We all like great stories. The seeming turnaround of the Atlanta public schools was one of them. Kids appeared to be doing better on standardized tests. Now it's come to light that principals and teachers were changing answers, with analyses of incorrect-to-correct erasures indicating astronomical odds against anything other than cheating going on. Even some Teach for America teachers have been questioned in all this.
Critics of NCLB and testing in general are claiming that this is the inevitable result of high-stakes situations. Hedge fund managers rewarded or fired based on performance may resort to insider trading. Athletes use performance enhancing drugs. I think these explanations go a little too easy on the culpable individuals involved, but I'm not particularly naive about human nature either. In situations where it is easy to cheat and cheating is rewarded, people cheat. By some estimates, about half of sole proprietor income (e.g. the guy who paints your deck) goes unreported in the US (and hence untaxed). Teachers judged on student test performance are subject to the same temptations.
I think there's plenty to criticize about NCLB -- not least of which is taking a focus away from high achievers who will always pass grade level tests, and concentrating teachers' time and attention mostly on children who are right around the edge of passing. Those children deserve attention, sure, but so do kids who are too far below grade level to catch up in the time any one teacher has them, and so do kids who will score in the 99th percentile.
On the other hand, in the absence of any accountability measures, you can wind up with kids graduating from high school who do not know how to read. Children are not necessarily served well by no testing either.
So how does one keep accountability while reforming some of the worst aspects of its unintended consequences? Long-time readers of this blog know I like a "value-added" assessment approach. Create tests that don't ceiling out at grade level (so you can get an accurate picture of gifted kids' needs as well). Then assess again throughout the year and judge teachers and schools on kids' improvement. In a digital era, this shouldn't be too difficult, and could in fact make testing fun. Frequent feedback can help teachers make spot changes (and is also more fair, given how much some students move around). Such assessments would reveal that a teacher who brought a fifth grader up from a 2nd grade level to a 4th grade level was in fact doing an awesome job -- and hopefully wouldn't put pressure on her to change a few more answers to get that kid up to 5th grade level. Such assessments would also show if a school's gifted students were merely treading water. Yes, they keep "passing" grade level tests. But that doesn't mean anything good is going on.