Two solutions:

--in a data step, take the earlier file and rename the earlier date/time and create a new variable with the original name, but with one second added. Merge this new file to the later file.

--SQL:

proc sql; create table merged as select a.*,b.* from earlier(rename=(datetime=earlytime)) a , later(rename=(datetime=latertime)) b where a.earlytime+1=b.latertime ; quit;

(This will produce a file with only the cases that match. If you want the non-matching ones in there, you'll have to do some other join.)

Roger

Kim Nhac Le wrote:

> Hello, > I'm sorry about this silly question but I have been looking the text > books to figure out how to merge these datasets and i'm still stuck. > suppose I have two data sets that have a date/time in each > observations. The problem is that I want to merge the two data sets by > the date/time. BUT, the date/time aren't really the same in the two > data sets. One would be 1 second ahead of the other. > > Thanks. > > >