I ran into this question in my class and am not sure how to solve it:

A positive test result gives you a Bayes factor of 71 in favor of being sick. If your prior probability of Being sick was 0.05, what is your posterior probability of being sick.

I understand that a Bayes factor is basically the likelihood ratio and I know that $p(M_2)=.01$. I guess I am looking for $p(M2|y=\text{testing positive for sick})$, but I am missing all the other variables.

Can someone please give me a hint how to proceed?

  • $\begingroup$ Please add self-study to your tags as it will avoid answers that provide a complete resolution of this problem, without letting you gain by solving the final step. $\endgroup$ – Xi'an Nov 18 '18 at 16:17

Given data $D$ and two models $M_1$ and $M_2$, the Bayes factor is defined as:

$$BF_{1,2} \equiv \frac{p(D|M_1)}{p(D|M_2)}.$$

So you can write the posterior probability of $M_1$ as:

$$\begin{equation} \begin{aligned} \mathbb{P}(M_1|D) &= \frac{p(D|M_1) \mathbb{P}(M_1)}{p(D|M_1) \mathbb{P}(M_1) + p(D|M_2) \mathbb{P}(M_2)} \\[6pt] &= \frac{BF_{1,2} \cdot \mathbb{P}(M_1)}{BF_{1,2} \cdot \mathbb{P}(M_1) + \mathbb{P}(M_2)} \\[6pt] &= \frac{BF_{1,2} \cdot \mathbb{P}(M_1)}{BF_{1,2} \cdot \mathbb{P}(M_1) + (1-\mathbb{P}(M_1))}. \\[6pt] \end{aligned} \end{equation}$$

Also note: Bayesian statistics is named after the statistician Thomas Bayes. Since this is a proper noun, it is always expressed as "Bayes", not "bayes" or "Bays".

| cite | improve this answer | |

Your Answer

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

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