Skip over navigation

Journal Issue: America's High Schools Volume 19 Number 1 Spring 2009

Finishing High School: Alternative Pathways and Dropout Recovery
John H. Tyler Magnus Lofstrom

Dropout Rates: The Magnitude of the Problem and Measurement Issues

Given the importance of graduation rates as a performance metric of the nation's high schools, one might assume the existence of well-defined, well-agreed-upon measures of that performance. One would be wrong. Although each state and the National Center for Education Statistics (NCES) all produce graduation and dropout statistics based on "standard measures," recent heated debates over the "true" rates underscore a general unease about how accurately and consistently officials are able to document school performance when it comes to graduating students.

The NCES provides the nation's most commonly cited dropout and school completion statistics. Using primarily two data sources, the Current Population Survey (CPS) and the Common Core of Data (CCD), the NCES provides four different statistics: event dropout rate, status dropout rate, status completion rate, and averaged freshman graduation rate. Table 1 defines these measures, along with the respective data sources. Figure 1 shows the trends in these four statistics from 1972 to 2005.

Based on figure 1, one might conclude that in terms of historical trends, schools are doing relatively well at moving students to graduation. School attainment appears generally to be on the rise—dropout event and status dropout rates are decreasing and school completion rates are steady (averaged freshman graduation rate) or rising slightly (status completion rate). In 2005, a relatively small share, 3.8 percent, of students dropped out of grades ten through twelve, and almost nine in ten (87.6 percent) of the country's eighteen- to twenty-four-year-olds held a high school credential. But conclusions based on these government statistics are controversial. Some observers feel that these measures paint too positive a picture of what some call a dropout "crisis," while those on the other side of the debate suggest that the government figures are at least close to the mark and that the "crisis" label is yet another undeserved black mark on the nation's schools. Driving the debate are questions about what data are used to calculate the relevant statistics and who is considered a "graduate."

Data issues primarily focus on the fact that three of the four widely used national measures—the status completion rate, the status dropout rate, and the event dropout rate—use the CPS. The CPS is a monthly survey of about 50,000 households conducted by the Bureau of the Census for the Bureau of Labor Statistics and is a primary source of information on the labor force characteristics of the U.S. population. But the CPS has some recognized deficiencies as a basis for calculating dropout and graduation statistics.

The more important issue, however, is which individuals are considered to be high school graduates. In particular, it matters substantially whether the data count individuals who leave school and later earn a General Educational Development (GED) credential as high school graduates or as dropouts. In terms of official NCES statistics, people who hold GEDs are not counted as graduates in the calculation of graduation rates, such as the averaged freshman graduation rates, but they are treated as completers in the status completion rate.4

The distinction between having a traditional high school diploma or a GED credential would be less important if the two differently credentialed groups had equally favorable outcomes in the labor market and higher education. But in terms of labor market outcomes such as wages and employment, GED holders fare consistently worse than do regular high school graduates, and GED holders also get less postsecondary education than do regular high school graduates.5 Given that dropouts who hold a GED are not the equivalent of high school graduates on two such important outcomes, it seems problematic to treat GED holders as "graduates" in official educational attainment statistics.6 Indeed, the adequate yearly progress requirements of the 2001 No Child Left Behind Act (NCLB) state that only students who receive a traditional diploma should be counted as high school graduates. Furthermore, although the GED program may be beneficial to some dropouts, it may have unintended consequences. Several studies, for example, find that the GED program may induce some students to drop out.7

Not surprisingly, there are competing views about the GED credential as a marker of successful high school "completion." For example, some states and local school districts count GED recipients as high school "completers" when computing their own administrative graduation statistics, while others stake out compromises between the two polar positions. In January of 2008, the state board of education in Virginia entertained a proposal to establish a school-level "Graduation and Completion Index" that would give regular high school graduates a weight of 1.0 and GED recipients a weight of 0.75. Satisfactory scores on this index by each school would then be a part of the state's accreditation process.8

Policies of the GED Testing Service (GEDTS) seem to establish a clear boundary between enrolled students pursuing a high school diploma and dropouts who pursue the GED credential. GEDTS policy states that the GED tests may be administered only to people who are "at least 16 years of age and not currently enrolled in an accredited high school...."9 There are, however, exceptions to the requirement that a candidate for the GED credential must be a school dropout.

In response to requests from state departments of education, the GEDTS has authorized in-school "GED Option" programs whereby some students may remain enrolled in their regular high school as they pursue a GED. Twelve states now have GEDTS authorization to offer GED Option programs to students who meet certain criteria, including credit deficiency, that place them at risk of dropping out.10 (Before 2002, a few states operated in-school programs that used the GED tests without the authorization of the GEDTS, but they no longer do so.) The ostensible purpose of these in-school GED programs, whether sanctioned or not, is to keep potential dropouts enrolled and involved in high school. Thus, even though the GED program was designed as a second-chance option for school dropouts, it has a secondary focus on dropout "prevention."11

