Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Supported by

Informed prior

In the current version of JASP, I realised that it is possible to center the Cauchy distribution on a specific value , instead of zero, thereby having an 'informed' prior.

However, I am really not sure how this is done. If a previous study reported an effect size of d = 0.9, do I simply center the Cauchy distribution accordingly? What about the prior width? Can I just leave it as 0.707? Is there any other information that I need?

Thank you in advance!


  • Hi esv,

    First off, there's more information here:

    Second, I think that once you step away from the default Cauchy centered at zero, you are now find yourself firmly in "informed prior distribution" land. That probably means you want to use a distribution that captures your prior expectations about effect size. Now you can do this empirically, and base the prior purely on an earlier study; this is the "Replication Bayes factor" (

    My own preference would be to use the "Oosterwijk" prior described in This seems a sensible prior for small-to-moderate effect sizes. Yes you might come up with something different, but my bet is that it will be very similar to the Oosterwijk prior, and that the end results will therefore also be similar.

    That said, I would recommend to execute both a "reference", default analysis (Cauchy with r=.707) and an informed prior analysis. This was the reader can judge the extent to which the results are robust. If you lack the space in the paper itself, you could consider to report the results in an online supplement on the OSF.


Sign In or Register to comment.