Update: Kathleen Porter-Magee looks at the Chicago school board's proposal for how to evaluate teachers. It includes standardized test scores in reading and math; students' scores on standardized performance tasks (ex; "Draw a table using two-point perspective;" "Recite a poem") created by a district-wide committee of teachers; classroom observations; and student surveys for teachers in grades 4-12. Documents released by the school board suggest teachers in untested grades and subjects will be judged in part on school-wide performance metrics. For example, a kindergarten teacher in a K-5 school would be evaluated on how well the school's fourth and fifth graders perform on standardized reading and math tests, the logic being that she helped prepare those students a few years earlier. This plan looks like a decent start, though it's important that classroom teachers retain the autonomy to create their own units, lessons, and assessments.
The Chicago teachers' strike has prompted a lot of commentary along the lines of — "Duh, judging teachers by how much students learn is a great idea! I get evaluated at my job!" — and while I agree that more stringent, detailed teacher evaluations are important, measuring how much students learn over time, and how much of their knowledge can be attributed to their classroom teacher (as opposed to their parents, tutors, or peers), is a complex proposition. I'd like to review what some of the challenges are.
Historically, when adult incentives are tied to children's test scores, the curriculum is narrowed to what is included on the test. The must-read book is Measuring Up, by Harvard psychometrician Daniel Koretz. No test can encompass the broad sweep of what we want children to know, not only in a given subject, but also in terms of critical thinking, decent behavior, curiousity, and so on and so forth.
That said, teaching to a good test–like an International Baccalaureate or Advanced Placement exam, with free-response questions and performance tasks–will lead to better instruction than teaching to a mediocre mulitiple choice test. Currently, however, most states' standardized exams remain pretty bad. Many states don't test writing or science at all, and hold kids to ridiculously low standards in math and reading. A new generation of tests, tied to the more rigorous, nationally-shared Common Core curriculum standards, has not yet rolled out, and we still don't know how far those tests will move beyond the multiple-choice paradigm.
But even if we adopt a wonderfully rich, broad new set of student tests, not all subjects and grade levels are or will be covered by standardized assessments. I've written about the challenges of assessing the youngest children, as well as those enrolled in art, music, and physical education classes. Should there be state testing days in kindergarten? In gym? In some states and districts, it's already happening — but you better believe it's controversial.
Not everything in education needs to be centralized and standardized. Teachers should always be able to create their own tests of student learning. For example, if a ninth grade French teacher chooses to do a special unit on Impressionism, she should be able to test her students on vocabularly related to painting and nineteenth century history, or instead assign an oral presentation in which each student talks about a painting of their choice. And this teacher should be rewarded if her students do well! This can coexist with state or district-level standardized testing, but if too much teacher evaluation weight is given to assessments out of the teacher's control, she will have to cede much of her creativity in choosing what to teach. That can take some of the joy out of teaching and learning.
Tests are not the only way to judge effective teaching. My colleagues at the New America Foundation released a fantastic paper last year about models for more rigorous classroom observations of teaching practice. In this article I visit a school in Denver that uses teacher peer review, and compare it to a Colorado Springs district that evaluates all teachers according to pencil-and-paper test scores, as well as through classroom observation.
The bottom line is that there is no one, agreed-upon way to measure student learning or to tie that learning to teacher performance. It isn't fair to claim that teachers' unions oppose evaluating teachers; what they do tend to oppose is evaluating teachers according to standardized test scores. In my view, a good evaluation system will include many different kinds of evidence, from standardized test scores, to scores on tests created by the teacher, to portofolios of student work, to classroom observations. This is a delicate balancing act. Weighing any one of these data sources too heavily has the real potential to distort students' learning experiences, which is why elite private schools tend to downplay the importance of tests.