Sportive Thoughts

While I am a broad supporter of standards based testing, I do have a few worries.

First, of course, are the questions which test and what standards? The difficulty of inventing a so-called "perfect" standardized test is one that requires many expensive investments and incentives. The best standardized tests--such as the SAT and ACT (for colleges), GRE (for graduate schools), MCAT (medical), LSAT (law), and GMAT (business)--are all proprietary exams. They are expensive to create, distribute, and evaluate, and they all require a hefty fee from students not only to take the exams, but also to distribute test scores to potential schools. The GRE, for instance, is computer adaptive: the test program selects questions from a bank of presorted samples; each question matches the student's current progress according to a statistical algorithm. The final score depends not only on how many questions the student answers correctly, but also the difficulty of each question AND the relative success of other test takers that same session. It is an enormously complex process, as well as enormously expensive.

When states turn to cheaper tests--usually produced with government subsidies or contracts--they do so to lower the tab for both students and the district. These cheaper tests are often less predictive of academic merit; they are limited to more basic subject matter and have less robust statistical sampling. As a result, it is very difficult to adjust standards up or down; if a state commissions a test, they must limit the relative difficulty of questions to a very specific range--otherwise the results will be either meaningless (too easy)  or politically embarrassing (too hard).

Hence, the tests given to 4th, 8th, and 10th grade students aim towards the middle 50%. This means two things: (1) if your child is in the bottom 25% of his peers (a "D" range), the primary goal of his teachers will be to bring him up to the next quartile (a "C" range). (2) if your child is in the top 25% of his peers (an "A" range), his teachers will not face direct pressure from the district until he falls two quartiles (the "C" range). Teachers will have not only an incentive to teach the test, but an incentive to teach mediocre scores rather than high scores.

This is all solved by a better test, but a better test is more expensive. State budgets are already restricted.

Secondly if teacher pay and teacher tenure is dependent upon student test scores, why would a new teacher select to teach in a "failing" school? The scenario is like this: a new teacher is trying to decide between job offers between school A and school B. School A has historically high-performing students, therefore new teachers reach their instructional targets sooner; school B has historically low-performing students, therefore new teachers reach their instructional targets later. The first choice will clearly be school A, since there is a stronger possibility of success and higher pay. School B is a riskier choice--perhaps a challenge for the ambitious education school graduate--but we can expect few people to volunteer for the risk. The result will likely show new teachers competing for spots at the better school. Risk will be portioned out to weaker teachers--exactly the opposite of what we want.

(One might easily counter that we don't want weaker teachers at all--but that's a different issue.)