This report summarizes and discusses the results of the Texas Teacher Effectiveness Project, a two-year, replicated, naturalistic-correlational study of the relationships between presage and process variables with student learning product criteria at the second- and third-grade levels. Both linear and nonlinear relationships between predictor variables and the student learning gains criteria (class means averaged across four consecutive years on each of five subtests of Metropolitan Achievement Test battery) are reported. In general, the data make more sense in separate analyses for low versus high SES (socioeconomic status) schools than they do for the combined sample, and a great many significant relationships proved to be curvilinear or otherwise nonlinear. Major findings consistently contrast with findings from instruction in higher grade levels, suggesting that teaching fundamental tool skills in the early grades is fundamentally different from verbal manipulation of symbolic materials that typifies teaching at later grades, and therefore it should be conceptualized and studies separately. Additional points are detailed in the report. (Author)