The current article presents additional analyses of a classwide mathematics intervention, from a previously reported randomized controlled trial, to offer new information about the treatment and to demonstrate the utility of different types of effect sizes. Multilevel modeling was used to examine treatment effects by race, sex, socioeconomic status, special education status, and achievement risk status and did not indicate differences in intervention effects on year-end state test scores by subgroup. Multilevel modeling analyses found score differences on spring curriculum-based measurements by race, treatment assignment, and prior educational risk, but only treatment assignment predicted differences in gains over time, favoring the intervention group for two of the three measures. Race and treatment assignment predicted differences in gains for one measure. Nonproficient scorers were proportionate by all demographic categories in the intervention group, but in the control group, there were disproportionately higher numbers of African American students, students receiving special education services, and students with an initial achievement risk who scored in the nonproficient range on the year-end test. Risk reduction analyses, including absolute risk reduction, relative risk reduction, and number-needed-to-treat (NNT) estimates, were computed overall and by subgroup. Overall, the data suggested stronger intervention effects for students who began the intervention at greater risk, including students of minority ethnicity (NNT = 4), students receiving special education (NNT = 3), and students with initial achievement risk (NNT = 3), in this sample.