Two-By-Two frequency tables with all variables

[JackyK]

Hi all, 

For simple 1-way frequencies I know Proc Freq can do all variables in a data set if I just don't specify anything in the table option.

But I would like to create 2-by-2 tables with say variable Z. 

So if my data set have columns W X Y Z, I would do:

Proc freq data = have;
     table W*Z X*Z Y*Z;
run;

But my data set has like 70 variables. I was wondering if there is a way to this with all variables, or all but certain variables. 

Thanks,

Jacky

[Reeza · Accepted solution]

You have a few options. 

1. If the variables are all in order in the table (column order) you can list the first and last variables only, with double hyphens in between (--).

proc freq data=sashelp.heart (obs=50);
table ageAtStart *(bp_status--smoking_status);
run;

2. If the variables have a similar prefix you can use the colon short cut notation

proc freq data=sashelp.heart (obs=50);
table ageAtStart *(chol:);
run;

3. If neither applies you can list all, though you'll get one extra table of X*X. _all_ refers to all variables, _numeric_ all numeric and _character_ is all character variables. 

proc freq data=sashelp.heart (Obs=50);
table ageAtStart * (_all_);
table ageAtStart * (_numeric_);
table ageAtStart * (_character_);
run;

---

@JackyK wrote:

Hi all, 

For simple 1-way frequencies I know Proc Freq can do all variables in a data set if I just don't specify anything in the table option.

But I would like to create 2-by-2 tables with say variable Z. 

So if my data set have columns W X Y Z, I would do:

Proc freq data = have;
     table W*Z X*Z Y*Z;
run;

But my data set has like 70 variables. I was wondering if there is a way to this with all variables, or all but certain variables. 

Thanks,

Jacky

---

[JackyK]

Thank you, it was really helpful to know these tricks with variables as they can probably be applied in other Proc's. 

[ballardw]

---

@JackyK wrote:

Hi all, 

For simple 1-way frequencies I know Proc Freq can do all variables in a data set if I just don't specify anything in the table option.

But I would like to create 2-by-2 tables with say variable Z. 

So if my data set have columns W X Y Z, I would do:

Proc freq data = have;
     table W*Z X*Z Y*Z;
run;

But my data set has like 70 variables. I was wondering if there is a way to this with all variables, or all but certain variables. 

Thanks,

Jacky

---

You may want to think about how many values per variable are involved and now many tables (70*70= 4900) you will generate as you quite possibly can fill up a results window or spend a long time waiting for the tables to render.

I remember a job similar to this at one place I worked when all output was sent to a line printer. And finding 8 full paper boxes of output waiting.