It's one of the most basic questions in education: What defines a successful school?

In Pennsylvania, at least, a new tool called the Pennsylvania Value-Added Assessment System is helping answer that question. The data system, known as PVAAS, tracks student growth instead of the traditional method of using an achievement standard.

While PVAAS has existed in pilot form since 2002, the full complement of data just became available to all school districts last fall.

"When you look at it from the perspective of growth, it's like pulling open a window shade," said Kristen Lewald, statewide project director for PVAAS. "It adds a whole new dimension."

While an achievement standard might measure test scores against an absolute number -- such as whether 63 percent of students are scoring proficient on the state reading exam -- a growth model looks at how much students have improved from one year to the next.

Ms. Lewald said that there are essentially four different types of schools: high achievement and high growth; high achievement and low growth; low achievement and high growth; and low achievement and low growth.

A growth model can help distinguish which low-performing schools are making improvements and which high-performing schools have stopped making progress.

"Achievement data only tells us where we were on a particular day," said Ms. Lewald. "Growth data tells us what direction we're moving in."

The state uses PVAAS data to determine whether schools and districts have acceptable rates of growth, but those findings aren't released publicly. That may change soon, however.

The federal government decided in December to open up its Growth Model Pilot to all states, rather than limiting the number, as it had done since the program started in 2005. The pilot program allows states to use growth data as a way to make adequate yearly progress under the No Child Left Behind law.

Pennsylvania applied last month to join the program, and, though its application has been rejected twice before for insufficient data, the state is now optimistic it will be able to join the program.

The growth model pilot is expected to help give high-poverty schools a chance to demonstrate their progress, even if they aren't meeting the absolute standard.

"There are massive differences in one school to the next in terms of the clientele served," said David Figlio, professor of economics at the University of Florida. "Schools that are bringing students closer and closer to proficiency can be rewarded even if the students aren't reaching proficiency."

States already in the pilot project that have used growth models to determine No Child Left Behind compliance have seen mixed results: A December study by the Editorial Projects in Education Research Center found, for example, that no schools in Alaska that made AYP under a growth model would have missed it under the previous criteria. On the other hand, 14 percent of Florida schools that made AYP made it on the growth model but not on achievement.

If Pennsylvania is approved to join the pilot program, the use of the growth model to meet AYP would likely start this year. And at least some of the data that the state uses to determine the appropriate level of growth would be made public.

With PVAAS, the district can get growth rates for the past year on the district level, the school level and broken down by student groups such as race or economic status. The school can also get predictions of individual student performance, or predictions of the performance of an entire grade, school or district.

Pennsylvania students also have a statewide identification number that allows the state to track growth and make projections even if they move between school districts.

Unlike other states, such as North Carolina, Pennsylvania is not using its growth model data to evaluate individual teachers.

The growth modeldata is based entirely on student scores on the Pennsylvania System of Student Assessment, and thus is only as good as the tests are, and only provides a snapshot of students on the few days of the year when they are taking the PSSAs.

"In order to draw valued conclusions, we need more than a single test," said Linda Hippert, superintendent of the South Fayette Township School District, which has been using PVAAS since its first pilot year in 2002. Currently, the state has no plans to include additional tests.

That said, PVAAS has helped South Fayette, which has some of the highest test scores in the state, make sure that its students -- even those already scoring proficient and advanced -- are still improving from year to year.

"We don't want them to slide and glide," said Mary Ravita, assistant superintendent for curriculum, instruction and assessment. "We want to push and challenge."

The PVAAS data helped the district make some curriculum revisions, said Dr. Ravita, such as reinforcing the Pythagorean theorem in eighth-grade algebra after PVAAS helped the district see that the same group of students who knew the theorem in seventh grade had forgotten it by the eighth-grade PSSAs.

One of the keys to PVAAS, said professor Suzanne Lane of the University of Pittsburgh School of Education, is that school districts like South Fayette actually use the data.

A Rand Corp. study last year found that 28 percent of principals in districts that had PVAAS at the time of their survey weren't even aware of it. Ms. Lewald, PVAAS director, said that the study was done too early in PVAAS's implementation for the data to be useful, and she expects the results to be significantly better when Rand releases another study next year.

Still, Dr. Lane emphasized the importance of continued professional development on PVAAS.

"Probably the hardest thing to deal with is to ensure that the results are being used in a way that's beneficial to the children," she said.