1
$\begingroup$

Simple question here. I'm using Bayesian Confirmatory Factor Analysis and I can get a log-Bayes Factor with the Laplace approximation. However, I'm wondering whether I can just exponentiate this value and get the "normal" Bayes factor?

Thanks for any help!

Cheers, Pedro

$\endgroup$
2
  • 1
    $\begingroup$ The direct answer is yes you can!, the indirect one is what you expect or fear from using the exponential transform. $\endgroup$
    – Xi'an
    Commented Oct 8, 2019 at 7:53
  • $\begingroup$ Hi Xi'an, thanks so much for the comment! I'm converting to the "normal" Bayes Factor because for me the interpretation is more straightforward compared to the log-Bayes factor :) $\endgroup$ Commented Oct 8, 2019 at 23:12

1 Answer 1

0
$\begingroup$

Yes, of course.

Though careful about interpreting the uncertainties on the Bayes factor itself. It's not common that the errors on estiamtes of log evidences and thus log Bayes factors are symmetric and roughly normal, thus leading to skewed, asymmetric uncertainties on the evidences or Bayes factor.

Though in any case, often we only care about the order of magnitude of the Bayes factor. See e.g., Jeffreys' famous (logarithmic) scale for ascribing qualitative meanings to Bayes factors of different magnitudes.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.