California's new tracking system represents a giant step forward in measuring the statewide dropout rate. But it is still an imperfect metric of what is widely considered the most sensitive indicator of school success.

To understand the reasons for this guarded appraisal, it's necessary to peel back the layers of statistical legerdemain that for too long have characterized the way data are collected and reported.

For most taxpayers, determining the rate at which students drop out is a straightforward matter. Quantify the number of students who enter high school as freshmen and then repeat the process four years later. Compare the two for the answer.

In fact, that's essentially the way the rate was ascertained.

But this simplistic method was always grossly misleading. For one thing, it counted as dropouts those students who moved out of the district, even though many may be attending school in another city. The growth of the school choice movement, in particular, has accelerated this trend. So although students didn't graduate from the same school in which they were initially enrolled, they hardly warranted the scarlet letter of dropout.

For another, the former metric didn't factor in students who earned their high school diploma in five years, instead of the customary four. The number of these students is increasing because social promotion is being phased out. Students must now demonstrate mastery of the knowledge and skills of each course they are taking. As a result, relying strictly on a four-year timetable skews the data.

To sweep away the confusion, California's new $33 million system, which began in fall 2006, issues an identification number to each student. It allows the Department of Education to follow students longitudinally. Although it is a marked improvement over the past, it still fails to present a totally accurate accounting.

One of the reasons is the existence of what researchers call pushouts. With pressure mounting to boost standardized test scores under the No Child Left Behind Act, principals have been given a strong incentive to encourage students who are likely to pull down their school's performance to enroll in alternative schools.

The trouble is that graduation rates at these schools are appalling. According to the National Center for Educational Statistics, students from the lowest-income quarter are more than six times as likely not to graduate as students from the highest quarter. Since poverty and performance are closely linked, school officials see poor students as the greatest potential liabilities on testing day.

This practice, however, is not limited to low-income students. According to Jobs for the Future, a nonpartisan research organization, 10 percent of students from families in the nation's highest socioeconomic status level also drop out.

While it's impossible to know exactly how many of these were pushed out, they too could have been perceived as a potential drag on the overall performance of their schools and therefore subject to similar prodding to leave school before testing day.

The credibility of reported dropout rates was further undercut by what is called credit recovery. This policy allows students who lack sufficient credits for graduation to make them up by means other than retaking a class or attending traditional summer school.

In an attempt to boost their graduation rates, principals in some high schools have helped struggling students make up missed work through such creative means as online courses. According to the National Dropout Prevention Center at Clemson University, roughly a third of states have either developed or endorsed this approach.

Although most states require students to earn credits for graduation by completing a stipulated number of hours of seat time and demonstrating subject competence, the practice of credit recovery is becoming more common. And so is criticism that credit recovery is casting aspersions on gains in graduation rates by resorting to gaming the system.

To restore confidence in what has become a crazy quilt of data regarding dropout and graduation rates, the U.S. Department of Education has stepped in to require all states to use one federal formula for the nation's 14,000 public high schools. The aim is to bring transparency to what has previously been an opaque system.

It's a long overdue step that couldn't come at a more crucial time in educational history.

Walt Gardner taught for 28 years in the Los Angeles Unified School District and was a lecturer in the UCLA Graduate School of Education. Reach him at walt.gard@verizon.net.