This study examined whether the combination of five charter school best practices could improve student achievement within traditional public schools in Texas. Results suggest that implementing charter school best practices in public schools significantly increased student math achievement, but had little effect on student reading achievement.
Policy Issue:
Evidence on the efficacy of charter schools in the United States demonstrates that there exist combinations of school inputs that can significantly increase the academic achievement of disadvantaged and minority children. However, such high-performing charter schools only serve a small share of U.S. students in kindergarten through grade 12. A potential strategy to more broadly improve student achievement and combat the racial achievement gap is to try to infuse the educational best practices of the most successful charter schools into traditional public schools. This study examined how five such practices affected student achievement within traditional public schools in Houston, Texas.
Context of the Evaluation:
Texas has been on the cutting edge of many education reforms, including the charter school movement. The Houston Independent School District (HISD) is the seventh largest school district in the nation, with 276 schools that enroll more than 200,000 students. Roughly 88 percent of the student population is black or Hispanic, 80 percent are eligible for free or reduced-price lunch, and 30 percent have limited English proficiency.
Details of the Intervention:
Researchers conducted a randomized evaluation to measure the impact of introducing best practices from charter schools into Houston’s public schools on student learning outcomes in math and reading. Among sixteen of the lowest-performing elementary schools, researchers randomly selected eight schools to implement the best practices; the other eight served as the control group. The two worst elementary schools were also included in the program at the insistence of the Houston school board.
Because the school board insisted that some of the lowest-performing schools be treated, researchers also made quasi-experimental comparisons where randomization was not possible. Five middle schools deemed “academically unacceptable” and four high schools deemed “failing” also implemented the best practices. For these schools, the entire Houston Independent School District (HISD) served as a control group.
Researchers introduced the following five charter school best practices in all selected schools:
Extended learning time: Extra days were added to the school year and learning time during the school day was increased by lengthening the day or eliminating breaks. In total, students received 21 percent more instruction time than they did in the year before the program began.
Human capital: Nineteen out of twenty principals and 46 percent of teachers were replaced with new staff vetted by the researchers for quality and increased value-added. Teachers also received additional training each summer and approximately ten times more observations and feedback than teachers in control schools. These observations were an important part of the teachers’ yearly evaluations, which also included standards on professionalism and student performance.
High-dosage tutoring: High-need fourth graders and sixth and ninth grade students received intensive math tutoring. In non-tutored grades, students who tested below grade level received an extra class in the subject in which they were furthest behind.
Data-driven instruction: The schools administered three benchmark assessments throughout the year to set individual performance goals for students for the end-of-year state exam.
Culture of high expectations: The Superintendent outlined a clear set of goals and expectations for schools; all teachers were expected to adhere to a professional dress code; and schools and parents signed “contracts” to honor school policies.
Researchers measured student learning outcomes based on state test scores administered at the end of every year. Administrative data provided by HISD included demographic information such as student race and gender as well as state math and reading test scores for approximately 200,000 students across the Houston metropolitan area.
Results and Policy Lessons:
Implementing five charter school best practices in public schools significantly increased student math achievement, but had little effect on student reading achievement. For every year spent in a treatment school, elementary school students experienced a 0.11 standard deviation increase in math scores compared to students in the control schools. Hispanic elementary school students experienced a larger gain of 0.23 standard deviations per year. However, students in treatment schools did not experience a significant change in reading scores.
The quasi-experimental comparisons showed similar results. Middle and high school students in treatment schools experienced a 0.15 standard deviation increase in math scores compared to students in control schools, but saw no significant change in reading scores.
High-dosage tutoring was particularly effective in secondary schools, where quasi-experimental comparisons demonstrated that students in tutored grades improved their math grades by 0.40 standard deviations more per year than students in non-tutored grades in treatment schools. However, there was no significant difference among students in tutored grades and non-tutored grades in treated elementary schools.
These results suggest that the set of charter school best practices, combined with high-dosage tutoring, was particularly effective in generating gains in math learning. The difference in achievement effects between math and reading is in line with existing research on charter schools, and could be due to the fact that reading scores are influenced by the language spoken when students are outside the classroom.
Related Papers Citations:
Fryer, Roland, Jr. 2014. "Injecting Charter School Best Practices into Traditional Public Schools: Evidence from Field Experiments." Quarterly Journal of Economics 129 (3): 1355-1407.
read more