Quote Originally Posted by CCFCC3PO View Post
The base rate would be the rate at which the sub population (i.e. age, gender, ethnicity, social situation, exposure, risk factor). So, you know how likely they are to have it before the test, now we do the test and then compare the figures. It is a critical piece of evidence.

But, clinical knowledge will always trump a test. So, if someone tests negative but the evidence is that they have all the symptoms of covid 19, then the clinical interpretation of those symptoms will trump the results of tests. Which is, I think, what you were saying earlier (about a doctor being more accurate than a test).

But, with regards to saying the test is 30% inaccurate, like I say, there are reasons that I don't readily accept that. I am not rejecting it completely, but there is a critical piece of evidence missing, and that is the base rate. The base rate will never be 0% or 100% and it will differ between sub populations (e.g. Asian women working in care homes have a base rate (known infection rate) of 32%). The issue we have here is that we do not know the actual value of the base rate because of asymptomatic cases. Also, since tests are not being done on everyone, but only on suspected cases, this leads to more false negatives.

In the same way, if they were testing solely the people who do not have the virus, we would get more false positives.
But weren't their samples weighted ?, why do they think their samples weren't weighted ?

The UK relied heavily on medical modellers and not enough on virologists was a common criticism of our lot, our lot seem to have got it wrong at every turn but the one area they should have some idea on was modelling, are you saying the UK modellers had no idea how to mathematically do their calculations ?

Yes they can make the wrong assumptions, yes they might not fully understand the characteristics of this new virus and if you put the wrong information into a computer you get wrong information out, no doubt that happened, but are you saying they don't know how to go about working it out statistically and mathematically (if they had put the correct info in) ?