GLM Mediation Model with ordinal DV?

Discuss statistics related things

by AnBraga » Tue Mar 16, 2021 3:19 pm

Hello! I dont know if this is the right place to post this, but here I go:

I would like to perform a GLM Mediation Model (jAMM module), but my dependent variable is an ordinal (ordered categorical) variable, with 5 categories. Is there anything against performing a GLM Mediation with an ordinal DV?

Thank you in advance!
Posts: 1
Joined: Tue Mar 16, 2021 3:13 pm

by » Wed Mar 17, 2021 10:42 pm

Ordered variables can be used in GLM mediation only if you can assume that they are continuous variables, otherwise, the GLM underlying the mediation analysis is not correct. If you really have ordered categories, you cannot use glm mediation. At the moment, there are not many options available in terms of software, because I'm not aware of any software (among the main statistical software) that allows doing categorical ordered dependent variables mediation.

Nonetheless, you can run a "logical" mediation analysis: First, establish that the IV has an effect on the DV, then on the Med, and then run IV+Med predicting the DV. Each of these models is run with different types of analyses (ordinal logistic or GLM) depending on the scale of the dependent and med variables. You have evidence of mediation if IV has an effect on Med, and Med has an effect on DV keeping constant IV. The limitation of this approach is that you cannot compute the mediated effect, because the coefficients (a and b in the slides) come from very different analyses, so they have different scales and the interpretation. Nonetheless, the logic of mediation holds.
User avatar
Posts: 251
Joined: Thu Mar 23, 2017 9:24 pm

by ASMA » Mon Mar 22, 2021 9:32 pm

hI Mcfanda, I also have an ordinal dependent variable, like the scale of 1-5, 6-10, 11-15 that software categorizes as 1, 2 and 3. Indepent variables and mediators are also categorical(not binary). Only control variables are continuos

If I assume it continuous variable, on what grounds can I assume it scale. I need to write a justification for the examiner.

When I put this ordinal variable in Jamm, It shows that the ordinal variable can also be inserted in Jamovi Jamm.

P.S. I used logistic regression to see the relation of IVs and DVs first because all my data is ordinal or nominal.

I would really appreciate to give me some clarity.
Posts: 1
Joined: Mon Mar 22, 2021 9:20 pm

by » Wed Mar 24, 2021 9:46 pm

The important thing is not what the software allows you to do, but what your theoretical arguments are. Two variables can be considered as ordinal by the software, but one would make sense and the other not. You can put it in this way: an ordinal variable can be considered as continuous when the distance between units (1 to 2, 2 to 3), is reasonably constant. If not, ordinal variables are not continuous. Nominals are always nominals
User avatar
Posts: 251
Joined: Thu Mar 23, 2017 9:24 pm

Return to Statistics