Here is one way- assuming you don't know how many sources that you have, first split the data set, then transpose the sources, then merge them back together again. You probably also want to check first that your other variables are identical on the different rows with the same ID, I'm creating a data set "duplicates" that checks to see if the select distinct on the other variables produces only 1 row for each id.

-Mary

data set1;

infile cards;

input id source $ var1 var2 var3;

cards;

1 V 1 2 3

1 S 1 2 3

2 V 10 11 12

3 V 20 21 22

3 S 20 21 22

3 R 20 21 22

;

proc sql;

create table sources as

select id, source

from set1;

create table vars as

select distinct id, var1, var2, var3

from set1;

run;

quit;

proc sql;

create table duplicates as

select sample_name

from vars

where sample_name in(

select sample_name

from vars

group by sample_name

having count(sample_name) >=2)

order by sample_name;

quit;

run;

proc transpose data=sources out=sources_transposed;

var source;

by id;

run;

data final;

merge sources_transposed vars;

by id;

drop _name_;

run;

----- Original Message ----- From: Erica To: SAS-L@LISTSERV.UGA.EDU Sent: Wednesday, April 30, 2008 12:12 PM Subject: Turn obs into vars

Hello all,

I have a SAS dataset that looks like this:

ID Source ...10 Other Vars... 1 V 1 S 2 V 3 V 3 S 3 R

I want:

ID Source1 Source2 Source3 ...10 Other Vars... 1 V S 2 V 3 V S R

Any help would be greatly appreciated!