0
$\begingroup$

Within a fixed effects approach, the effects of invariant variables cannot be estimated. Their effects are captured by the fixed effects. However, when I estimate following Bayesian fixed effects model in R

bayesglm(y ~ variant variables + invariant variables + factor(identity variable 1), family = gaussian())

by means of the function bayesglm, I get estimates for the invariant variables, too.

I am not familiar with Bayesian approaches, hence my question: Why is it possible to get these estimates in a Bayesian fixed effects model?

Thanks in advance,

Sören

$\endgroup$
  • 1
    $\begingroup$ Could you describe your model a little bit further? Have you checked if the "estimated" effects are not exactly the same as your prior distributions and/or do not have very large variances..? What R package do you use for estimation? $\endgroup$ – Tim Feb 6 '15 at 14:39
  • $\begingroup$ Hi Tim. I am using the R-package arm. The estimates for the fixed effects are identical with the estimates of a glm estimation. However, in the glm estimation I cannot estimate the effects for invariant variables. This I can only do with the Bayesian function bayesglm. $\endgroup$ – Sören Feb 6 '15 at 14:47
  • 1
    $\begingroup$ and my question was about the estimates of invariant variables $\endgroup$ – Tim Feb 6 '15 at 14:50
  • $\begingroup$ Please provide the general framework. If the question is solely about the R package, it should migrate to Stack Overflow. $\endgroup$ – Xi'an Feb 6 '15 at 15:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.