Stardate 20010630.1218 (On Screen): There are a lot of problems with multiple-choice computer graded achievement tests, but they have two virtues: they can be graded efficiently, and the grading is objective. It is true that it is possible to learn methods which will boost your score, but on the math SAT, there's still going to be a difference between a student who scores 750 and one who scores 450, and no amount of how-to-take-the-SAT training will alter that.
The biggest flaw of those kinds of tests is that they are not appropriate for testing certain kinds of things, and the ETS is now going to use an essay in some of its tests. I have deep misgivings about this. How do you objectively grade essays? And how do you grade 400,000 essays a year regardless? It can't be automated, and it isn't going to be economically possible for the grader to spend more than five minutes per essay. Also, one grader would give a different score to a given essay than another would, so the result will contain a strong element of chance.
The fundamental problem with the whole concept of continuously testing students for achievement in order to track the effectiveness of schools is that it can't really be done. You can try, but it isn't really possible to test effectively, efficiently and objectively. (You can have any two of those.) (discuss)
