Shared data: A new system could provide advance warning on school dropouts

The state Department of Education's recent announcement that one in four students drops out of school was grim news for Californians but a promising step forward in understanding the scope of a crisis. Instead of making an educated guess as they had been forced to do in the past, education officials were able to calculate the dropout rate by using a new system that tracks students as they move from district to district.

When fully implemented, the California Longitudinal Pupil Achievement Data System can do much more than tell us the bad news after students drop out. It can help educators identify strategies likely to help children before they fail. Now, for example, we don't know which teaching method might best prepare a child for the new eighth-grade algebra requirement because we haven't been able to track students from first-grade arithmetic through middle school math.

A debate in Sacramento about access to the new data will determine whether the tracking system reaches its potential-to help educators close a wide achievement gap and policymakers ensure that billions of dollars in education funding are spent on programs that work. Some want to limit access to government or contractors. But as the new dropout statistics show, California faces urgent education challenges that require resources both inside and outside government. The experience of other states has shown that researchers from universities and think tanks can play a valuable role as independent analysts of taxpayer-funded programs.

Like California, a number of states assign each public school student a unique, non-identifiable identification number. Unlike California, these states have a process to make this student data available to outside researchers. Here's a snapshot of some of their findings:

• Texas eighth-graders' test score gains between 1996 and 2000 were likely the result of test-score inflation, not gains in student learning.
• Some-but not all-changes in teaching practices in low-performing schools in Florida seem to improve achievement.
• North Carolina teachers with higher scores on licensing exams are more effective on average in raising student test scores, particularly in math.

Here in California, individual districts have been collecting longitudinal student data, but their policies for sharing it with researchers vary. The San Diego Unified School District made its data available to Julian Betts, an adjunct fellow at the Public Policy Institute of California and economics professor at the University of California, San Diego, and Andrew Zau, senior statistician at UC San Diego.

Betts and Zau were able to analyze years of student test results, grades, and demographic characteristics. They found that San Diego children likely to fail the state's high school exit exam can be accurately identified as early as fourth grade, and that tutoring currently provided to 12^th-graders is probably too late to improve passing rates. Studies like these-in a single district or another state-can offer valuable lessons. But the only way to understand what works for this state's diverse population is to study California students.

Valid privacy concerns have been raised about sharing student information with outside researchers. It's important to note that the system has already been designed to maintain confidentiality: student identifying numbers do not include names or Social Security numbers. More safeguards could be built in; data made available to legitimate researchers can be stored in a secure location on servers that are not connected to the Internet. An independent review board could be created to ensure that the quality and usefulness of a study-not political partisanship-determine which researchers may use the data.

Californians have a large stake in research conducted independently of government. The tracking system is itself the result in part of outside pressure from groups concerned about the achievement gap. In science, research based on data that cannot be audited and used to replicate an author's findings is considered unacceptable. Taxpayers and parents should insist on the same high standards for California's children.