Alfie Kohn's take on the failure of "value-added" teacher evaluations.

today's Huffington Post contained an essay from education expert Alfie Kohn, offering some good points on the dialog about measuring value-added data for teachers, "What Passes for School Reform: 'Value-Added' Teacher Evaluation and Other Absurdities." If you're interested in this nationwide discussion of how to improve teaching and learning by ensuring effective teachers, I recommend you read it.

The most interesting section (in my opinion) is this (reprinted):

What if we asked other questions instead? We could do so about any of the policies I've mentioned, but for now let's consider the idea of judging teachers with a "value-added" method.

Question 1: Does this model provide valid and reliable information about teachers (and schools)? Most experts in the field of educational assessment say, Good heavens, no. This year's sterling teacher may well look like crud next year, and vice versa. Too many variables affect a cohort's test scores; statistically speaking, we just can't credit or blame any individual teacher.

Unfortunately, many of the experts who point this out tend to stop there, even though the problem runs far deeper than technical psychometric flaws with the technique. For example. . .

Question 2: Does learning really lend itself to any kind of "value-added" approach? It does only if it's conceived as an assembly line process in which children are filled up with facts and skills at each station along a conveyor belt, and we need only insert a dipstick before and after they arrive at a given station (say, fourth grade), measure the pre/post difference, and judge the worker at that station accordingly. The very idea of "value-added measures," not just a specific formula for calculating them, implicitly accepts this absurd model.

Question 3: Do standardized tests assess what matters most about teaching and learning? If not, then no value-added approach based on those tests makes any sense. As I've argued elsewhere -- and of course I'm hardly alone in doing so -- test results primarily tell us two things: the socioeconomic status of the students being tested and the amount of time devoted to preparing students for a particular test.

Regarding individual students, at least three studies have found a statistically significant positive relationship between high scores on standardized tests and a relatively shallow approach to learning. Regarding individual teachers, let's just say that some of the best the field has to offer do not necessarily raise their kids' test scores (because they're too busy helping the kids to become enthusiastic and proficient thinkers, which is not what the tests measure), while some teachers who are very successful at raising test scores are not much good at anything else. Finally, regarding whole schools, if test scores rise enough, and for long enough, to suggest a trend rather than a fluke, the rational response from a local parent would be, "Uh-oh. What was sacrificed from our children's education in order to make that happen?"

It won't do to fall back on the tired slogan that test scores may not be perfect, but they're good enough. The more you examine the construction of these exams, the more likely you are to conclude that they do not add any useful information to what can be learned from other, more authentic forms of assessment. In fact, they actively detract from our understanding about learning (and teaching) because their results are so misleading.

Notice, by the way, that everyone who declares that we ought to reward good teachers and boot the bad ones is assuming that all of us agree on what "good" and "bad" mean. But do we? I'd argue that a dipstick, test-based model is endorsed by newspapers, by public officials, and by billionaires who have bought their seat at the policy-making table (seat, hell; they own the table itself) precisely because we often don't agree.

More good points to ponder...