3
$\begingroup$

I have been researching for my thesis/dissertation but I guess my knowledge about this is not so wide. I am just quite confused about my model, so if you think this question is easy to answer just doing more research, I would really appreciate some references, because I really did a research and I am still lost.

I made a survey experiment, 2x2 between subject design. I have two categorical/dummies independent variables and the dependent variable is a 7-point Likert Scale (it was a single question, so maybe a likert item). Now I am not sure if I have to perform a 2x2 ANOVA due to my experiment design, or there is another way. My hypotheses make reference first to main effects (with that two independent variables) and then I have to introduce other control variables and gender as a moderator.

So, can I use ANOVA with an ordinal variable if I violate the assumptions, or should I use Kruskal-Wallis H test, or directly an ordinal regression analysis? Sorry for the long post. I would really appreciate some guidance.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

You should use ordinal logistic regression, at least as a starting point. If the proportional odds assumption is violated, you may need to go to some other method such as multinomial logisic.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Thank you for answering! Hope you do not mind I ask you something else. Can I study with ordinal logistic regression the main effects and interactions as I can do with ANOVA? My independent variables are coded as dummies, and I wanted to test weather there are differences between my treatments/independent variables (2x2 experiment). I do not know if I can achieve it with the ordinal model. Thank you! $\endgroup$
    – Irene
    Commented Aug 19, 2016 at 14:37
  • $\begingroup$ Yes. There is no reason that is not possible. $\endgroup$
    – Peter Flom
    Commented Aug 19, 2016 at 16:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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