R tutorial: Ordinal regression

This tutorial will show you how to run an ordinal regression in R and write it up. It covers model fit, pseudo-R-squares and regression coefficients, plus an explanation of how to interpret the regression coefficients. I will also show you how to produce confidence intervals and odds ratios. Finally I will explain how you can test the assumption of proportional odds/parallel lines. Data and code can be found here
drive.google.com/drive/folder...
Table of Contents:
00:00 - Introduction
13:25 - test of parallel lines/proportional odds

[@thetruth5708]

Thank you very much this video! It's brilliant and it really safed me!!!!

[@adamholden2558]

Your videos have helped me so much! Wrapping up my masters and you've been a life saver

[@DrPC_statistics_guides]

Thank you!

[@Jockeymahorn]

Thanks, I´ll try it with the data!

[@charlesbwilliams]

This was super useful for me. I had a research paper a few semesters ago and I decided to go back and fix it. Before, I only used frequency charts but using ordinal regression for my paper helped me better answer my research question. Thanks for the video and the links you provided!

[@DrPC_statistics_guides]

Really great to hear, thanks for commenting!

[@rakelbeserra]

Thank you very much!!!🤩

[@kirstyfinlayson9726]

Thank you so much you are a life saver for non-sciencey scientists...

[@DrPC_statistics_guides]

Thanks!

[@carlacadena6525]

wooow thank you!!!

[@Mspersadr]

Hi, this video is super helpful, thank you! I was wondering if you are planning to make a video for ordinal regression with mixed effects (clmm)? Thanks again!

[@lealemler2967]

Great video, thank you. I just wonder in minute 12:23 how you know that the estimates are for liking coldplay (so assume category=pay to see them?)? You say here "There was as significant positive association between stupidity and liking Coldplay B=1.38, SE=0.55,...In basic logistic regression it is possible to run "levels" so I know what I am comparing to what, but i dont quite understand how I can see the effect of e.g. stupidity in case of 4 categories on one category? Hope this makes sense.

[@nilenefer]

what would work for test of parallel lines if you have a clmm model (ordinal logistic with mixed effects)? brant() does not work.

[@amogh2101]

Awesome video! I have question regarding comparing models. In the video, you compare the model to a null model. What if I want to compare one model (say with two independent variables) with other model (with three predictor variables). Is that something that I can do to **choose** one model for regression?

[@DrPC_statistics_guides]

Yes you can do that and then compare the models in exactly the same way. That's essentially a hierarchical model

[@amogh2101]

@@DrPC_statistics_guides Thank you! Just a follow-up question regarding Parallel Lines assumption. I see that in your model, the female variable's p-value was 0.08. What if that value was less than 0.05 (e.g., 0.03)? Is it okay to proceed with ordinal regression if Omnibus test's p value was greater than 0.05?

[@krimathakker6236]

What should I do if my likelihood ratio tests of cumulative link models does not have significant p-value (0.7233)? Thank you!

[@DrPC_statistics_guides]

Just write it up the in the same manner, essentially your model is not predicting a significant amount of variance to the null (intercept only) model

[@ozbirog]

What if only one (out of 9) of our variables doesn't pass brant test (p-value in this test 0.04)? Should we resign from conducting ordinal regression and choose MLR instead or exclude this variable although it's significant in the model (p - value 0.0003) or are there any other ways do deal with that?

[@DrPC_statistics_guides]

I would exclude it as it doesn't meet assumptions, the p value associated with it may be significant but the critical value that p value comes from cannot be trusted ergo nor can the p value

[@Mavis_1991]

Hi Prof. Paul Christiansen. I tried to copy what you did in 5:48 of the video wherein you typed the modelnull and model1 and you performed, "anova(modelnull,model1)."
When I performed the anova function in R, this is the error message that prompted, "Error in UseMethod("anova") : no applicable method for 'anova' applied to an object of class "c('double', 'numeric')."
Could you please help resolved this error message? Thank you in advance!

[@danibeanz]

thanks for such a useful video. question - does the response variable have to be a factor? I have likert results (1-10) as an integer in my dataset.

[@DrPC_statistics_guides]

It must be labelled as a factor to run but you can just label your variable as a factor and it will run

[@danibeanz]

@@DrPC_statistics_guides thank you that's what I did :)

[@julieroudaut9307]

Hello,
Your video helped a lot but I have some problem with my data. I hope someone can help me here, so here is my problem:
I try to treat ordinal data as a function of time in order to analyze if there is a significant difference between two different modalities.
My dataset consists of 30 readings taken each week in two different modalities. The ordinal data are scores given to individuals according to the abundance of aphids found on them (absence, a little, a lot).
I performed a linear regression using the polr function but time is taken into account here as a fixed factor.
How can I correctly interpret such a data set.
Thank you

[@landersebastian7886]

Good day sir, may I ask for your help

[@landertarroza.sebastian1340]

good day sir. may I ask for your help

[@TammyWilkinson-ed9km]

How should this code be cited? Thanks!

[@DrPC_statistics_guides]

Just cite the r package used

[@TammyWilkinson-ed9km]

@@DrPC_statistics_guides Thanks!

[@Chris-er5wz]

ordinal package wont load...