but that imperfection extends to the full domain of social affairs (i try to avoid “social science”), because even where you think you have a big sample, you almost never have construct and external validity. 2/