When the Census Bureau is wrong
July 13, 2005
What does it take to kill a damaging and misleading falsehood? For years, respectable researchers and advocacy groups from left and right have been trying to quash and correct the misleading high-school graduation rate figure put out annually by the Census Bureau's Current Population Survey (CPS). Yet the exercise feels like shooting at a cloud of dust; no matter how many times you hit it, the thing just keeps rolling on. You might suppose that embarrassment, if not professional research standards, would be enough to get the Census Bureau to stop publishing it, or at least express doubts about its accuracy. Yet there it was again the other day, a new ridiculous figure based on 2003 data to replace its ridiculous 2002 figure.
The new bogus figures claim that 85 percent of all U.S. adults (those over 24) and 80 percent of black adults have completed high school. These numbers are hailed as all-time highs by the Census Bureau's press release. For those more interested in the current graduation rate, the Census Bureau claims that a full 87 percent of young adults (ages 25-29) have completed high school. You can read the report for yourself by clicking here.
This would be fantastic news if it were true, but it isn't. Since many news outlets report these figures as actual graduation rates because of the credibility that always attends Census Bureau numbers, it's important to once again explain why they can't be believed.
First of all, the CPS counts those who receive a high-school equivalency certificate (like the GED) as identical to regular high school graduates. This badly inflates the graduation rate, since earning such a certificate is decidedly not "equivalent" to graduating from high school. Indeed, research shows that life outcomes for GED recipients are far closer to those of high school dropouts without GEDs than to those of true high school graduates.
The underlying problem here is that CPS's survey question is too vague. It asks, "What is the highest grade of school [this person] has completed, or the highest degree [this person] has received?" Since the question does not ask the respondent to distinguish between a GED and a real diploma, CPS has no choice but to report all high school "completers" in the same category. Everyone from the Business Roundtable to the Urban Institute has denounced this practice. Yet the CPS plugs on, year after year, asking the same defective survey question, oblivious to the urgent national need for accurate high-school graduation rate figures.
Second, the CPS does not include in its survey America's sizeable population of people living in institutional settings. This means, among other things, that approximately 1.3 million prison inmates are not included. Since it is well established that high school dropouts are over represented in prison, CPS thus overestimates the high school graduation rate.
Third is the inherent problem of self-reporting. A survey asking about educational outcomes of respondents and their family members is vulnerable to inaccurate reporting, given the stigma associated with dropping out of high school. Respondents may also be confused due to the vagueness of the CPS's survey question; a person who completed 12th grade but did not receive a diploma, for example, perhaps because he failed an exit exam, may well report that he completed high school.
These problems explain the huge discrepancy between the CPS figures and everybody else's estimates, which are based on more solid data. For example, the U.S. Department of Education collects state enrollment figures and diploma counts every year. These data tell a different story. In 1998-99, for instance, there were about 3.8 million 9th graders in public schools in the United States, but when those 9th graders should have graduated (in 2001-02) only about 2.6 million high school diplomas were handed out by U.S. public schools. In 1998-99, U.S. public schools enrolled about 661,000 black 9th graders, but in 2001-02 only 325,000 diplomas were awarded to black students. Analysts have devised various methods for adjusting these figures to account for population changes and other problems. You can read Jay Greene's and my preferred method at http://www.manhattan-institute.org/html/ewp_03.htm. But you don't have to take our word for it. The Business Roundtable, the Harvard Civil Rights Project, and the Urban Institute, using different methods, have produced almost identical results. Indeed, a surprisingly broad consensus has emerged among researchers across the political landscape: the overall U.S. high-school graduation rate today is close to 70 percent and the black graduation rate is about 50 percent. That's compared with the Census Bureau estimates of 85 and 80 percent.
Sure, some small part of this discrepancy occurs because Census includes graduates from private schools as well as public schools, and private schools have a somewhat higher graduation rate. But this can't come close to accounting for the yawning gap. We know that public high schools gave out about 2.6 million diplomas in 2001-02, and private high schools gave out approximately 280,000 diplomas, adding up to about 2.9 million diplomas total. In June 2001, according to the Census, there were almost exactly 4 million 17-year-olds in the U.S. That comes out to about 72 percent. Even counting GEDs would only get us up to 80 percent.
The bottom line is that the CPS graduation rate isn't consistent with the number of young people in the population and the number of diplomas we give out. Are extra graduates parachuting in from high schools on the moon?
It's time for the Census Bureau to come clean about the unreliability of its numbers. Ideally, it would fix these problems. But that would mean changing the CPS, and the Census bureaucracy is far more interested in maintaining the year-to-year comparability of its numbers. In any case, while there are obvious improvements the Bureau should make (a more precise survey question to distinguish GEDs from real graduates, for starters), it may just not be feasible to produce a fully satisfactory graduation rate figure based on a survey. A second-best solution would be for the Bureau to stop publishing its graduation rate. But if it must put out reports claiming to provide information on educational attainment, it should at least make clear the limitations of its data - and acknowledge the wide consensus in the research field that what CPS measures doesn't reflect the important realities of American secondary education.
If the Census Bureau enjoys being the slowest, fattest target on the educational shooting range, that's its business. It should (and as a government agency probably will) keep right on doing what it's always done. But it shouldn't expect the rest of us to quit taking shots at its work any time soon.
Greg Forster is a senior research associate at the Manhattan Institute's Education Research Office (www.miedresearchoffice.org).