topic Re: Ordinal Logistic Regression with likert scale outcomes in Statistical Procedures

Ordinal Logistic Regression with likert scale outcomes

nickgarza19 — Wed, 17 Aug 2022 14:56:27 GMT

Hey all,
I haven't worked with ordinal logistic regression before so I am trying to clear something up. I am looking at various demographic independent variables that may or may not predict a likert scale outcome. (disagree, neutral, agree). I have these coded as disagree=1, neutral=2, and agree=3. Should I model these in my class statement? I am inclined to - but I have seen examples where its treated as a continuous outcome. Any tips are appreciated.

Thanks,

Re: Ordinal Logistic Regression with likert scale outcomes

StatDave — Wed, 17 Aug 2022 15:04:21 GMT

The outcome (response) variable in PROC LOGISTIC is always considered categorical, not continuous. It is never necessary to put it in the CLASS statement.

Re: Ordinal Logistic Regression with likert scale outcomes

Reeza — Wed, 17 Aug 2022 15:08:28 GMT

@nickgarza19 wrote:

Hey all,
I haven't worked with ordinal logistic regression before so I am trying to clear something up. I am looking at various demographic independent variables that may or may not predict a likert scale outcome. (disagree, neutral, agree). I have these coded as disagree=1, neutral=2, and agree=3. Should I model these in my class statement? I am inclined to - but I have seen examples where its treated as a continuous outcome. Any tips are appreciated.

Thanks,

Independent variables ordinal/categorical data should go in your class statement.

Your outcome does not go in the class statement.

Re: Ordinal Logistic Regression with likert scale outcomes

nickgarza19 — Wed, 17 Aug 2022 17:17:53 GMT

Thank you! Sorry its been ages since I've used this.