Try

PROC SQL; CREATE TABLE ALL2 AS SELECT A.YR AS YR,A.EN AS EN,A.REP AS REP,A.A,A.B,COALESCE(A.C,P.C) FROM ALL A, NEWP P WHERE A.YR=P.YR AND A.EN=P.EN AND A.REP=P.REP ;

The COALESCE() function takes the first value unless missing, else the 2nd value. Sig ______________________________ Reply Separator _________________________________ Subject: SQL JOIN - Still having problems. Author: Dianne Brooks <dbrooks@GAES.GRIFFIN.PEACHNET.EDU> at Internet-E-Mail Date: 3/12/99 5:22 PM

Thank you, Robert Abelson Ian Whitlock, couldn't get over the quick response time.

You've gotten me along a little father... but I still can't figure out how to replace ONLY the missing values in C with the predicted value of C.

PROC GLM NOPRINT DATA = ALL;BY YR; CLASSES EN REP; MODEL A B C = EN REP; OUTPUT OUT = NEWP PREDICTED = PA PB PC; RUN;

PROC SQL; CREATE TABLE ALL2 AS SELECT A.YR, P.YR, A.EN, P.EN, A.REP, P.REP, A.A, A.B, A.C, P.C FROM ALL A, NEWP P WHERE A.YR=P.YR AND A.EN=P.EN AND A.REP=P.REP; SELECT YR, EN, REP, A, C, A.C, P.C FROM ALL2;

Thanks, Dianne