In this study on K-12 schools in the U.S. Florida Heartland, I take a QuantCrit approach to uncover how processes of data transformation, which I call 'racial re-formation', shape the utilization and reporting of racial and ethnic representations of students. To understand actual data use at schools, I apply QuantCrit's principles on how numbers are not neutral and how categories/groups are neither 'natural' nor given. I introduce 'racial re-formation' to identify the mechanisms behind expression of the QuantCrit principles in the data. Analyses of observations demonstrate that definitions of race/ethnicity are malleable, as personnel changed reported definitions of students as they used data in school decisions. Contrast of school records with teacher-created grade-level cards shows mismatch in student race/ethnicity between multiple school measures, while comparison of state vs. school records illustrates how policy affects demographic portraits. These practices led to varying representations, especially for American Indians and other groups, though ways to improve data collection, utilization, and reporting are offered.