New study urges caution in using “value-added” measures

The last five weeks have seen tremendous debate on the topic of teacher effectiveness, spurred in large part by recent events in the public-school systems of Washington, D.C. and Los Angeles.

Michelle Rhee, the D.C. Schools Chancellor, fired hundreds of teachers in July, and some of these educators were let go specifically because they were deemed ineffective. On the West Coast, meanwhile, the Los Angeles Times unleashed a firestorm this month with its series entitled “Grading the Teachers,” one part of which is an attempt to assign every elementary teacher (grades 3-5) in Los Angeles Unified School District a “value-added” score. (Note: A grant from The Hechinger Report helped fund the Times analysis, although we didn’t participate in the analysis.)

The Times explained “value-added” in the following way: “a statistical approach … [that] rates teachers based on their students’ progress on standardized tests from year to year. Each student’s performance is compared with his or her own in past years, which largely controls for outside influences often blamed for academic failure: poverty, prior learning and other factors.”

A new briefing paper, released yesterday by the Economic Policy Institute, calls into question “value-added” measures of teacher effectiveness. Its title, “Problems With the Use of Student Test Scores to Evaluate Teachers,” captures well the skepticism of its distinguished co-authors, who include Linda Darling-Hammond, Helen Ladd, Diane Ravitch, Richard Rothstein and Lorrie Shepard — all major players in the worlds of academia and education policy.

A central question in this debate has been, “Can we currently measure teacher effectiveness in fair, reliable ways?”

No one seems to have wondered whether we do measure teacher effectiveness in fair, reliable ways in the current system; there appears to be universal agreement the answer is “no.”

But can we? Or could we? And, if so, what would such a system look like? How big a role would, and should, value-added models play? These are questions that keep statisticians, psychometricians, economists and others awake late at night. These are questions that matter because the answers to them have the potential to affect, for ill or good, every student and teacher in the country.

But a question of equal or greater importance — and one that’s less frequently asked — is, “What does it mean for a teacher to be effective?”

Colorado took up this critical question in January, when Gov. Bill Ritter created the “Governor’s Council for Educator Effectiveness,” tasking the council with defining teacher and principal “effectiveness” by the end of 2010.

Defining, and then measuring, teacher and principal “effectiveness” is no small task. That’s one reason Ritter gave his council twelve months to do its work.

Educator “effectivness” is hugely important, which is something we learned the hard way from the No Child Left Behind Act. NCLB was a disaster in part because it fixated on teacher “quality” rather than teacher “effectiveness,” which meant the focus was on teacher inputs (degrees, certification) rather than student outputs (learning).

The discussion now seems to be firmly — and rightly — about outputs, not inputs. But which outputs matter? And how can we measure them in fair, reliable ways?

A widely shared view among educators is that a great many outputs matter. And some of the most vital outputs aren’t easily measured. Did a student come to love learning in a certain teacher’s class? Was a student motivated or inspired in important but intangible ways? Did a student gain much-needed self-confidence? Did he or she learn valuable lessons about working collaboratively, or sharing, or having empathy, or putting others before him- or herself?

What we can easily measure is how students perform on standardized tests in English and math. But what do these tests say about how much a student has actually learned? And how much of a student’s performance on such standardized tests is attributable to his or her English or math teacher? How can we fairly and reliably divvy up responsibility for what a student does or doesn’t learn? Presumably, learning is a joint venture — especially as children get older — and responsibility for learning must be shared among the student him- or herself, the student’s parents, the student’s teachers, the student’s administrators and perhaps even the student’s community.

As with so much of what matters most in life, we’re left with more questions than answers. But asking the right questions is essential. Answers don’t matter if we’re fixating on the wrong questions.

Asking how we could or should best define “teacher effectiveness” is of paramount importance. I have yet to see a good, comprehensive definition, but I’ve seen many less-than-good, incomplete definitions.

(Note: This post has been edited since it was first published on August 30, 2010.)