How can we prove anything in the human sciences when it is virtually impossible to run reproduceable experiments the way you can in the natural sciences? Every so often a “natural experiment” presents itself and we can come to conclusions about truths in fields such as economics. The problems of economics are present in fields throughout the human sciences for the same reasons. Below is an article that discusses the issue the effects of raising the minimum wage.
“In statistics, ‘identification’ just means separating two groups in order to tell if a treatment works. You give Group A the pill and you give Group B a placebo, and you see if Group A does better than Group B. In laboratory experiments this is usually possible to do. In the real world, it’s a lot harder — you have to wait for a policy to bring about a difference between two areas that are roughly comparable.”