What value-added models can—and can’t—tell us about teaching and learning

Getting your middle-schooler in front of a high-quality teacher for even one year will improve his or her chances of going to college and earning a good salary later in life, according to a recent study. The study’s authors used value-added modeling—predicting how well a given student will do on a standardized test, controlling for variables such as past scores and individual characteristics—to reach their conclusions.

Nationally, the question of whether value-added calculations should be included in teacher evaluations remains controversial. Critics of value-added models argue that they’re not reliable enough to be used in high-stakes decisions, and that the tests on which they’re based are themselves a poor measure of student achievement.

In the latest issue of Education Next, the study’s authors present their findings. Raj Chetty, John Friedman and Jonah Rockoff write that the students of teachers with high value-added scores “are more likely to attend college, attend higher-quality colleges, earn more, live in higher socioeconomic status (SES) neighborhoods, and save more for retirement. They are also less likely to have children during their teenage years.”

The authors analyzed two decades of test-score data on more than 1 million children in grades 4 through 8.

“I think we’ve shown pretty convincingly that there is likely a role for student tests in teacher evaluation,” said Rockoff, a professor at Columbia Business School. “However, nothing we’ve said in our paper says that the role has to be very large.”

Having a teacher with a high value-added score increases a student’s cumulative lifetime earnings by $50,000, according to the researchers. Teen pregnancy rates were also found to be lower, on average, for the students of teachers with high value-added scores.

“We can say that the teachers who tend to raise their students’ test-scores also provide them with something that is going to help them down the road,” said Rockoff.

Though he doesn’t believe value-added scores should be the sole measure of a teacher’s quality, Rockoff said he takes offense at “people who say it should go down to zero.”

The study adds to a growing body of research on the validity and reliability of value-added modeling, as educators and policymakers work to change how teachers are evaluated across the country. Many schools are in the midst of overhauling their evaluation systems as part of the federal School Improvement Grants program.

Rockoff and his colleagues acknowledge some of the potential pitfalls that could come with a greater focus on value-added scores, such as greater teaching to the test and cheating. Rockoff also said that knowing one’s value-added score doesn’t do much to help a given teacher improve his or her teaching.

“One really important reason why value-added can’t be the only component in evaluations is that it provides no feedback to teachers,” Rockoff said. “It’s like telling a baseball player what [his] batting average is. You know you batted .320 last year and that’s good, but it doesn’t tell you what you did that made you do poorly or do well, or what you have to change to get better.”