Solved
New Contributor
Posts: 2

# Output dataset for multiple chi squared test for association

Hi everybody,
I have a data set containing 35 categorical variable that I want to test with other 33 dichotomous variables using chi squred test. If I run
Proc freq data=mydata;
Table = c_: * d_: / chisq;
Run;
It works but i was struggling how to create a simple output data set with just three information: the two variables used in every test and the pchi.

Accepted Solutions
Solution
‎01-19-2016 03:53 AM
Posts: 5,524

## Re: Output dataset for multiple chi squared test for association

Use ODS OUTPUT to get the table :

ods listing select none;
proc freq data=sashelp.heart;
table (status sex) * (chol_status bp_status weight_status) / chisq;
ods output ChiSq=outChiSq;
run;
ods listing;

proc sql;
create table myChiSq as
select
scan(Table,2) as v1 length=32,
scan(Table,3) as v2 length=32,
df,
value,
prob
from outChiSq
where statistic="Chi-Square"
order by v1, v2;
select * from myChiSq;
quit;

Note : You need parentheses around the variable lists in the TABLES statement.

PG

All Replies
Solution
‎01-19-2016 03:53 AM
Posts: 5,524

## Re: Output dataset for multiple chi squared test for association

Use ODS OUTPUT to get the table :

ods listing select none;
proc freq data=sashelp.heart;
table (status sex) * (chol_status bp_status weight_status) / chisq;
ods output ChiSq=outChiSq;
run;
ods listing;

proc sql;
create table myChiSq as
select
scan(Table,2) as v1 length=32,
scan(Table,3) as v2 length=32,
df,
value,
prob
from outChiSq
where statistic="Chi-Square"
order by v1, v2;
select * from myChiSq;
quit;

Note : You need parentheses around the variable lists in the TABLES statement.

PG
New Contributor
Posts: 2

## Re: Output dataset for multiple chi squared test for association

This is great! Thank you so much!

Posts: 5,524

## Re: Output dataset for multiple chi squared test for association

PG
🔒 This topic is solved and locked.

Discussion stats
• 3 replies
• 243 views
• 1 like
• 2 in conversation