The Center for Analysis of Longitudinal Data in Education Research (CALDER) at the Urban Institute (UI) proposes a RAPID proposal to address an urgent Congressional request for information on successful STEM schools. Specifically, the Center, through the access it has to longitudinal state databases from a US Department of Education award, will conduct a preliminary analysis of how to create a survey using current state data and will query the data for what constitutes successful STEM schools. Through the use of hierarchical linear modeling (HLM) methods, the project will identify school-level residual effects of STEM fields, holding prior student achievement constant. This work should contribute substantially to an approach for measuring efficacy across diverse programs.
?In 2005 and 2010, the National Academies' Committee on Prospering in the Global Economy of the 21st Century released two influential reports focusing national attention on Science, Technology, Engineering, and Mathematics (STEM) education. These documents highlighted persistent deficiencies in K-12 education in STEM fields and advocated federal intervention to remedy these deficiencies, particularly among underrepresented minorities (URMs) in STEM occupations. Yet, in spite of the growing urgency to make targeted interventions, relatively little empirical research has been conducted to assess the relationship between STEM educational inputs and student learning in STEM subjects. Empirical research is therefore urgently needed to properly tailor policy interventions in STEM, particularly analyses using longitudinal data that can trace the growth of students' learning trajectories over time. This project used longitudinal education data from two states (Florida and North Carolina) to investigate the relationship between school and instructional inputs and student achievement on standardized tests in STEM subjects. Two studies were conducted under this project, described below. The first study, titled "Characteristics of Schools Successful in STEM Subjects," categorizes schools in FL and NC on the basis of average student growth on math and science tests, then compares school-level STEM measures across these performance categories. This study finds several school-level variables show a significant association with school effectiveness in STEM subjects, including student participation rates in calculus and early algebra, and math and science instructional indices of classroom practice. A surprising finding, consistently documented in both states, is that of a negative association between students' STEM course participation and effective schools in STEM subjects, in addition to significantly lower participation rates of URMs in successful schools compared to average an low-performing schools. The second study, titled "Investigating the Relationship between STEM Learning Principles and Student Achievement in Math and Science," estimates the correlation between various learning activities reported on student surveys and learning gains on North Carolina's standardized math and science tests. This study finds that among the math-related classroom activities included in the survey, indicators for whether a student 'listened and took notes in math class' and 'used computers, calculators, or other machines in math class' were most positively correlated with student gains in math achievement. Similarly, indicators for whether a student 'completed a science experiment or project in science class' and 'listened to the teacher explain something about science' had the greatest positive correlations with performance on the eighth-grade science exam. These relationships held when restricting the analysis sample to URM students.
