Beyond the IRB: Selection bias and the direction of innovation in applied economics