The post describes a "data interleave" technique, not a "data merge" requirement? I really don't understand why you would want to create an output file having observations where any one contributing variable name will be blank -- that will be the behavior result when using SET instead of considering somehow doing a MERGE (however you would need to have some BY variable list to use with a merge process.
Honestly, the purpose of this example, as described in your post, appears somewhat curious, given that you have mentioned "set", even though you really mean to do some type of merge, right?
I'm a bit baffled how this point wasn't raised previously with the post/thread, frankly.
Scott Barry
SBBWorks, Inc.