3
$\begingroup$

Any help would be appreciated, as I'm not sure how to proceed. I have a dataset with binary data and have been able to fit a logistic regression model to it (presence of disease, for example, as a function of age). The data coverage is large, spanning ages 0-90, and I believe I can then use the output betas to predict the probability of disease to other people in the population. I've been considering these predicted probabilities as "priors". Suppose there is then a different study on the same topic, but samples only include children. This study uses a uses a totally different methodology, but the methodology provides a probability of disease as well. What is the proper way to update the priors for children with this new dataset output? I was thinking that I could just use Bayes' theorem, but the second dataset is not in any way related to the first (so that I cannot update the model parameters) and I only have the output predicted probabilities. Thank you in advance.

$\endgroup$
  • 3
    $\begingroup$ When you say that the second dataset is not in any way related to the first, do you mean that the datasets have different covariates, or that they're arising from entirely different situations? $\endgroup$ – P Schnell Aug 21 '14 at 0:11
  • $\begingroup$ To @PSchnell's comment I'll add that if the second dataset is not in any way related to the first how can it yield predicted probabilities for the cases from the first dataset? $\endgroup$ – rolando2 Aug 11 '18 at 12:30

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.