A lot of social scientists do understand selection bias.
Look up Perry Preschool and Abecedarian for randomized experiments on early childhood education and their positive findings. (Yes, academic score increases diminish over time, mostly gone by 8th grade, but long-term the treated kids are doing MUCH better in terms of things like incarceration rates, teen pregnancy, dropping out etc., especially the girls). Both Michal Anderson (Berkeley) and Jim Heckman (UChicago) have recent work looking at the long-term outcomes of those experiments.
I can't remember specifically who did the work (other than Carolyn Hoxby and Jesse Rothstein, but I don't want to go there... there are less controversial studies), but there are many relatively clean papers looking at charter schools and school choice using things like school lotteries and other instruments as exogenous variation to get around problems with selection bias. It's not a solved problem, but not all charter schools and school populations are the same either. The current conventional wisdom among experts (who understand how to find causation) is that some charter schools are better than regular schools and some are worse. On average they're about the same in terms of quality, though charter schools seem to be doing about the same at a lower expense. Why and when that is is still an area of intense study.
