:         Wed, 20 Jan 1999 11:49:41 +0100
Reply-To:     Fabrizio <AvStrien@USA.NET>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         Fabrizio <AvStrien@USA.NET>
Organization: KLM Royal Dutch Airlines
Subject:      Re: Question about PROC FREQ

Major differences between your different PROC FREQs are:
 
1) the edu*age*sex one needs a sort by CLUSTER only,
   the other one requires a sort on CLUSTER, AGE and SEX.
 
2) the edu*age*sex one may fail when the number of possible
   combinations exceeds some 32756
 
3) the edu*age*sex one definitely wins on performance when
   the number of possible combinations remains below those
  32756.
 
Brgds, Fabrizio
 
 
 
christopher wrote in message <36a5859f.1204566697@news>...
|I have tested the difference is the cumulative freq (with different
|basis), the count remain the same.
|On Wed, 20 Jan 1999 07:04:46 GMT, appreciate@bigfoot.com (christopher)
|wrote:
|
|>Hi,
|>Is any different between
|>
|>proc freq;
|>tables edu*age*sex/out=xxx;
|>by cluster;
|>
|>and
|>
|>proc freq;
|>tables edu/out=xxx;
|>by cluster age sex;
|>
|>Thanks
|