Analyses of tests of models using different cohorts showed that even the best outcomes showing a connection between socioeconomic status and delinquency have not equalled those for IQ for the same prevalence rates. The socioeconomic variables responsible for the best fits of the models were also the two that are best qualified to act as surrogates for IQ in the adult population. In addition, using just any set of correlates of socioeconomic status does not achieve the extraordinarily good fit to race differences in the prevalence of delinquency that IQ achieves. The author states that any effort to shift the analysis from IQ to schooling must confront the fact that the black-white IQ difference exists even before pupils enter school, that it holds for tests whose content is related to the school curriculum only remotely, and that it is not changed in the course of schooling. Finally, he contends it does not depend on differences between blacks and whites in mastering specific information, but rather on differences in the same underlying factor that accounts for individual differences within each race. Tables and 78 references.