In this study, longitudinal cohort-sequential latent growth modeling was used to determine the effects of (a) socioeconomically mixed preschool and kindergarten classrooms, (b) the implementation of an education program aiming to accelerate socioeconomically disadvantaged children's emergent academic skills, and (c) the amount of teacher-managed academic activities on 3- to 6-year-old disadvantaged children's emergent literacy and math skills (n = 91). The results indicate that disadvantaged children in mixed preschool and kindergarten classrooms gained more in literacy and math than disadvantaged children in targeted classrooms. The results also indicate that the use of a special education program to promote disadvantaged children's emergent literacy and math was not effective, probably because of a lack of implementation fidelity. However, the extent to which teachers engaged in the kind of activities that were intended by the education program, in particular frequent initiation and guidance of language, literacy, and math activities, was significantly related to disadvantaged children's outcomes. (author abstract)