# Three views of a statistics puzzle

## 3 thoughts on “Three views of a statistics puzzle”

1. Trent says:

So for me, at first, this seems to yield a counterintuitive result: the positive predictive value of independent tests goes up as the incidence goes down!

But this an artifact of the specific example artificially fixing the posterior probabilities; for any given fixed PPV, as the incidence rises the test characteristics have to get worse (either SN or SP or both) to keep the same PPV.

I could see where this would trip most people up, as it did me at first, because having a fixed PPV is such a weird and un-realistic phenomenon.

1. Thanks! I’ve updated the post.