I read through the thread, and most of them don't seem to be pulling data out of relational databases; a few that were indicated that the were doing the SQL work in the language of the database system (i.e. Transact SQL for SQL Server, PL/SQL for Oracle). It does seem like most of them are dealing with smaller tables, not big relational database systems. I think if they were dealing with big relational database systems as their source of data then their answer would be different- R seems very much like IML, and if I had to program only in SAS IML only then I'd push my data manipulation back to the database system like Oracle, but since I have base SAS I can do much of what I might have done in PL/SQL in SAS.

So I think they are dealing with different kinds and sources of data; SAS users tend to be dealing with large relational databases that lend themselves to using Base SAS and SAS SQL, whereas R users seem to be dealing with data that lends itself towards matrixes. Sort of explains why there aren't many SAS IML users- those were the matrix oriented types.

You could challenge them to a few of the problems on SAS-L; such as "last value carried forward", to see whether they could really do those sorts of things in R. They currently probably don't see problems like that on R help, because those people trying to do those things are trying to do them in SAS (or a database language like Oracle's PL SQL).

But it probably is easier to do matrix manipulation in R than it is in Base SAS, though the more correct comparison might be R to SAS IML.

-Mary

----- Original Message ----- From: "Peter Flom" <peterflomconsulting@MINDSPRING.COM> To: <SAS-L@LISTSERV.UGA.EDU> Sent: Wednesday, May 06, 2009 4:45 PM Subject: Another SAS v. R thread

> Over on R-help, there is a big discussion on how great R is for data > manipulation. > > Most of it is just "R does it all, nothing else is needed" some of it, > though, is > "Why would anyone use SAS?" > > I don't know enough about data manipulation in either program to say why > SAS is better. > > Peter > > > Peter L. Flom, PhD > Statistical Consultant > www DOT peterflomconsulting DOT com