One thing I'm getting at here is that even if we discount things like demographic shifts, there is the factor that the kids are one year older and are getting the harder tests. The new grade 3 kids being tested are last year's grade 2; the 4's are last years 3rd graders and so on. If you don't teach the kids anything all year, push them onto the next grade and then test them at the next level, they will do worse. Based on this idea, a year-over-year performance should be able to detect a real, instantaneous problem.