2
$\begingroup$

I am analysing multiple Likert Style statements with an Ordinal Logistic Regression. I have many different factors that I am taking into account, from job to research area to location.

I read that it is good practise to put results in a table, with the coefficient, CIs and the P-Value. However I do not want the reams of tables this will generate (as the way to show significance with the OLR is by looking at all factors' P values rather than one for the whole group). Do you think it would be acceptable just to write whether there was or was not significant results for each likert outcome, elaborating further in a text passage for the factors that influenced the result (i.e. reporting the coefficient / CI and P).

Apologies I know this is a simple question I just wanted opinion on what is deemed acceptable. Many thanks.

$\endgroup$
1
  • $\begingroup$ I really like Garcia's (2021) chapter (Chapter 8) on OLR. I don't think that I can copy what he'd say here due to copyright, but if you are from the human sciences and if you have the chance, he details how to perform n' how to report it. Overall, he brings the coefficients into the text by "an ordinal regression was performed n' the X condition increases the probability of being more certain (b = , p < 0.05) and Y condition decreases this probability (b = , p < 0.05) so on" (179). He also brings a line plot with the mod's predictions (using predict()) for each condition, which I really liked! $\endgroup$ Sep 8, 2022 at 13:14

1 Answer 1

3
$\begingroup$

Some journals will accept it, some won't. When I review papers that just say whether variables were significant or not, I require that they be rewritten to include effect size. It need not be "reams of tables" and, in nearly all cases, should not be. You generally need one table for each regression. If you have many bivariate regressions it is often possible to combine them into one table, provided that you make it clear what you have done and why you have done it.

$\endgroup$

Your Answer

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

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