Different code producing same results

[jaimedba]

Below is my code, which is producing the same results. With the different output file names, and different regression variables, I would expect different results. I separated the two with #### to indicate a 'break' in the two requests. The bold/underlined text is what changes between the two.

Any help is greatly appreciated! I've been looking over this forever...does anyone see the problem that I cannot find?

/*BB: 3e*/
ods exclude all; /* suppress tables to screen */
ods output FitStatistics=RegEq3e_Output; /* contains results for each BY group */
proc reg data=temp.Equation3_interactions plots=none;
by datacqtr;
model lead_prccq = epsq bvpsq BB_Real_GDP_Q
BB_EPS BB_BV
cfpsq cashpsq dvpspq rdpsq
intanpsq spipsq ocipsq revpsq atpsq change_revtpsq
capxpsq cogspsq xsgapsq FF;
quit;
ods exclude none; /* no longer suppress tables */

proc print data=RegEq3e_Output noobs;
where Label2="Adj R-Sq";
var datacqtr Label2 nValue2;
title 'Rsquared Values for Full Regression Equation 3 BB';
run;

/*To keep only the Adj Rsquared Rows*/
data temp.RegEq3e_Output; set work.RegEq3e_Output;
where Label2="Adj R-Sq";
keep datacqtr Label2 nValue2;
run;

/*Merge the Adj Rsquared data with the Quarter from Econ Vars*/
proc sql;
create table temp.RegEq3e_QTR
as select *
from temp.RegEq3e_Output as a, work.EconVarFile as b
where a.datacqtr=b.datacqtr;
quit;

proc print data=temp.RegEq3e_QTR noobs;
where Label2="Adj R-Sq";
var datacqtr Label2 nValue2 datacqtrQ;
title 'Rsquared for Full Equation with Quarterly Time Variable BB';
run;

/*BB (Bloomberg Real GDP):*/
proc reg data=temp.RegEq3e_QTR plots=none;
model nValue2 = datacqtrQ;
title 'Full Regression Equation 4 BB';
run;
quit;

/*###################################################################*/

/*Jensen: 3f*/
ods exclude all; /* suppress tables to screen */
ods output FitStatistics=RegEq3f_Output; /* contains results for each BY group */
proc reg data=temp.Equation3_interactions plots=none;
by datacqtr;
model lead_prccq = epsq bvpsq jen_calc
jen_EPS jen_BV
cfpsq cashpsq dvpspq rdpsq
intanpsq spipsq ocipsq revpsq atpsq change_revtpsq
capxpsq cogspsq xsgapsq FF;
quit;
ods exclude none; /* no longer suppress tables */

proc print data=RegEq3f_Output noobs;
where Label2="Adj R-Sq";
var datacqtr Label2 nValue2;
title 'Rsquared Values for Full Regression Equation 3 Jensen';
run;

/*To keep only the Adj Rsquared Rows*/
data temp.RegEq3f_Output; set work.RegEq3f_Output;
where Label2="Adj R-Sq";
keep datacqtr Label2 nValue2;
run;

/*Merge the Adj Rsquared data with the Quarter from Econ Vars*/
proc sql;
create table temp.RegEq3f_QTR
as select *
from temp.RegEq3f_Output as a, work.EconVarFile as b
where a.datacqtr=b.datacqtr;
quit;

proc print data=temp.RegEq3f_QTR noobs;
where Label2="Adj R-Sq";
var datacqtr Label2 nValue2 datacqtrQ;
title 'Rsquared for Full Equation with Quarterly Time Variable Jensen';
run;

/*Jensen (Monetary Policy):*/
proc reg data=temp.RegEq3f_QTR plots=none;
model nValue2 = datacqtrQ;
title 'Full Regression Equation 4 Jensen';
run;
quit;

[Patrick]

First things I'd be doing:

- Check the log for Warnings and Errors - and also read the SAS Notes if there is something "telling" in it.

- Execute the two versions in fresh and independent SAS sessions. This to ensure that steps in the 2nd version can't just pick-up work tables created by the first version.

[jaimedba]

Thank you for your help! I do get this warning in the log, but I think that's okay?

WARNING: Variable datacqtr already exists on file TEMP.REGEQ3F_QTR.

[ChrisNZ]

> I do get this warning in the log, but I think that's okay?

Probably not. It means that the variable REGEQ3F_QTR from table work.EconVarFile  is dropped when you join the tables:

  proc sql;
    create table temp.RegEq3f_QTR as
    select *
    from temp.RegEq3f_Output as a, work.EconVarFile as b
    where a.datacqtr=b.datacqtr;
  quit;

If this what you want?

Never leave warnings in the log. Never.