topic Re: Iteratively running a procedure, substituting variables from a list? in SAS Programming

Iteratively running a procedure, substituting variables from a list?

SAS93 — Wed, 10 Nov 2021 14:05:31 GMT

I'm trying to run a simple proc surveylogistic using one variable at a time to retrieve trends overtime for each variable. I don't want to have to copy/paste the variables each time and re-run (there are 55 of them, total). So I thought about creating an ARRAY containing the list of variables, and telling SAS to iterate through that list, one at a time, then finally print each of those "final outputs" into the results window.

But I'm not sure how to do this. Or if there is a better option, considering I have to create the ARRAY in the data step, then somehow pass it to the proc step? Below is an example of what I've been attempting, based off a suggestion I saw elsewhere on another forum:

*Create array list;
ARRAY roll	[4]	
var1 var2 var3 var4;

DO i=1 to 4;

    *Select only table of interest;
	ODS Select  ParameterEstimates;
	ODS output  ParameterEstimates= roll[i];
	PROC SURVEYLOGISTIC Data=rollrpt3 Total=TOTALS_DATASET NOMCAR;																		
	Weight wghtvar;
	Strata strat_var;
		Model roll[4] = year;
	Run;  
	ODS close;

	*Print estimate and its associated P-value for trend;
	Proc print data=roll[i];
	Var Variable ProbT;
	Where Variable="year";
	run;
Quit;

Re: Iteratively running a procedure, substituting variables from a list?

PaigeMiller — Wed, 10 Nov 2021 14:04:54 GMT

You could use a macro, such as this example:

%let varnames=var1 var2 var3 var4;
%macro dothis;
    %do i=1 %to %sysfunc(countw(&varnames));
         %let thisname=%scan(&varnames,&i,%str( ));
         proc surveylogistic data=rollrpt3 /* whatever other options you want */ ;
               model &thisname = year;
               /* Any other needed statements in PROC SURVEYLOGISTIC */
          run;
     %end;
%mend;
%dothis

I am skeptical that SURVEYLOGISTIC is a good tool to model a variable named YEAR as a predictor variable. But I don't know the goals of this study ... why don't you tell us the goals?

Re: Iteratively running a procedure, substituting variables from a list?

SAS93 — Wed, 10 Nov 2021 14:09:13 GMT

Thank you for the suggestion! I'm using survey data to assess change in overall prevalence of various dichotomous indicators (the variables in my list) over a continuous variable for time (years) instead of setting it as categorical. The prevalence estimates need to be properly weighted, so survey procedures it is.
Would you recommend using surveyREG instead of logistic?

Re: Iteratively running a procedure, substituting variables from a list?

PaigeMiller — Wed, 10 Nov 2021 14:29:11 GMT

To tell you the truth, I was thinking year should be a CLASS variable if you are doing a study with random sampling over many years. But I don't know your data. And I don't know your study design.

Also, there's no way I can tell what PROC you should be using since I don't know what VAR1 is. For PROC SURVEYLOGISTIC, you need to have VAR1 as categories. If VAR1 is continuous, then you would want to use PROC SURVEYREG.