Proc SQL Left Join Based on a Condition

[E_W]

I have one dataset which is missing some coefficient data.

I have a separate dataset with a similar structure which contains much of the missing data, but when I left join them, I don't want it replacing the rows with existing coefficients. I only want the 2nd dataset's numbers to fill in the blank rows.

I've been trying out conditional proc SQL's, but nothing has worked so far. Any help would be appreciated.

Dataset1

State              City/Town      Coefficient

ALABAMA	AUTAUGA	1.01473
ALABAMA	BALDWIN	19.33969
ALABAMA	BARBOUR

[Tom · Accepted solution]

Sounds like you want the COALESCE() function.

proc sql;
create table want as
select a.state
     , a.city
     , coalesce(a.coeff,b.coeff) as coeff
from a 
left join b
on a.state=b.state and a.city=b.city
;
quit;