It seems like we can learn so little by selecting only for jobless people. The best way to do a BI test seems to me to get a random sample so we can test how i

It seems like we can learn so little by selecting only for jobless people. The best way to do a BI test seems to me to get a random sample so we can test how it affects people in various circumstances. Selecting a specific demographic like this seems more a political move than a scientific one.