Both arguments are based on mistaken ideas about estimating casual effects in nonexperimental research. To realize unbiased estimates, social scientists often must control for a dominant confounding ...