Estimation of the relative validity of employee selection tests from information commonly available in the presence of direct and indirect range restriction.