What You Should Think About the New Version of No Child Left Behind

The House Education and Labor Committee released a discussion draft of a new version of the No Child Left Behind Act this week. It's an important milestone, being the first concrete proposal from one of the committee chairmen who will ultimately write the law. Overall, it goes a long way in addressing the single biggest legitimate complaint about the current law--lack of nuance--but in doing so pushes accountability-based school reform into some uncharted and potentially treacherous waters.

The current system is pretty simple. Once a year, you give every student two standardized tests, one in reading and one in math. You establish a target percentage of students who have to pass the test--say, 50%. As the years pass, that target increases until it reaches 100% in 2014. If a school misses the target in a given year--or if a subgroup of students (i.e. poor, minority, special ed) misses the target--then the school is identified as not having made "adequate yearly progress," or AYP. As a school misses AYP for multiple consecutive years, it's subject to increasingly stringent consequences.

This simplicity--and thus, lack of nuance--drives most of the not-insane critiques of NCLB. Because school identification is binary--you either make AYP or you don't--NCLB is bad at distinguishing between massively dysfunctional schools and those that just need fine-tuning. Because AYP is based on the percent of students who achieve a state-defined "proficient" score, NCLB disincents schools from concentrating on students who are far above or below that line. Because NCLB only tests reading and math (science is soon to come) it short-changes history, art, foreign language, music, civics, etc., as well as the teaching of important non-academic skills. Because NCLB is based on a "status" measure--the percent of students proficient on a given day--it doesn't take into account growth in student learning from one year to the next. Because it's based on a single statewide test, local evaluations are ignored. Etc., etc.

The discussion draft tries, somewhat valiantly, to address all of these concerns. There are two levels of insufficient achievement instead of one--"Priority" and "High Priority" schools, the latter identifying more serious failure. There are three levels of performance that contribute directly to acccountability determinations instead of one--"Basic" and "Advanced" now bookend "Proficient"--which states can use to create a blended "performance index" which takes the percent of students reaching all three into account (some states already do this). States can use "growth models" where student achievement is deemed sufficient if they're proficient or their rate of improvement over time puts them on a trajectory to become proficient within three years (some states already do this too). States can also implement systems of "multiple measures" whereby a portion of a schools' rating is based on host of possible measures beyond standardized tests in reading and math--tests in other subjects, like history, civics, and writing; AP and IB test scores, college enrollment rates, dropout rates, etc.

And in the most radical departure from the current law, up to 15 states could participate in a "pilot program" whereby state tests are jettisoned in favor of "locally-developed, classroom-embedded assessments." There's intuitive appeal to better integrating accountability with the tests that teachers themselves create and use in their teaching. But despite loads of mandatory peer reviews, certfications of alignment and rigor, etc. etc., designed to ensure that this process is legit, the idea of local accountability skirts dangerously close to oxymoron--when you ask people if they themselves are doing a good job, they almost always say "yes." The entire two-decades-long push for educational accountabilty rests on the idea that schools peform better when there's some external check on their performance, and even the most ostensibly well-designed local system puts that correct and important idea at risk.

One thing's for sure: the accountability regime envisioned in this law would be exponentially more complicated than what we have now. So complicated, in fact, that very few people will be able to understand it fully. That's not necessarily disqualifying; very few people fully understand how a Boeing 747 works, but they fly in them anyway because they understand that difficult tasks--like flying across the world or holding schools accountable--often require sophisticated systems to work well, and they can see the end results for themselves. But there's a real tradeoff here--it's for hard for parents and educators to develop a strategy for improving on a measure they can't understand.

The draft would also take the nation from having 50 different versions of one accountability system to 50 different accountability systems. Again, not necessarily the end of the world in theory. But states have proven themselves to be wily innovators when it comes to exploiting their flexibility under NCLB to undermine the law's intent, including (but not limited to) excluding large numbers of minority students from subgroup calculations and using unusually large "confidence intervals" to give schools the statistical benefit of the doubt. All these new options create wide vistas of opportunity for that kind of mischief, or much worse, which puts a huge enforcement burden on a U.S. Department of Education that often hasn't been up to the task enforcing the much more prescribed rules we have today.

Some other things of note:

* The draft cracks down on some of the afore-mentioned statistical trickery--no more "99-percent confidence intervals" and excluded sub-groups larger than 30 students (or in same cases 40). That's a good thing.

* The graduation rate requirements are a lot better than the essentially non-existent standards in the current law.

* While the 2014 goal of 100 percent proficiency is nominally preserved, the combination of the "growth model" and "multiple measures" options effectively moves the deadline out closer to the end of the next decade.

* NCLB currently has an "out" at the very end of the accountability system--once a school has missed AYP for six consecutive years, it's got to shut down, fire all the staff, reconstitute as a charter school, or (and I paraphrase only slightly) "anything else you can think of that might work." Unsurprisingly, most states and districts have been choosing the latter option, which would be eliminated for the "High Priority" (i.e. really failing) schools. That's a good thing too.

* As Eduwonk noted, the draft eliminates a major loophole in the "comparability" provisions that dictate how Title I funds are distributed among schools within districts. NCLB currently requires that before federal funds are sent out, state and local funds must first be distributed so that students in high-poverty schools recieve services that are "at least comparable" to those provided to lower-poverty schools. This kind of "supplement not supplant" provision is designed to ensure that poor children get additional resources above and beyond what they normally receive.

But NCLB undermines that goal by requiring that when districts compare school resources, they pretend that all teachers are paid the same in each school. Of course, they're not--high-poverty schools disproportionately hire younger, lower-paid teachers, because they're often the least desireable place to work. As a result, per-student spending can often be thousands of dollars less in high-poverty schools compared to low-poverty schools--a disparity the law currently requires districts to ignore. Teachers unions, wary of anything that would infringe on the ability of their more senior members to teach where they please, will fight this as reauthorization moves forward.

* In requiring states to create "longitudinal data systems" that track student performance from school to school and year to year (you can't have the above-mentioned "growth models" without them), the draft includes a requirement that student data be matched to a "unique statewide teacher identifier." The unions will oppose this too, and here's why: matching student and teacher records is a precondition for the kind of test-score-based "value-added" measures of teacher effectiveness (essentially, evaluating teachers by calculating the improvement on their students' test scores from the beginning of the year to the end) that many reformers would like to use as part of merit pay reforms. No teacher identifier=no value-added measure=no merit pay.

*There's a huge amount of verbiage in the draft that tries to describe how schools should go about improving, via the submission of hyper-detailed improvement plans and the adoption of various improvement strategies that tend to read like a collection of education jargon words strung together in random order. This is unhelpful; such provisions are largely unenforceable and cut against the important principle that the federal government should be defining the parameters of educational success, not the means.

All in all, this draft is a reputable first shot at preserving the underlying principles of NCLB while addressing its most obvious flaws. But a good accountability system is a fragile thing, and making the law more complex also makes it more vulnerable to those who disagree with the principles themselves. It will take a lot of additional hard work to make sure that doesn't come to pass.