proc means + ways + _type_ + ods output

[xxformat_com]

Hi,

I would like to know if there is a way to get something similar to the _TYPE_ variable in proc means output to identify the levels when using ODS OUTPUT instead of OUTPUT OUT=.

When having missing values in the variable listed in the class statement, there is no direct way to distinguish level 0 from level 1.

A workaround is to change the missing value to something else before proc means.

Here is some code to illustrate the issue:

data class;
    set sashelp.class end=eof;
    output;
    if eof then 
       do;
          sex=' ';
          age=.;
          output;
       end;
run;   

proc means data=class;
	ways 0 1;
	class sex /missing;
	var height weight;
	output out=test1;
run;
	
ods exclude all;
	ods output summary=test2; 
	proc means data=class;* stackods;
	    ways 0 1;
	    class sex /missing;
	    var height weight;
	run;
ods exclude none;

proc print data=test1 noobs;
run;
proc print data=test2 noobs;
run;

[PaigeMiller]

When having missing values in the variable listed in the class statement, there is no direct way to distinguish level 0 from level 1.

There are two methods I can think of to eliminate this issue. The first is that _type_=0 and missing SEX is different than _type_=1 and missing SEX. The second is shown below.

data class;
	length sex $ 7;
    set sashelp.class end=eof;
    output;
    if eof then 
       do;
          sex='Missing';
          age=.A;
          output;
       end;
run;   

proc means data=class;
	ways 0 1;
	class sex /missing;
	var height weight;
	output out=test1;
run;

proc means data=class;
	ways 0 1;
	class age /missing;
	var height weight;
	output out=test2;
run;

[xxformat_com]

Thanks but it exactly what I said in the question. There are workaround like replacing the missing value by another value. But I'm not looking for a workaround.

[ballardw]

If you want _type_ like behavior please describe why OUTPUT OUT= is undesirable.

[Tom]

Not sure why you want it but it looks like it is easy enough with  a single class variable.

Let's use the AUTONAME option on the OUTPUT statement so we are comparing similar outputs.

ods exclude all;
	ods output summary=test2; 
	proc means data=class;* stackods;
	    ways 0 1;
	    class sex /missing;
	    var height weight;
    	output out=test1 n= mean= std= min= max= /autoname;
	run;
ods exclude none;

data test2;
 set test2;
 by sex ;
 _type_=not (missing(sex) and first.sex);
run;

proc compare data=test1 compare=test2;
run;

The COMPARE Procedure
Comparison of WORK.TEST1 with WORK.TEST2
(Method=EXACT)

Variables Summary

Number of Variables in Common: 12.
Number of Variables in WORK.TEST1 but not in WORK.TEST2: 1.
Number of Variables in WORK.TEST2 but not in WORK.TEST1: 3.
Number of Variables with Differing Attributes: 10.


Observation Summary

Observation      Base  Compare

First Obs           1        1
Last  Obs           4        4

Number of Observations in Common: 4.
Total Number of Observations Read from WORK.TEST1: 4.
Total Number of Observations Read from WORK.TEST2: 4.

Number of Observations with Some Compared Variables Unequal: 0.
Number of Observations with All Compared Variables Equal: 4.

NOTE: No unequal values were found. All values compared are exactly equal.