3 ms·
Some statistician correct me if I'm wrong, but that doesn't work. You are assuming the likelihood of a test being wrong on the same individual is random every t
by didibus 6y ago
Some statistician correct me if I'm wrong, but that doesn't work. You are assuming the likelihood of a test being wrong on the same individual is random every time it is taken. Which I don't think is something we know yet.
- tankenmate 6y agostatistically it depends on whether tests are independent; i.e. if the cause of the false positive is to do with the subject then using three tests won't help. if the cause of the false positive is entirely random then doing three tests will help improve the accuracy. if the correlation between false positive occurrence and the testing parameters are somewhat correlated then it may help but maybe not as much as you'd want.
- loeg 6y agoEven worse than that -- even if statistically independent, 3 instances of a 33% individual test chance of true positive does not mean 100% chance of a true positive result after 3 tests. Something like (1 - (1-33%)^3) gives you a 70% chance of true positive after 3 totally independent 1/3 change trials.