Standardized tests are optimized for grading.
They are designed to be evaluated quickly, and objectively. Scantrons can be graded mechanically, and these essays described in the article can be graded with no thought and a minimum of judgment. These goals, efficiency and objectivity, impose constraints on how you can test competence, and as far as I can tell those constraints are simply insurmountable.
I have taken one and exactly one category of standardized test that I respected, and that was the AP tests. The essay questions for AP Lit was graded by having three English teachers read the essay and evaluate whether your answer indicated you understood the meaning of the passage, and how it relates to the work as a whole. In other words, it was done the hard way. And it was better than any other standardized test I've ever taken at correlating performance with understanding.
The problem is scale. Scaling is the only advantage of the current form of tests, and that's enough. Any replacement is going to have to address scaling.
(My hare-brained solution: Grading essays is what teachers do over summer vacation. It's enough excess labor to remove that design constraints.)
They are designed to be evaluated quickly, and objectively. Scantrons can be graded mechanically, and these essays described in the article can be graded with no thought and a minimum of judgment. These goals, efficiency and objectivity, impose constraints on how you can test competence, and as far as I can tell those constraints are simply insurmountable.
I have taken one and exactly one category of standardized test that I respected, and that was the AP tests. The essay questions for AP Lit was graded by having three English teachers read the essay and evaluate whether your answer indicated you understood the meaning of the passage, and how it relates to the work as a whole. In other words, it was done the hard way. And it was better than any other standardized test I've ever taken at correlating performance with understanding.
The problem is scale. Scaling is the only advantage of the current form of tests, and that's enough. Any replacement is going to have to address scaling.
(My hare-brained solution: Grading essays is what teachers do over summer vacation. It's enough excess labor to remove that design constraints.)