4
$\begingroup$

I am currently doing my fourth year thesis examining the moral stages of children/adolescents. The DV is the moral stage (categorical variable, stages 1, 2, 3, or 4) and the IVs are age group (I currently have 2, but may potentially have 3) and gender, so my design will be either a 2 x 2 or a 3 x 2.

I have been told by my supervisor that a logistic regression is the way to go, but it hasn't been covered in our previous statistics classes (at least not multinomial logistic regression) and he said I can otherwise use 2 chi square tests, one for each gender. I then wouldn't be able to examine any interaction between the age groups and gender.

Could anyone tell me what the best method to use would be, and if I was to use logistic regression, would it be multinomial rather than binomial, considering my DV is not dichotomous?

$\endgroup$
  • $\begingroup$ You are right that you can't use binary logistic regression because your dependent variable takes on 4 possible values. $\endgroup$ – Michael Chernick Sep 3 '12 at 14:21
4
$\begingroup$

Actually, the best thing would be to use ordinal logistic regression, since moral development stage is ordinal. You don't say what software you are using, but this can be done in most software (certainly in R and SAS). I wrote a presentation about ordinal and multinomial logistic regression that also (very briefly) covers binomial logistic. It uses SAS, but still may be helpful to you.

Also this thread on CV has some good info.

$\endgroup$
  • $\begingroup$ Thanks @peter, that is so, so helpful. I'm using SPSS. If you happen to know of any 'ordinal logistic regression for dummies' sites, please throw them my way :) I'm googling it to find out as much as I can, it was not a topic that was covered in my third year statistics course. $\endgroup$ – Melissa Tso Sep 3 '12 at 23:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.