given your data, i think the code would do what you want. I changed option from missover to truncover.

data have; infile datalines truncover; input id name1 $char8. name2 $char8. name3 $char8.; flag=(name1 max name2 max name3)=(name1 min name2 min name3); datalines; 1 mike mike mike 2 joe joe joey 3 andy andy 5 danny dan 6 sean shawn 7 aaron ; run; proc print;run;

On Mon, Mar 14, 2011 at 4:09 PM, Wei Wang <weiwangum@yahoo.com> wrote:

> Thanks Joe. In this case, as long as there are any different > letters between two names they are treated as different names. For > instance, Danny and Dan are two different names. > > Wei > > --- On Mon, 3/14/11, Joe Matise <snoopy369@GMAIL.COM> wrote: > > > From: Joe Matise <snoopy369@GMAIL.COM> > Subject: Re: Find Different Names > To: SAS-L@LISTSERV.UGA.EDU > Date: Monday, March 14, 2011, 3:53 PM > > > Wei, > This is a fairly trivial task if you're just comparint identical (Joe, > Joe), > and even if it's just shortenings (Joe, Joey; Dan, Danny). However, > Sean-Shawn begins to add a significant layer of complexity to it - how > would > you tell a computer algorithmically those are identical? Same goes for > "Marge/Margaret", "John/Jack", "Chris/Krystal", etc., particularly when you > have some that are sometimes valid and sometimes invalid (Chris/Krystal for > example). Some of the luminaries on SAS-L have addressed this sort of > issue > in the past, and hopefully they can assist you (or you can google the > listserv's history for 'fuzzy matching'). In general, I would say that the > easiest way to approach this is one of the SOUNDEX type functions, but any > of those will have some risk of false positives and false negatives - the > only true way to do this is to make the list by hand. > > -Joe > > On Mon, Mar 14, 2011 at 3:39 PM, Wei Wang <weiwangum@yahoo.com> wrote: > > > Hi guys, > > > > data have; > > infile datalines missover; > > input id name1 $char8. name2 $char8. name3 $char8.; > > datalines; > > 1 mike mike mike > > 2 joe joe joey > > 3 andy andy > > 4 > > 5 danny dan > > 6 sean shawn > > 7 aaron > > ; > > run; > > > > I want to create a flag varialbe indicating different non-missing > > names. Here is the data I need. > > > > id name1 name2 name3 flag > > 1 mike mike mike 0 > > 2 joe joe joey 1 > > 3 andy andy 0 > > 4 0 > > 5 danny dan 1 > > 6 sean shawn 1 > > 7 aaron 0 > > > > Thanks, > > Wei > > > > > > > > > > > > >