Mikhail Zinshteyn at The Atlantic: “What if U.S. students took fewer tests that measured their ability to understand academic concepts far more deeply than current tests permit?”
Uh, yeah! What if! That’d be pretty sweet! Pardon me for getting all 2006 on you, but… and a pony. I’m all for calling for better assessments– especially when done in service to testing fewer students and less often, as Zinshteyn is here– but this is functionally the same as saying you want your cellphone to be faster, have more features, and be cheaper. Sounds nice, but the difficult part is actually making it happen. I don’t mean to be harsh to Zinshteyn here. Talk like this is very common in popular education media. It’s reflective of two problems: one, the Big Think phenomenon, where the only kinds of ideas that get bandied about in educational debates are very big ideas that call for all kinds of revolutionary change or dramatic improvement, when the history of education is a history of gradual change, failed revolutions, and incremental improvement. Second, of thinking that our educational problems are problems of will– that we simply have to decide to fix them– rather than of ability, or inability. It’s the Green Lantern Theory of education.
In writing my dissertation on the CLA+, I’m frequently frustrated by those who assume that criticisms of these kinds of assessments are necessarily political or self-interested, rather than practical or empirical. From the most apolitical, ruthlessly empirical standpoint, there are many problems with large-scale assessments. That doesn’t make them not worth attempting. But it does mean that we need to reckon with that difficulty when we propose making sweeping changes to public policy based on them, and it means that we need to make those changes in a way that is minimally invasive. The standardized test movement has done neither of these.
Besides, as I’ve said many times: we already have the tools necessary to dramatically reduce the total testing load on our students. Those tools are called inferential statistics, and they are very powerful indeed. With careful sampling, stratification, and responsible inference, we can understand state and national trends with remarkable accuracy, and use those trends to drive policy responsibly. The NAEP is the single most effective educational assessment we have in this country today. It is not a census measure. Instead, it uses sampling and inference appropriately, and it does so with remarkable explanatory power while placing minimal burdens on students, teachers, and schools. Such assessments are well within our ability to create. Typical reasons to oppose that kind of sampling include a) not understanding the power of inferential statistics or b) because you want to sell pricey tests and test materials for tax dollars, and you want to have as large of a customer base as possible. It’s almost as if the profit motive and what’s best for our schools and students are not well aligned!
Update: The thing about the Atlantic‘s educational coverage is not just that it’s so filled with problems. It’s that they’re always the same problems. And those problems are such massive cliches that I don’t understand how nobody at the magazine ever thinks to say to themselves, “gee, maybe disruptive innovators utilizing dynamism with Web 3.0 by teaching a kid to code through MOOCs taught by thinkfluencers won’t solve all of our problems.” It’s exemplified in that awful Graeme Wood piece on Minerva, which reads like somebody watching Silicon Valley and not getting that it’s satirical: it’s the absolute and utter credulity towards those who use the right buzzwords. These people are selling you something, Atlantic crew. Try a little skepticism. Every once in awhile. Just for fun.