"Subclass" with proc tabulate

[dariatri]

Using proc tabulate, I want to create a frequency distribution table (n, %) of a char variable "var1" (which gets 'A' or 'B'). However, 'B' value can be further subdivided by another variable, let's say "var2" (which gets 'b1' or 'b2'). How can I get a table that shows the frequency distribution of 'A' (with no subdivisions) and 'B' (subdivided by var2) all in one table.

Thanks!

[Astounding]

You can only do this partially.  You would have to use both VAR1 and VAR2 as class variables, and nest them:

var1 * var2

The values of VAR2 would need to be constant for all VAR1=A observations.  That constant value would be the only subcategory available when VAR1=A.

[dariatri]

Ok, thanks!

[Reeza]

If you do it by VAR2 as the sub variable, you can keep it constant for group A and then it'll show as desired, though you'll still have an entry. I set it to missing and added a format to have it show up as a space. If this doesn't meet your expectations please post details of what you're looking for and what you're started with. 

*sort data;

proc sort data=sashelp.class out=class;
    by sex age;
run;

*create sublevels with age missing for female;

data class_demo;
    set class;

    if sex='F' then
        age=.;
run;

*format to display missing as a space;

proc format;
    value age_fmt low-high=[8.] .=' ';
run;

*Tabulate example;

proc tabulate data=class_demo;
    class sex age / missing;
    table sex*age, N=Count / misstext='';
    format age age_fmt.;
run;

---

@dariatri wrote:
Using proc tabulate, I want to create a frequency distribution table (n, %) of a char variable "var1" (which gets 'A' or 'B'). However, 'B' value can be further subdivided by another variable, let's say "var2" (which gets 'b1' or 'b2'). How can I get a table that shows the frequency distribution of 'A' (with no subdivisions) and 'B' (subdivided by var2) all in one table.

Thanks!

---

[dariatri]

Thank you! Very close to what I was looking for.