Let us have 2 groups - treatment group (1) and control group(2). Each group has survival probability $p_1$ and $p_2$ respectively. Ofc each patient survives or dies independently given $p_i$ of his group. Let us define $y_i$ as number of survivors in group i, $n_i - y_i$ - number of deceased in group i.

So, I am setting up bayesian model for this:

$p(p_1, p_2| Y) \propto p_1^{y_1}(1-p_1)^{n_1 - y_1} * p_2^{y_2}(1-p_2)^{n_2 - y_2} * 1 $

1 comes from $Beta(1, 1)*Beta(1, 1)$ which I assume as prior for $p(p_1, p_2)$.

As I understand the posterior will be a product of 2 Betas?

Now, how can I summarize the posterior distribution for the odds ratio, $\frac{(p_2/(1 − p_2))}{(p_1/(1 − p_1))}$?

Also is the choice of noninformative prior correct in this case? Or is there a way to claim that this or that particular prior is better than others?


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.