Under the federal No Child Left Behind law, states are required to submit student data, disaggregated by a list of federally defined groups, to the U.S. Department of Education. Fueled by more than $500 million in grants from the Obama administration, all 50 states and the District of Columbia now have data warehouses that allow them to track each student's academic growth from the time he or she enrolls in school.

Collecting data is one thing, but there's growing evidence that school districts often aren't using it to improve student achievement.

A new study from the Data Quality Campaign finds that most states aren't sharing what they collect with parents, policymakers, principals and, most importantly, teachers. In many cases, the data could provide information that would allow teachers to focus instruction directly on individual students' needs.

Teacher quality is the school-based variable (outside factors such as parental achievement levels, income and race are most important) that has the greatest impact on student performance, and so-called value-added assessment data is the tool that allows us to measure the comparative effectiveness of different teachers over time.

Since 2005, Massachusetts students have been the nation's best performers on national tests. By any measure, the Bay State's tests, which were first administered in 1998, have produced a wealth of data. But even there, the record is spotty when it comes to using the data to improve learning.

Massachusetts' Office of Educational Quality and Accountability developed specific indicators that measured school districts' use of disaggregated student assessment data to inform programs and services that affect the quality of teaching and learning. In 2006, the Pioneer Institute (a think tank I'm affiliated with as a senior fellow) gathered data from EQA evaluations of 76 Massachusetts school districts. During 2005, 71 percent of the districts reviewed received a grade of "Below Satisfactory" for student assessment and evaluation.

Commenting on the Bay State's immense data warehouse, one school-district official made the point that districts "need 10 or 15 defined pieces of data to help kids improve, not 100."

In addition to being more selective about data collection, the Pioneer Institute and Data Quality Campaign studies make other promising recommendations. The Pioneer study calls for Massachusetts to require school districts to indicate the process they will use to analyze data in a way that informs instruction. It also recommends that districts earmark part of their state education aid to hire a data specialist to develop data-driven school-reform models that teachers can use in the classroom. In an effort to improve teacher-preparation programs, the Data Quality Campaign calls on districts to share data on teachers' impacts on student achievement with the institutions that prepared each teacher.

Student data is only as good as school districts' ability to use it to boost performance. To achieve that goal, states may have to be more targeted about the data they collect, or at least what they share. But now that the data is there, it will certainly require shifting the focus from collection to how to apply what has been collected.