The increasing complexity of No Child Left Behind.

It was supposed to be simple. Former Secretary of Education Rod Paige said it clearly when discussing No Child Left Behind in an August 2004 speech: "Its goal is simple: All students read and do math at grade level."

Is it that simple? More than four years after the passage of NCLB, implementation is more complicated than ever. An apt metaphor might be taxes. The law set out to create the equivalent of the flat tax proposals with one simple form. Instead we have the 1040 form, and it's not the EZ. To find out whether a school is achieving under NCLB, parents and policymakers alike must sort through multiple iterations of data, if/then scenarios and much more.

Multiple Layers

For example, many states use a confidence interval to increase the reliability and validity of the results of their state assessments. It has the effect of creating a margin of error around test scores so the score on paper may not be the score figured into adequate yearly progress.

Further, many states now have a confidence interval around their safe harbor calculation. This means states must calculate whether there was a 10 percent decrease in the number of students performing at the lowest level, plus or minus a certain percentage. It's not so simple.

Subgroup sizes have gotten more complicated as well. In early implementation, most states used a group size of 30 or 40 students to ensure the data would be valid and reliable. Recently, an increasing number of states have begun using a minimum subgroup size proportional to the overall student population. For example, instead of 30, a state's subgroup size might be 30, or 15 percent of the student population in question, whichever is greater, with a cap of 75. That's definitely not simple.

What about special education? First, there are the alternate assessments for students with the most severe cognitive disabilities. There is a cap on the number of proficient and advanced scores on those assessments equal to one percent of the student population. If more students take the alternate assessment and pass, some of those scores have to be changed--for the purposes of AYP only--to not passing.

Then there's the alternate assessment to modified standards for students with persistent academic difficulties--students who cannot achieve grade-level proficiency even with the best interventions. For them, there is a cap on the number of proficient and advanced scores equivalent to 2 percent of the student population.

The bottom line? To determine whether the special education group of students made adequate yearly progress, states need to develop two new assessments. It must be determined which students can and should take which test within the limitations of the cap, and AYP must be calculated using those results. While this policy results in more appropriate options for students with disabilities, it is also incredibly complicated.

Further, for this year only, because no state currently has modified standards and assessments, states can use something called a "proxy" that is yet another percentage added onto the number of students scoring at the proficient level. While better for children than the previous policy, can it get more complicated than this?

On the issue of graduation rates, it's not just a matter of looking at four-year intervals. Many states now allow extended timelines to students with disabilities of five years and longer.

Schools in several states also can count retesting when students have multiple opportunities to take an assessment, and students with limited English proficiency now are counted in that subgroup even after they have "tested out" and are taking regular classes.

These are two worthwhile ideas that are advantageous to schools and students, but they add yet another layer of complexity to the calculation of AYE

Necessary Tweaks

All of these modifications to the federal law are intended to make the system more fair and accurate and to help reduce the chance of over-identifying schools for not making progress. But this means the AYP system is increasingly complicated, especially for parents for whom it was supposed to be easy to understand.

This exposes a fatal flaw in NCLB--the considerable challenge of obtaining valid and reliable results through the sundry types of tests and the "status model" in use, a system that compares groups of children to past performance.

This is why we're seeing increasing interest in Washington in the notion of a growth model. Growth models, which are in place in several states, look at the change in a school's performance over time. While these models have great value, they are not acceptable currently to the U.S. Department of Education because they do not require all students to be at grade level by 2014.

One possible alternative is a model that considers individual student growth and school performance, similar to the process used by 1,900 school districts that work with the Northwest Evaluation Association. The Department of Education has had one task force on growth models and is currently reviewing proposals from states to develop them. The department also gave $52.8 million in grant money to 14 states for the design and implementation of statewide longitudinal data systems for the purpose of growth models.

Growth models are certainly not simple, but they would add value in terms of meeting the goal of NCLB by establishing a clearer way to gauge a school's performance. Watch for growth model discussions to permeate the upcoming reauthorization of the Elementary and Secondary Education Act in the next few years.

Terri Duggan Schwartzbeck is a policy analyst at AASA. E-mail: [e-mail]