Statistical Assumptions and Social Reality: A Critical Analysis of Achievement Models

The statistical assumptions of asymmetry and linearity in path models of academic achievement are critically examined using data from a sample of Connecticut high school seniors. Analysis suggests that student self-evaluations of academic capability and school performance may be reciprocally related and that academic achievement may be more resistant to change among poor than among average students. Both findings would be masked in simple recursive path models. The data are consistent with the Equality of Educational Opportunity Report's conclusion that different predispositional factors affect achievement among blacks and whites.