Parameter estimates not corresponding to chosen priors

Interfaces brms
1

Please also provide the following information in addition to your question:

  • Operating System: Windows 10
  • brms Version: 2.8

I am fitting a non-linear multilevel model using brms.
I chose my priors and fit the model.
However, the parameter estimate I get is very extreme,very unlikely to be drawn from the prior distribution I chose for this parameter. Also, the plot of the prior for this estimate looks different from the originally chosen prior.
Yet, when doing the pp_check, my model seems to be okay.

So I have two questions:

  1. Why does this happen?

  2. Is this something I should worry about? Does it mean the prior I chose is not a good one and I should revise my choice and find a better one?

2

Would be helpful to share your model, and priors.

A really high level and non technical answer is:

In principle with a well working model, you can have enough data to push an inferred parameter far from your prior knowledge/assumption.

3

Hi,
Thanks for your reply!
My question was mostly theoretical.
I was mostly curious why it happens in general.
This seems to be answering my question:

you can have enough data to push an inferred parameter far from your prior knowledge/assumption.