How would they measure from simulations, if he can't have the exact set of same people on different situations and your results might vary?
Social science isn't "experimentable" as biology or physics... And if you try hard to force it you will end up with what's called social engineering instead...