Adverse reaction databases could show that some batches are deadlier than others, but that does not prove that other batches are safer.

There is additional factor that heavily influence the data in such databases - was it allowed to create a records for adverse effects or not.

If different batches had different ingredients and so different outcomes, it is certainly a factor too.

So you have at least two unknown factors that could interfere with results to produce results you observe.

If we suppose that they know which batches are more harmful beforehand, then why they need multiple different batches at all?

And if they need different batches to find out most harmful, they obviously couldn't know beforehand what batch will be more harmful and couldn't administer them accordingly.

I think they didn't know outcomes, all experiment was about finding them, so average distribution of harmful and not so harmful results among the batches is constant, but in some places it was prohibited to even admit officially a possibility of adverse effects while other places post them into database. Obviously different batches was distibuted to different places.

Adverse reaction databases could show that some batches are deadlier than others, but that does not prove that other batches are safer.

There is additional factor that heavily influence the data in such databases - was it allowed to create a records for adverse effects or not.

If different batches had different ingredients and so different outcomes, it is certainly a factor too.

So you have at least two unknown factors that could interfere with results to produce results you observe.

If we suppose that they know which batches are more harmful beforehand, then why they need multiple different batches at all?

And if they need different batches to find out most harmful, they obviously couldn't know beforehand what batch will be more harmful and couldn't administer them accordingly.

I think they didn't know outcomes, all experiment was about finding them, so average distribution of harmful and not so harmful results among the batches are constant, but in some places it was prohibited to even admit officially a possibility of adverse effects while other places post them into database. Obviously different batches was distibuted to different places.