Our district has changed its approach to placing students in grades 6 and 7 math classes. Students considered for placement above grade level must now take a test composed of problems drawn from the next-level CMP textbook: for instance, in the spring of grade 5 they take a test based on problems from the CMP 6 textbook. Students must score 95% or better on this test to be placed above grade level; otherwise they will be in CMP 6 in 6th grade. Teacher input and scores from standardized tests are not considered. I cannot find data supporting the use of this method. Do you have any comments on the usefulness of this approach to placement?
There are many flaws in this system, but there aren't many other ways.
As @celeriko pointed out, testing on material that has not been taught yet is dangerous. It can result in not promoting a fast learner who is truly "gifted", but rather promoting a kid who worked for hours the week before the test on the next year's textbook, on the way attaining many holes in their math and hurting them in the long run.
On the flip side, if a student is truly talented and learns fast, they should have taken initiative, possibly through their parents, to explore higher math by now and should be comfortable with the next year's concepts.
Your concerns are valid, and they should definitely be also integrating teacher input and grade on the current class to ensure social maturity, capability of learning, and consistent performance. In summary, the test itself is fairly useless without external validation, at which time it becomes useful enough to make the decision.
In America at least, there is a strong emphasis on "social" promotion, and keeping children with their "peer group."
Thus, they won't allow promotion to the "second" level above the current one, unless a student can get a 95% on a test before going into the "next" one. This is a difficult (but not insurmountable) hurdle.
Basically, the American grade school systems want to make special provision only for students that are clearly "gifted," not merely "good."