How precise are the tests?

In an earlier post, I expressed my unease at using a single test to make definitive conclusions about either a school or a student. Part of the problem is that individual students' results can vary quite a bit from one test to another. When MPS first introduced annual testing using the Terra Nova, it started with second graders. Wisconsin tested third using its own third grade reading test (called the WRCT). Thus for that first year MPS third graders were given two different reading tests. How well did the tests track each other?

See this figure comparing the third grade reading scores for the two tests. On average, students who did well on one also did better on the other, but the points do not fall on a straight line. For any individual student the story may be quite different: a student may do superbly on one and dismally on the other. Some of the difference may reflect differences in the tests: whether they were timed and differences in the kinds of questions. Some may reflect differences in the student's mood on test day.

Some opponents of testing might argue that that this lack of precision argues for doing away with the tests. I would argue the opposite: that it calls for more testing to better understand the student. The aim, after all, is not high test scores but skilled readers.

To take the medical analogy, if a test for a disease is one hundred percent accurate there is no reason to do additional tests. If however, the test often produces false positives or negatives, additional testing is called for.