The present research evaluated whether behavioral tasks ("direct assessments") commonly used to assess young children's social cognitive development in laboratory studies could have utility for measuring and predicting U.S. children's outcomes in educational contexts. To do so, children (N = 95; 49 boys, 46 girls; 41.05% White, 16.84% Hispanic, 14.74% Black, 13.68% Asian, 11.58% Multiracial, 2.11% American Indian/Alaska Native) in a publicly funded pre-kindergarten (4K) program in the United States completed 9 direct assessments that capture important skills in early childhood (e.g., task switching, group conformity preference, theory of mind). The school district also provided children's 4K and kindergarten grades (assessed through teacher ratings of children) to evaluate whether direct assessments had additional explanatory power over-and-above existing metrics of children's aptitude. Across the direct assessments, children's group conformity preferences (i.e., the extent to which children preferred members of the same group to behave in the same way) were most reliably correlated with concurrent (4K) and predictive of future (kindergarten) grades, even when controlling for teacher ratings of children's concurrent performance. Interestingly, teacher ratings of children on each assessment loaded onto a single factor despite the intention to capture theoretically distinct components of children's school performance. Discussion focuses on the implications of direct assessments in educational contexts and critical areas for future research at the interface of psychology and education.