To read a decent and fair summary of the Vergara v. California trial, please see this piece posted by The Pew Charitable Trust. Also covered on VAMboozled! was a summary, but also the summarized testimonies of Linda Darling-Hammond and David Berliner.
It seems Jesse Rothstein, an Associate Professor of Economics at the University of California Berkeley, also testified for the defense. His research has been critical in terms of the evidence about bias in VAM estimates, and whether the sophisticated controls are indeed sophisticated enough to control for the nonrandom assignment of students to classrooms.
As per Rothstein’s testimony (again summarized by Pew) he cautioned “against placing too much weight on the “value-added models” now being used by many states to try to determine the impact of a teacher on student learning. The models try to tease out how much an individual teacher has contributed to a student’s academic progress.
Rothstein’s research has found that such models can yield very different findings for the same teacher from one year to the next, in part because 25 students or so are not a large enough sample size to create a reliable estimate of a teacher’s teaching ability. Value-added models also assume that students are assigned to teachers randomly, which very often is not the case.
“I think we want to think hard about how to use them,” Rothstein said of value-added models. “What I’m seeing is there are a lot of places that got way out ahead of the evidence and passed policies saying we should fire teachers based on value-added. It’s not as easy as just saying ‘just do this.’”
Rothstein also testified about possible unintended consequences of changes to teacher tenure and dismissal rules, saying it could be harder for school districts to attract and retain good teachers and encourage teaching to the test, for example.”