The Growth in Growth

Prior to the 2005-2006 school year, schools around the country were required to count how many students in each school were "proficient" to determine whether a school made Adequate Yearly Progress (AYP). That year, two states, Tennessee and North Carolina, began the Growth Model Pilot program to allow schools to meet AYP another way, by moving children sufficiently along on a path to proficiency. No more was it just about status; growth and progress mattered too. It was a fairly inauspicious start: in that first year, a grand total of seven schools nationwide made AYP by growth targets alone. But in 2007, 353 did, and last year, 1,571 schools did it. This represents just a fraction of our country's 90,000 schools, but the growth in growth is a positive sign.

The difference between old and new is one of "status" versus "growth." Status models count as making AYP only those students who score above a pre-determined cut-off. Schools have a strong incentive only to worry about those students close to meeting (or close to failing to meet) that bar. These are the "bubble kids," the ones where extra attention may be just enough.

In contrast, the best growth models encourage adults to watch the progres of all students. If a student fails to meet proficiency standards but makes sizeable gains in the process, that student counts towards a school's proficiency. In Tennesee, where they have a particularly rigorous growth model, a student who scores in the "proficiency" range is not counted as such if she has slid backwards to the point where her future proficiency is in doubt**. No more ignoring kids at the poles.

The growth models as currently implemented are by no means perfect. Several states, for example, have implemented a system where students have to cross thresholds into newly created tiers to qualify as making growth. These are more or less arbitrary and not much more than additional layers of status--a student could still meet AYP by making a small jump across a threshold or fail to make AYP despite large gains within a tier. As more states get approved to use growth models--15 will be using them in 2009--further study will be needed on their impacts (see the first program evaluation here). Yet, growth models are already adding needed flexibility to the unpopular No Child Left Behind Act, flexibility that may better identify successful schools that help students below proficient make larger than expected gains.

**To clarify, I meant this sentence to suggest only that Tennessee's growth model was rigorous in this specific context. Whereas North Carolina and other states count all kids above the target plus those on pace to be above it as making AYP, Tennessee adds another wrinkle. It looks like this:

AYP = kids above proficient + kids on pace to be proficient - kids above proficient on pace not to be

This is just one among many trade-offs each growth model makes. North Carolina, for example, chose a relatively simple model that takes the amount a student is behind and divides it by three or four years, depending on the student's grade. If the student made gains higher than that, they count towards AYP for growth. Tennessee, on the other hand, uses multiple regression to predict future scores, a far more complicated and difficult to explain system. These issues are complicated, and the implementation choices states make matter. Look for more from Education Sector in the near future on this very topic.