JayUtah
Penultimate Amazing
Let's assume...
No, let's not invent hypotheticals. I'm talking about the tests you've already done, which use different sample pools, different trial probabilities, and different methods of filtering the data. Your past tests are incongruent and therefore cannot be aggregated.
If you would like to propose new tests, using the same protocols and the same sample pools, then you can make a case for aggregation. But that's not what you've done. And as has been belabored, your method is irreproducible. Since you rule on each data point using subjective ad hoc methods, you don't even have one dataset for which a valid statistical model exists.