@elmoreoocyte wrote:
I tried it myself with the data I'm working with. Doing a left merge with SAS I had the 25k results I expected, no duplicated rows. Using proc sql join, SAS merges on the missing ID and duplicates rows.
Which is what happens in a m:n situation. The DATA step MERGE will result in max(m,n) observations, the SQL JOIN in m*n observations. In both cases, missing values are used as valid keys.
... View more
