This implies that statements can be disproven on a scientific basis. Sometimes we don't have the metrics, and we gather them too late. Sometimes statistics can

This implies that statements can be disproven on a scientific basis. Sometimes we don't have the metrics, and we gather them too late. Sometimes statistics can be interpreted either way. Look at the IQ-bias in non-western countries. Racists take it to mean that people are less 'smart' in non-western countries, the scientists working in the field have a number of opinions, from considering it a cultural bias on part of the test, to considering the bias to be correctible, and a problem created by poor material circumstances such as poverty, lack of nutrition, and inadequate education. The problem with the former, is that because the test itself is the metric, you can't 'prove' that it is biased. Are these people less intelligent, or is the test at fault? We need to devise another test to see if we can correct for the bias. But wait, that means that the second test might be biased towards non-western cultures! Societal bias in general becomes ridiculously fiddly and difficult to prove either way, simply because there are so many different places in the process where something can go wrong. From the test itself, to the pruning of the numbers, to the slice of the community selected for the tests (historically 'gender difference' studies have had such small -- and inequal test groups that they are laughable), and so on -- all can have bias. It is to some degree impossible to prove whether your test is biased or if the bias is in culture, because cultural bias can lead to material effects. How do you isolate for how the researchers treat the subjects? Are women being driven out by lack of skill, or because the male-domination of the environment has led to an inhospitable culture. How do you test for that in the first place? Etc. There are much too many variables to make it a 'simple' job.