T-Test Conditional Logic Using Two Datasets

[naomip]

Hi all,

I am running a t-test for variables NHH and AGWRKRTYP_G. I want to modify the output from PROC TTEST using conditional logic so that if ProbF > 0.05 (from the equality output dataset) then display the Pooled method p-value and not the Satterthwaite p-value (from the ttest dataset).

Am I able to use conditional logic using two datasets if I do not want to merge them? Is there a way to do this with SQL? 

 

ODS TRACE ON;
ODS OUTPUT	TTests		=	work.ttest_nhh;
ODS OUTPUT	Equality	=	work.equality_nhh;
PROC TTEST	DATA	=	import.agworker_sub118	ALPHA	=	0.05;
	VAR	NHH;
	CLASS	AGWKRTYP_G;
	RUN;
ODS TRACE OFF;

[PaigeMiller · Accepted solution]

Seems like you could use a simple macro %IF.

From the equality output data set, test to see if ProbF>0.05, assign a macro variable to contain a 1 or 0 depending on ProbF>0.05 or not, and then display the pooled or Satterthwaite p-value depending.

Example: 

ods output ttests=ttest;
ods output equality=equality;
proc ttest data=sashelp.class;
    class sex;
    var height;
run;

data _null_;
    set equality;
    call symputx('equality_of_variance_test',probf>0.05);
run;

%if &equality_of_variance_test %then %do;
    proc print data=ttest(where=(method='Pooled'));
    run;
%end;
%else %do;
    proc print data=ttest(where=(method=:'Satt'));
    run;
%end;

[naomip]

Thank you! That worked perfectly.