How not to rate a teacher
Standardized test scores don’t tell us much about instructors’ skill.
Wwas the general counsel of the Los Angeles Unified School District, it was extraordinarily difficult to dismiss underperforming teachers who had tenure. One major problem was that we lacked objective measures of teacher effectiveness. So when the 2001 No Child Left Behind Act brought the nation annual standardized testing for math and reading, I applauded.
Congress is now seriously considering new legislation, the Every Child Achieves Act, which would continue the same testing program. But 14 years on, I think that’s a mistake. I believe our exam system is deeply flawed, especially when it comes to teacher evaluation.
First, the results are too variable. Teachers may one year be rated “highly effective” while the next year they are merely “effective” or worse, even though there are no observable changes in their teaching skills or strategies. And researchers have shown that even small variations in the evaluation formulas can produce disparate results. This seemingly uncontrollable variability produces great teacher anxiety that is not worth the damage.
Second, there is reason to doubt the relationship between test scores and an individual teacher’s competence. When we had one teacher per classroom, it might have been fair (or at least fairer) to ascribe poor test results to a particular teacher. Now we have more team teaching, and reading and math specialists who, if they are good, can pull up test scores notwithstanding the primary teacher’s lack of skill.
We also have come to recognize that merely by changing principals, school scores can rise or fall dramatically. This phenomenon suggests that the synergy created by high-quality leadership (or the dysfunction of bad leadership) may have a significant impact on teacher ratings.
Third, we have the vagaries of student class assignment. In places that still track students on ability, some teachers get highachieving kids, and a virtual pass, while others have to sweat every day. Similarly, principals often give newer teachers a disproportionate share of students with special needs or disciplinary problems because older teachers don’t want them.
None of the above even takes into consideration the segregation by race or class of school populations because of the continued (indeed, increasing) segregation of housing patterns. Without some means of modifying expectations to the degree of difficulty associated with teaching a particular group of kids, the results penalize teachers who are willing to work in tougher schools. Some states and districts have tried to make adjustments, but no one has developed a widely accepted formula that educators see as fair.
Fourth, the tests are too narrow in scope. They largely focus on math and reading, which obscures the value of elementary teachers who do a great job teaching history, science or music, and makes evaluation of secondary teachers of non-tested subjects like “apples” in comparison with math and English teacher “oranges.”
Finally, there is the little matter of the “cut score,” or standard for proficiency. At one point, it looked as if Common Core was going to impose a national standard. That did not happen. Instead, over the last decade, states have lowered the cut scores when legislatures want to cover up failing schools, and increased them when they want to show their commitment to more rigorous education. So teacher evaluations are at times as much a statement about politics as teaching ability.
Of course, the tests are problematic for students as well as teachers. Just one example: Standardized tests are generally given in the spring, disrupting weeks of learning.
If Congress acknowledges these issues and decides not to continue the current testing regime, that will leave us with two questions.
One is how to evaluate teachers. Do we need standardized test results to distinguish good teachers from bad? Fair and accurate tests could be helpful, but the answer is “no.” Before standardized tests, some districts had great evaluation and professional development programs that weeded out low performers. Others did not. Adding test data can’t turn weak programs into effective ones, as is reflected by the lack of a significant increase in teacher terminations in most districts in recent years.
The other is whether there is any useful role for standardized tests at all. Civil rights advocates worry that without standardized tests, the troubling disparities in our public education system will sink back into the mists and be hidden from public view. I concur. But we don’t need annual testing to demonstrate the problem. Testing at the end of fourth and eighth grade can meet that need, especially if coupled with college matriculation and dropout rates, SAT or ACT scores, and Advanced Placement or International Baccalaureate program results.
Holding teachers and schools accountable is important, but the means should be accurate and fair. The current standardized test program doesn’t pass muster.
is the author of “Renewal: Remaking America’s Schools for the 21st Century.”