Although early mathematics instruction is predictive of future mathematics achievement, the effects of STEM-based mathematics instruction on mathematics gains in elementary school have been largely unexplored. The purpose of this quantitative study was to determine whether mathematics scores from third grade student state-mandated standardized mathematics test differ between students who were enrolled in STEM schools and students who were enrolled in non-STEM schools in the largest school district located in a Southwestern state in the United States. Polya's problem-solving heuristics formed the theoretical framework because of their relevance to concepts on the third grade mathematics test. Two research questions focused on intraindividual changes and interindividual changes over time in standardized mathematics test scores of third grade students who were enrolled in 18 STEM and 18 non-STEM schools. Analyses included growth curve modeling and a one-way random effect ANOVA to determine individual growth trajectories of mathematics test scores from individual schools over time from 2012 through 2017. The results indicated that there were no intraindividual differences in growth over time within schools, and there were interindividual changes in growth over time between schools, but the changes could not be explained by the independent variables, STEM and non-STEM schools. Findings were not consistent with the literature, which indicated early STEM-based mathematics instruction is more beneficial than traditional instruction. This study offers implications for positive social change by demonstrating equivalent results of STEM to non-STEM instruction, which may encourage more hands-on, inquiry-based learning for all children.