03-21-2017 03:24 PM
I would like to run a logistic model in the binary outcome (Y). In the multivariates, I only need the condition 1 and 2. Therefore, I use "and" to select all of them. Am I correct? Or should I use "or"? Thanks.
proc logistic data=test;
class Y (ref="1") X1 (ref="2") X2 (ref="2") X3 (ref="2") X4 (ref="2") X5 (ref="2");
model Y = X1 X2 X3 X4 X5;
where X1 in (1,2) and X2 in (1,2) and X3 in (1,2) and X3 in (1,2) and X3 in (1,2);
03-21-2017 03:41 PM
Only you really know based on your logic. Make some test data and ensure it behaves the way you expect.
Scenarios to test:
X1, X2, X3, X4, X5 all are either 1/2
Any of X1-X5 is not a 1/2
All of X1-X5 is not a 1/2
You may need to break these out into smaller cases but you'll probably understand your logic better at that point.
03-21-2017 04:59 PM
Get more insight about your data with:
proc sql; select x1, x2, x3, x4, x5, count(*) as n, case when X1 in (1,2) and X2 in (1,2) and X3 in (1,2) and X4 in (1,2) and X5 in (1,2) then "Yes" else "No" end as included from test group by x1, x2, x3, x4, x5; quit;