Publication: Ordered Beta Regression: A Parsimonious, Well-Fitting Model for Continuous Data with Lower and Upper Bounds

Hi all -

Happy to share that the brms-based model I developed for modeling bounded outcomes/proportions has been published in Political Analysis. This model had its first genesis in posts about issues with beta regression and discrete values on this forum, so it is great to post this more complete version!

Ungated version here: https://osf.io/preprints/socarxiv/2sx6y/

You can run the model via my R package ordbetareg, which is a front-end to brms and allows you to use all the many brms specification options: Introduction to ordbetareg.

Thanks to @matti @lindeloev and @stijn for helpful comments on my early ideas!

12 Likes

fyi the link you provided for the Political Analysis paper seems to be connected to your web browser’s proxy service (… proxy.library.nyu.edu …), its asking me to sign in

sorry about that! I changed the link and also added the ungated version too.

Great work!

1 Like

I am using this to analyze three outcomes for a Morris water maze experiment, to wit:
Proportion of total time in maze spent in “Target Quadrant”.
Proportion of total time in maze spent in “Target Annulus”.
Proportion of total time in maze spent in “Thigmotactic behavior”.
The first two are estimates of how well mice may remember a location. The last can reflect levels of anxiety in the mouse.

1 Like