Press Release

Conducting a pilot growth-model program for accountability under the No Child Left Behind Act (NCLB) has the potential to provide valuable information to educators, policymakers, and the public, and it is encouraging that the plans under consideration for this pilot program, by and large, represent serious proposals.  Credit is due to both the U.S. Department of Education, which developed clear standards for the pilot, and the state officials who have been working in good faith to meet those standards.

One indication that the process is being taken seriously is that only eight of the initial 20 submissions by states have proceeded to the peer review process.  Utah, for example, unsuccessfully sought consideration of a plan that would have expected only 75 percent of its students to ever meet standards. The Department was clear from the outset that growth models could not be used to undermine the goal of near-universal proficiency by 2014. Again, the Department gets some credit for rejecting plans that clearly did not comply with the eligibility criteria announced in November 2005.

We are well aware that growth models have been used in the past to avoid responsibility for educating all students to high standards. While the plans under consideration represent a good start, they all have room for improvement, and some pose serious concerns. Accompanying this statement is a chart that tracks the differences across the plans, highlights their strengths, and raises concerns about key components of the plans.

Among the concerns that cut across the plans:

Half of the states — Alaska, Arkansas, North Carolina, and Oregon — propose to hold schools accountable for putting students on a path to proficiency within four years. That’s too long. Think of what’s missed by the fourth-grader who is not expected to achieve proficiency in reading or math until eighth grade. While the point of a growth model is to acknowledge that it takes time to catch up the students who are furthest behind, a four-year proficiency target lets states and districts off the hook for ensuring these children get the help and resources they need.

The situation could be even worse for mobile students within North Carolina or Oregon.  In addition to their four-year timeframes, these states propose to reset the clock every time a student transfers between school districts.

Another critical issue is whether states expect that students will become proficient by a set date, or only expect that students remain “on track” to proficiency. This is left unaddressed in most state plans. Two states, Arkansas and Arizona, address this issue directly but in very different ways.

Arkansas would never hold schools accountable for actually bringing students to proficiency, only for ensuring that they are on track to be proficient four years in the future.  In the state’s own words, if a school does not make adequate growth with a student one year, “it would be inconsistent with the growth model to expect the school that the student attends in the next year to make not only significant growth toward proficiency within four years…but also to compensate for the shortfall in achievement from the prior year.”

This stands in stark contrast to Arizona’’s proposal, which sets a firm expectation of proficiency within three years from the first indication that the student is behind.  If a student does not meet his growth target in one year, the growth target for the next year becomes more aggressive to ensure that he is on track to meet the goal in that third year.  After that catch-up period has elapsed, the system doesn’’t award credit for growth. This system is much fairer to students.

Another distinction among the plans is whether school accountability determinations are based on the growth made by individual students or the average growth made by student groups. This is significant because when the unit of analysis is the group, schools can mask the stagnant or slow growth of some students – and even the decline in academic performance – by getting very high growth for others. North Carolina and Oregon both propose to make determinations based on school and group averages.

In addition to timeframes, it is also important to consider the standards to which students are expected to grow under these proposals. The attached chart compares the proportion of students proficient on the state’s tests to the percentage of students proficient on the National Assessment of Educational Progress, which reveal that some states have standards that are far below the federal standards.

Four states — Alaska, Delaware, Florida, and Oregon — propose to evaluate high schools with growth models – even though their assessments do not go past the 10th grade. As a result, a school gets credit for a 10th-grader who has not yet met the 10th-grade standards but appears to be on track to pass the 10th-grade assessment by the end of his senior year.

Given the national concern about raising achievement in high schools, plans that peg growth expectations to below-grade level standards are a bad idea. It would be better simply to exclude high schools from growth-based accountability unless there are annual assessments in every grade of high school.

Finally, states offer very little information on expected impact of growth models on accountability determinations. Only three states — Florida, North Carolina, and Tennessee — provide any information along these lines. These states estimate that a relatively small number of schools that did not make proficiency targets under NCLB would meet their accountability goals under the proposed growth models.

While useful, the limited projection data in these states are not nearly enough.  The Department should require both that states answer more questions themselves and that they make their data available to third-party researchers for thorough evaluation. An independent evaluator should be hired to track implementation of the pilot so that lessons can be learned and disseminated in a credible and efficient process.

It’s important to keep in mind that the goal of the growth-model pilot is to investigate the best ways to refine accountability systems. All of this is a paper exercise unless it is used to identify the students and schools most in need of help and ensure they get the assistance and resources they need. In the end, we hope that states that are allowed to use growth models also will serve as pioneers in providing better interventions and school improvement plans.

Click here for the grid on growth models

Click here for the Ed Trust November 2005 statement on the pilot growth-model program




Note: Education Trust Director Kati Haycock is serving as an official peer reviewer for the growth-model pilot program. She did not participate in the development of this statement, nor did she review it before its release.