Do high-value-added teachers improve student results?

Responsible researcher: Viviane Pires Ribeiro

Article title: MEASURING THE IMPACTS OF TEACHERS II: TEACHER VALUE-ADDED AND STUDENT OUTCOMES IN ADULTHOOD

Article authors: Raj Chetty, John N. Friedman and Jonah E. Rockoff

Location of intervention: United States

Sample size: Approximately 2.5 million children         

Big theme: Education

Type of Intervention: Teachers' impacts on students' long-term outcomes
Variable of Primary Interest: Teacher Value Added

Assessment method: Score Test

Assessment Context

One prominent but controversial method is to evaluate teacher effectiveness based on their impact on student test scores, commonly referred to as the Value Added (VA) approach. Major school districts have already embraced using VA to evaluate teachers, such as districts from Washington, DC to Los Angeles. Some authors argue that selecting teachers based on their added values ​​can generate substantial gains in student performance, while others, more critical, state that these are not good measures to assess the quality of teachers.

Intervention Details

Chetty et al. (2014) studied the long-term impacts of teachers using school and tax records for more than a million children. Information about students—including enrollment history, test scores, and teacher assignments—was obtained from the administrative records of a large urban school district. These data span the school years 1988–1989 to 2008–2009 and cover approximately 2.5 million children in grades 3–8. On the other hand, information on students' adult outcomes was obtained from United States federal income tax returns for the period 1996 to 2011.

School district records were linked to tax data using an algorithm based on standard identifiers (date of birth, state of birth, gender, and names), after which individual identifiers were removed to protect confidentiality. In total, the authors combined approximately 90% of observations from school district data with fiscal data, thus allowing them to track nearly a million people from elementary school to early adulthood in order to measure outcomes such as salary, college attendance and teenage birth rate .

Methodology Details

Two research designs were used to estimate the long-term impacts of teacher quality: cross-sectional classroom comparisons and a quasi-experimental design based on teacher turnover. The first project compares student outcomes for teachers with different VA, controlling for a set of student characteristics such as prior test scores and demographics. The authors implemented this approach by regressing long-term outcomes on VA estimates from the test score. The identifying assumption underlying this design is the selection of observables: unobserved determinants of adult outcomes (such as student ability) should be unrelated to teacher VA, depending on observable characteristics. On the other hand, the second project relaxes this assumption of selection of observables, exploring teacher turnover as a quasi-experimental source of variation in teacher quality.

Results

In the first research project, the results found by Chetty et al. (2014) show that a one standard deviation improvement in teacher value added in a single grade increases the probability of students attending college at age 20 by 0.82 percentage points, relative to a sample average of 37%. Improvements in teacher quality also increase the quality of the colleges students attend, as measured by the average earnings of previous graduates.

Students who study with teachers with higher VA have steeper earnings trajectories in their 20s. That is, at age 28, a one standard deviation increase in a teacher's VA in a single grade increases students' annual earnings by 1.3%. If the earnings impact remains constant at 1.3%, students will receive an average of $39,000 in earned income over their lifetime. Discounting at a rate of 5% generates a present value gain of $7,000 at age 12, the average age of students.

Improvements in the quality of teacher instruction significantly reduce the likelihood that students will have a child during adolescence, increase the quality of the neighborhood in which a student lives as an adult (measured by the percentage of graduates from that zip code), and increase participation rates in retirement plans.

In the second project, the results show that an improvement of one standard deviation in the teacher's VA increases the probability of students attending college at age 20 by 0.86 percentage points, a value almost identical to the estimate obtained in the first project. Improvements in teachers' average VA also increase the quality of the colleges students attend.

Public Policy Lessons

Do high-value-added teachers improve students’ long-term results? The results found by Chetty et al. (2014) show that test score value-added measures are a good proxy for teachers' long-term impacts. In this sense, students assigned to high-value-added teachers are more likely to attend college, receive higher salaries and are less likely to have children as teenagers. Replacing a teacher whose value added is less than 5% with an average VA teacher would increase the present value of students' lifetime income by approximately $250,000 per classroom.

The authors state that two issues must be resolved before determining the use of value-added in educational policymaking. First, using this measure to evaluate teachers may induce responses such as teaching to the test or cheating. Thus, it is possible to estimate the magnitude of such effects through the analysis developed by the authors in a district that evaluates teachers based on their added value. Second question, one must compare the long-term impacts of teacher evaluation based on VA to other metrics, such as principal evaluations or lesson observations.

However, regardless of whether or not value-added is used as a policy tool, the results show that parents should value having their children in the classroom of a high-value-added teacher and the improvement in teaching quality is likely will bring economic and social benefits.

References

CHETTY, Raj; FRIEDMAN, John N.; ROCKOFF, Jonah E. Measuring the impacts of teachers I: Evaluating bias in teacher value-added estimates. American Economic Review, vol. 104, no. 9, p. 2593-2632, 2014.