Proc logistic for three levels of outcomes?

ybz12003 — Mon, 16 Jan 2017 14:16:56 GMT

Hello:

I have the following modeling would like to use proc logistic to analyze. I don't have SAS in installed in my computer, my data set is mega-dataset. I am not sure whether the code I write is correct? Please correct me if I am wrong. Thanks.

Three categorical response variables: Y1 (normal), Y2 ( medium), Y3 (Severe)

Explanatory variables:

variable X1 (categorical): yes or no

variable X2 (categorical): yes or no

variable X3 (categorical): little, medium, most and all.

proc logistic data=test descending;

class X1 (ref="no") X2 (ref="no") X3 (ref="little") / param=ref;

model Y = X1 X2 X3;

format X1 yesno. X2 yesno. X3 degree. ;

title3 "Ordinal outcome (cumulative logit)";

run;

Re: Proc logistic for three levels of outcomes?

clim072 — Tue, 17 Jan 2017 05:39:39 GMT

What you had was fine. I assume that Y which your dependent variable has three levels (normal, medium and severe). In that case, Y should be listed in the class statement too.

PROC LOGISTIC DATA = test DESCENDING;

CLASS X1 (ref="no") X2 (ref="no") X3 (ref="little") Y (ref = "normal")/ param=ref;

model Y = X1 X2 X3;

format X1 yesno. X2 yesno. X3 degree. ;

title3 "Ordinal outcome (cumulative logit)";

run;

# Modify Y (ref = "normal") to Y (ref = "medium")/ if you want to change your reference group

Hope it helps!

topic Proc logistic for three levels of outcomes? in Statistical Procedures

Proc logistic for three levels of outcomes?

Re: Proc logistic for three levels of outcomes?