Graduation Statistics: Caveat Emptor
Lack of candor about our educational problems produces the most serious problem of all: the minority students whose shockingly low graduation rates we are failing to admit, let alone address.
As the pundits have recently reminded us, honesty is an early casualty of war. Alas, it also appears to be a frequent casualty of K-12 education data. Graduation statistics reported by federal, state, and local school districts are especially confusing, misleading, and implausibly optimistic.
Even the federal government's normally reliable National Center for Education Statistics is wont to inflict such casualties. That agency recently issued its annual report on dropouts and high school completion rates, and again we find that it paints a blurred and falsely cheerful picture of how U.S. schools are performing. According to "Dropout Rates in the United States: 2000," 86.5 percent of young Americans are completing high school, up from 85.9 percent the previous year.
Would that it were so. By my calculations, however, U.S. rates of high school graduation are considerably lower. Using a transparent and easily checked method of comparing 8th grade enrollments in 1993-94 with high school diploma counts in spring 1998 (and adjusting for student-population changes), I put the national graduation rate at about 74 percent. The situation for minority students is far bleaker. I find that only 56 percent of African- American students and 54 percent of Latino students graduated from high school in 1998. This contrasts with claims by the NCES of a 83.7 percent completion rate for African-American students and a 64.1 percent rate for Latino students.
Why are the NCES numbers so much higher than mine? The main reason is that the federal report includes among high school completers those who have passed an "equivalency" test, such as the General Educational Development exam. This is misleading for several reasons, beginning with the fact that recipients of a GED credential are not graduates of the public high school system. They have received their diplomas with the assistance of community colleges, the prison system, or vocational schools, or through their own, independent efforts. Since most people view the NCES completion data as attesting to the performance of our high school system, we should not include people who dropped out of that system and later received diplomas from some other system. (In computing a physician's cure rate, we would not want to count patients who had transferred to the care of another doctor.)
A considerable body of research also suggests that the life outcomes of GED recipients are more like those of dropouts than those of regular high school graduates. Economists James Heckman and Stephen Cameron find that "exam-certified high school equivalents are statistically indistinguishable from high school dropouts." Other researchers find moderate benefits for GED recipients, but no one claims that they are truly "equivalent" to regular high school graduates. To lump GED holders with regular graduates, therefore, is to combine fundamentally dissimilar groups while misleading people about the school system's ability to produce high school graduates.
In previous years, NCES reports included data (in small print) that distinguished regular-diploma recipients from those with "equivalency" certificates. The latest report, however, does not provide this information at all. But we can compute the difference using the 1999 report and removing GED recipients. We find that the NCES calculates the "true" high school graduation rate at 76.8 percent, a little higher than my (1998) finding of 74 percent. The Latino rate drops to 54.9 percent, close to my finding of 54 percent. The education statistics center's African-American graduation rate, sans GED holders, however, was 72.9 percent for 1999, considerably higher than my (1998) estimate of 56 percent.
That discrepancy probably has to do with the different methods by which the information was collected. My method simply involves a comparison of 8th grade enrollments with diploma counts five years later, adjusting for enrollment and population changes. The NCES figures come from the Current Population Survey of the U.S. Census. The survey, known as CPS, tries to phone a representative sample of households to ask a number of questions, including whether young people in the house have finished high school.
While the CPS is a well-run survey, this methodology may inflate the graduation-rate estimate for young African-Americans. For example, the survey excludes prison populations. To the extent that young black dropouts are disproportionately likely to be incarcerated, the CPS will overstate African-American graduation rates. Second, households containing minority dropouts are particularly difficult for phone surveys to reach. Deep in Appendix D of the new NCES report, we learn that this "coverage bias" could overstate black graduation rates by 9 percent to 18 percent. Third, the Current Population Survey depends on people forthrightly describing their own levels of educational attainment. Such self-reporting could significantly distort graduation rates for groups that may overstate their educational achievement to compensate for workforce discrimination. One of the advantages of my simple method of calculating graduate rates is that it requires only enrollment and diploma counts, which tend to be reliable and easily checked figures.
Yet for all their flaws, the NCES graduation statistics are light years ahead of the numbers usually reported by states and school districts. Rather than relying on surveys or enrollment counts, most states and school districts calculate graduation statistics by trying to track individual students over time. Few, however, have either the resources or incentives to track individual students successfully. Faced with ambiguous or missing information about the whereabouts of individuals, districts may be prone to provide the most benign explanation for a student's absence. This tendency to produce rosy results is exacerbated by the bizarre definitions of dropouts that many districts use, commonly excluding not only GED-seekers but also students who leave high school to go to jail or join the military.
While it sounds more precise to track individual students, the failure of districts to do so successfully and the inability of outsiders to check district accounts of student whereabouts can lead to graduation statistics that are grossly in error. Take, for example, the Dallas school district, which reports an annual dropout rate of 1.3 percent. Presenting dropout rates in annual terms is like reporting credit card interest rates in monthly terms; it just makes the number feel smaller. If we convert the annual rate into a cumulative rate (which is how everyone thinks about dropouts), we would expect about 8 percent of an 8th grade class to drop out before graduation. Yet, according to my calculations, only 52 percent of 8th grade students in Dallas manage to earn diplomas. The 1.3 percent rate reported by the district has to be a fantasy in a district with half as many graduates as 8th graders and a growing student population.
Not to pick on Dallas; the sad fact is that the misreporting of dropout statistics is all too common across the country. And at least Dallas is aware that its numbers are off and that the district truly has a serious problem. As a spokesman for the district told the Dallas Morning News, "[Superintendent Mike] Moses has said the dropout problem is probably a lot bigger than what any of the other figures report. We know it's a major problem, and we're trying to do something about it."
Most school officials are not so candid. This lack of candor about the extent of problems in U.S. education produces the most serious casualty of all: the minority students whose shockingly low graduation rates we are failing to admit, let alone address. If people only realized that we graduate barely more than half our minority students, there would be demands for dramatic efforts to remedy the situation. Instead, the problems are being glossed over and the failures of the status quo maintained.
Jay P. Greene is a senior fellow at the Manhattan Institute for Policy Research in New York City. He lives in Weston, Fla.
Vol. 21, Issue 18, Pages 37, 52Published in Print: January 16, 2002, as Graduation Statistics: Caveat Emptor