2016 - The Impact of Inclusive STEM High Schools on Student Achievement

Attribution: Gnagey, Jennifer, & Lavertu, Stéphane
Researchers: Jennifer GnageyStéphane Lavertu
University Affiliation: Weber State University; The Ohio State University
Email: jennifergnagey@weber.edu
Research Question:
To estimate the impact of "inclusive" science, technology, engineering, and mathematics (STEM) high schools.
Published: Yes
Journal Name or Institutional Affiliation: American Educational Research Association
Journal Entry: Vol. 2, Issue 2, pp. 1-21
Year: 2016
Findings:
  1. The results indicate that only one school has had a positive effect on achievement- a positive effect in science that appears to have come at the expense of achievement in social studies. On the other hand, three of the six schools are associated with substantial negative effects across both STEM and, particularly, non-STEM subjects.
  2. The analysis reveals the critical importance of accounting for students’ prior test scores in science, in addition to math and reading, when estimating the impact of these STEM high schools. Doing so alters significantly the estimated STEM school effects across all subjects.
  3. Generally, student achievement suffered most in non-STEM subjects and among African Americans.
  4. Inclusive STEM-schools focus on problem-solving skills and group work that standardized exams may not capture. If some of these schools simply help foster student interest and attainment in STEM fields (via early college opportunities, for example), they may very well achieve their goals of improving and broadening the STEM workforce pipeline in spite of lower student achievement.
Scholarship Types: Journal Article Reporting Empirical ResearchKeywords: Academic AchievementPreparationRaceSTEMSTEM SchoolRegions: MidwestMethodologies: QuantitativeResearch Designs: Administrative DataAnalysis Methods: Descriptive StatisticsInstrumental VariablesOLSPropensity Score Matching Sampling Frame:Students in Ohio
Sampling Types: Non-Random - PurposiveAnalysis Units: SchoolStudentData Types: Quantitative-Longitudinal
Data Description:
  • Data comes from students that attended 6 inclusive STEM high schools and feeder district students.
  • The analysis employs student-level administrative data collected by the Ohio Department of Education from the 2005-06 through 2012-13 school years. Some of the schools in this study are chartered, independent high schools and others are schools established and run by traditional public school districts.
  • Because all students are tested in both 8th and 10th grade, the authors focus on the estimation of two-year STEM high school treatment effects. This study tracks six cohorts of students at each platform school from 8th through 10th grade.
  • Of the six STEM schools in the study, the 4 district STEM schools draw primarily from the districts that operate them, whereas the 2 independent STEM schools draw from numerous districts. Feeder district students (which are non-STEM schools)- the control group in this study- are those residing in districts where at least one student in their cohort attended a STEM school for two full academic years.
  • Essentially, students in these models are exactly matched based on their cohort and district of residence. The lagged test scores control for the pre-treatment effects of observable inputs and the time-constant effect of the unobservable initial endowment that affects student achievement levels. Similarly, in subsequent models, the authors further disaggregated estimates by cohort within schools in order to examine trends in STEM school impact over time. Finally, they estimated all models individually using student samples restricted by race and gender to examine how STEM school attendance affected the achievement of student groups that are thought to benefit from inclusive STEM schools.
  • Sample sizes: School A=6,205 and 30,307, School B=9,407, School C=3,526, School D=6,344, School E=460, School F=220.
  • Propensity Score Matching Variables:previous achievement in reading, science and math, gender, race, SES, limited English proficiency, disability, giftedness, attendance to charter school.
  • For the instrumental variable analyses for School A the instrument utilized were admission lottery results.
Theoretical Framework:
Relevance:STEM-focused Schools
Archives: K-16 STEM Abstracts