I read that this is the intention, so when some content that can be used for harm (in a discriminatory or racist way for example), authors should make all the needed considerations, including confounding factors, in their research and their paper.
For example, say someone publishes statistics about school grades and race, and, maybe unsurprisingly, black people in the US turn out to have lower average grades, what does that say about black people? Here lies the responsibility of the researcher to make it clear that the problem is social rather than genetical, either using statistical means to remove confounders, or by highlighting this when discussing the numbers. Actually unless the purpose of the research itself is to extract those social differences, why would you stratify your cohorts by race in the first place?
Even seemingly innocuous decisions, like stratifying (because, why not?, you have the data) may have consequences, or may hide biases, if they are not considered with care. Ask anybody doing ML how confounders are behind every corner in whatever messy human data you collect!
Then there are other topics which specifically look into, say, sex and brain composition etc. Those may be scientifically sound, but ethically more controversial. I am happy I don't work in that field...
For example, say someone publishes statistics about school grades and race, and, maybe unsurprisingly, black people in the US turn out to have lower average grades, what does that say about black people? Here lies the responsibility of the researcher to make it clear that the problem is social rather than genetical, either using statistical means to remove confounders, or by highlighting this when discussing the numbers. Actually unless the purpose of the research itself is to extract those social differences, why would you stratify your cohorts by race in the first place?
Even seemingly innocuous decisions, like stratifying (because, why not?, you have the data) may have consequences, or may hide biases, if they are not considered with care. Ask anybody doing ML how confounders are behind every corner in whatever messy human data you collect!
Then there are other topics which specifically look into, say, sex and brain composition etc. Those may be scientifically sound, but ethically more controversial. I am happy I don't work in that field...