How to create robustness plots

Interfaces brms
1

Hi All,

My apologies if what follows seems ignorant or simplistic. I am new to the world of bayesian statistics and of brms. I wonder if you could provide me with guidance on how to create a figure that shows how the Bayes factor varies in accordance with prior width. In other words, and Ralph that would allow us to examine whether a particular value of a Bayes factor is robust to changes in the width of the prior.

Is there a pre-existing R package or function for this?
Thank you very much in the advance, for your help.

Best wishes

Tarandeep Kang

2

I am not aware of anything prepackaged, but it should be relatively easy (but computationally demanding) to create your own. Essentially, you would run a for loop over a set of prior widths you are interested in and in each iteration fit a model with different width and compute the Bayes factor. Then you just plot the resulting data with any plotting package you are familiar with.

3

Thanks, that will be easier than expected. The tip is much appreciated!