Many aspects of American public education that attract attention for one reason or another, no matter what their seeming subject, turn out at some level to be about the American Dilemma: race. Take, for example, the current set of discussions concerning how to calculate high school graduation rates.
One way in which the quality of schools and school systems can be compared with that of others is the rate at which students move through high school from grade 9 to graduation. This is a rather crude measure—it says little or nothing about what those students have learned—but it is useful in a sheep-and-goats way: Districts that graduate very few of their students in four years are unlikely to perform well in other ways.
The point of such a measure is comparability. We will not find it a useful comparison if some districts report dogs and goats and others report sheep and cats. High school graduation rates that include General Educational Development, or GED, credentials, for example, might be interesting, but they are not useful comparisons with those that do not.
There is now a consensus that the ideal method is to trace individual students from grade 9 through to graduation. As this system will take some time to put into place, various approximations are under consideration. The most straightforward way of approximating graduation rates is to count the students entering grade 9 one year, count those receiving diplomas four years later, and divide the former into the latter. This is, after all, what most people mean by the term “graduation rate.”
The National Center for Education Statistics recently published a study of some methods of approximating graduation rates in lieu of actual individual student data. Noting that in the aggregate there is a “bulge” in enrollments at grade 9, the NCES recommends estimating a base cohort by averaging grade 8 through grade 10 enrollments.
The grade 9 bulge is thought to be caused by the common practice of restricting promotion from grade 9 to grade 10 by an examination. In some places, this gateway is placed at grade 8, rather than at grade 9, creating a grade 8 bulge. In other places, there is no gateway. But, looking at national data, the grade 9 bulge is quite apparent and, by increasing the size of grade 9 enrollments over those of grades 8 and 10, seems to require adjustment.
Looked at more carefully, the matter is more than a problem in statistical methodology.
When data are disaggregated by race/ethnicity and gender, which was not done by the NCES study group, we find that states with large numbers of black students have large grade 9 enrollment bulges; those with smaller numbers of black students have smaller grade 9 enrollment bulges. According to NCES data for the country as a whole, nearly a third of black male students repeated grade 9, while 15 percent of white male students were forced to repeat grade 9. In many states, more than 30 percent of black male students were forced to repeat grade 9 in 2004-05 (and in New York, where 5 percent of the white male students were held back, the figure for male black students was 44 percent).
More research is needed (as researchers always say), but it does appear that the grade 9 “bulge” is largely attributable to the failure to educate black students (and, to a lesser extent, Hispanic students). These deleterious effects are concentrated in states where black students are concentrated. (Where there are lower concentrations of black students, black graduation rates are higher.) “Smoothing the bulge” statistically, as advocated by the NCES and others, conceals this effect and therefore makes estimated cohort graduation rates less useful as an analytical tool.
This is a matter of concern on both ends of the causation arrow. Why, in these states, are black boys so ill-prepared for high school that a third or more must repeat grade 9? And, given the dropout rates in these states for black male students, is this policy of erecting a grade 9 gateway useful educationally, or simply effective as a way to push male black students out of school?
