Teaching to the Test Is Not Teaching

Today, more than ever, teaching and learning are heavily influenced -- if not totally dictated by -- standardized tests. Even in my very highly regarded suburban St. Louis school district, students and teachers are subjected to a "benchmark diagnostic test system" (developed by Tungsten, a division of Edison Schools, Inc.) that provides a web-based diagnostic test that features a series of multiple-choice questions designed to help children practice for the state standardized test and to give teachers feedback on how well children are doing. Seems great in principle, but even the asst. superintendent of curriculum and instruction admits that the curricula and the test preparation efforts are now the same thing. That's troubling. You can only assess that which can be measured, and because you can only assess that which can be measured, you're likely to teach only that which can be assessed.

So it's not merely that I oppose large-scale testing that relies heavily/exclusively on multiple-choice questions. Rather, what I oppose most stringently is the extent to which multiple-choice questions in monthly "practice standardized tests" now completely dominate what happens in the classroom.

The famous Einstein quote is relevant here: "Not everything that counts can be counted, and not everything that can be counted counts."

My concern is that the over-reliance on the statistical and psychometric techniques that buttress the claims that multiple-choice questions actually test analytical and conceptional skills will produce classrooms that are not consistent with a broader vision of teaching and learning. Even in using more valuable kinds of assessment, e.g., classroom-based formative assessment, there's a tension between assessment for learning and assessment of learning for documentation and accountability purposes. In other words, it's hard to care about students when you're so busy writing down observable performance data about them that ties into State Standards CA42.A1, SS16.B12, and M27.J4. Learning vs. proving you have learned are two different objectives. In the former, both the student and the teacher may actually care about the outcome. And they may care less whether it can be quantified and recorded.

But in this new quantifiable game, "proving I have taught well" or "proving I have learned" are euphemistic covers for "please don't fire me" and "please don't fail me" respectively. Under NCLB, even really good assessment practices, when operating under the weight of "accountability," can become about covering one's derriere. Inevitably, and quite logically, students may focus only on those things they can demonstrate they know and that they are good at. Teachers may focus only on those things they can demonstrate they can teach with predictable, positive outcomes. Neither can afford to show process or ambiguity, and certainly neither wants to show a lack of knowledge or competence or even – heaven forbid – that they are wrong about something. Moreover, if the measures they use to capture and record knowledge and performance are biased towards reliability instead of validity, such measures as "process" and "growth" do not even register as possible options.

So what effect might this have on quality, substantive, in-depth teaching and learning? It's not hard to imagine.