I colleague has come to me with a question which I have answered for him but the only statistics I have done was what I did at school and a one semester course on Bayesian methods at university, so I wanted to post the question here and see if my answer is correct.

  • I have a population of $10,000$ animals.
  • I want to test for a disease for which I have a test which is $100$% effective.
  • If the disease is present in the population then prevalence of the disease will be $20$%.

How many tests must I perform to be sure to $3\sigma$, and $5\sigma$ significance that the disease is not present?

I suggested applying Bayes theorem, I have typed up my attempted solution here

  • 2
    $\begingroup$ The long Bayesian interlude seems unnecessary, one can go directly to if the disease is present, then the probability of $n$ negatives in a row is $(0.8)^n$. $\endgroup$ – André Nicolas Nov 15 '13 at 20:25
  • $\begingroup$ That's kind of obvious now you say it, I should have realised that when I got to my result! Thankyou! $\endgroup$ – rgvcorley Nov 15 '13 at 20:36
  • $\begingroup$ You are welcome. When one has a powerful tool such as Bayesian ideas, it is tempting to use it. $\endgroup$ – André Nicolas Nov 15 '13 at 20:55

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.