Schools have strong incentives to participate in the GED Option program, because it allows them to continue to receive average daily attendance funds for participating students, funds they would lose were the students to drop out and leave the school rolls. A close inspection shows a lack of consistency nationwide in how GED Option students are treated in calculating graduation statistics. Some states may award students who successfully complete the GED Option program a regular high school diploma and count them as high school graduates; others count them as "completers" but not as "graduates" in calculating graduation statistics.12 No good national data on the size of the GED Option program exist, but a 2007 document from the Mississippi Department of Education concludes that if the state were to count GED Option students as high school graduates, the state graduation rate would rise from 61.1 percent to 62.9 percent.13

Comparing the NCES status completion rate, which treats GED holders as completers, with the NCES freshman graduation rate, which does not count GED holders as graduates, makes clear the importance of this issue. In 2005, the NCES status completion rate was 87.6 percent, 13 percentage points higher than the freshman graduation rate of 74.6 percent. The GED is not the only reason for the discrepancy but it may plausibly be the most important. This view is consistent with the work of James Heckman and Paul LaFontaine, who report that graduation rates estimated using data from the October CPS—the data used by NCES to generate the status completion rate—are upwardly biased by 7 to 8 percentage points because they count GED holders as high school graduates. Heckman and LaFontaine conclude that this is the most important source for overstated U.S. graduation rates.14

The GED, however, is not the only source of bias in the measures of school attrition and completion. Heckman and LaFontaine report that sample coverage (that is, inclusion or exclusion) of people who are incarcerated, people in the armed forces, and immigrants also creates a bias in secondary educational attainment measures.15 These coverage issues affect both trends and differences across groups.

Heckman and LaFontaine also find that official statistics that show white and minority graduation rates converging over time are inaccurate, particularly so for males. They note that young black and Hispanic men have been incarcerated at increasingly higher rates than young whites. Such men are not counted in the CPS-based status completion rates because the CPS sample excludes people who are incarcerated. In addition, blacks in particular have been earning GEDs at higher rates in recent years than have whites. Heckman and LaFontaine contend that white-black differences in graduation rates are roughly the same as they were thirty years ago, about a 15-percentage-point difference favoring whites.

Finally, Heckman and LaFontaine show that when comparable measures are used on comparable samples, a consensus of the graduation rate can be reached across data that have been used by various researchers—for example, the Current Population Survey, the Common Core of Data, the National Education Longitudinal Study of 1988, High School and Beyond, and the National Survey of Youth.16

The work by Heckman and LaFontaine helps to reconcile the competing dropout and graduation rate figures computed by researchers such as Jay Greene, Christopher Swanson, and Lawrence Mishel and Joydeep Roy.17 On balance, the Heckman and LaFontaine estimates suggest that today's overall graduation rates are in the 75 to 78 percent range, with white rates at 84 percent, Hispanic rates at 72 percent, and black rates at about 65 percent. These figures tend to be lower than both official government figures and those put forth by Mishel and Roy. Heckman and LaFontaine's overall graduation rate estimates are higher than the roughly 67 percent rate suggested by Greene, and their minority graduation rates are not as dire as Greene's 50 percent rate.

The apparent confusion and resulting debates over how well U.S. schools are graduating students leads one to consider what kind of data set might be a "gold standard." One possibility would be a national student ID system that would follow students no matter where they were enrolled. Thus, a student who left a school or a district or a state but re-enrolled in another school would remain in the system until graduation or until he or she otherwise left the system. But even an effective national student ID system would not reveal that some ninth graders had dropped out and left the system until they failed to show up as graduates with their age cohort. Of course, the missing graduates could not be counted as "dropouts" without giving them an extra year or two to graduate in case they had been held back a grade in high school or had decided to return to school. The problem is that even with a very good, individual student ID data system, a dropout becomes a dropout when he or she leaves school and the school-leaving often happens without the kind of consultation that would allow for accurate data coding as to dropout status.

If schools are to do a better job at having up-to-date information on dropout and graduation rates, they must have more accurate and more appropriate data. And although the Heckman and LaFontaine effort may go a long way in quelling the "dropout debates," it provides no information that a state, school district, or school can use to inform practice and policy. Most researchers who have explored this topic agree that the starting point for quality data is with a student-level ID that would allow states to follow students across schools and districts at least within states and over time.

One national effort to promote consistent state information on student performance is the Data Quality Campaign, which provides guidelines for what constitutes "good graduation and dropout data."18 But even when acceptable data systems are in place, the question remains how administrative units at the state, district, and school level will use those data in their reporting. For example, how will these units count GEDs when the incentives from virtually all sources are to have graduation rates that are as high as possible? New Jersey might provide some insight on this question. Heckman and LaFontaine report that in New Jersey, an individual need only mail in GED test scores that meet the state GED score requirements to qualify for a state-endorsed high school diploma. These newly credentialed individuals are then included in the official state diploma counts. The critical issue here, as Christopher Swanson and Duncan Chaplin have pointed out, is that under the federal system it is states, not the federal government, that have final authority in determining requirements for a high school diploma.19 Thus, agreement across states on what represents high school completion may be as important as data development and consistency when it comes to developing "well-defined and well-agreed-upon" measures of schools' performance in graduating students. The U.S. Department of Education recently recognized the need for action and tightened the NCLB regulations regarding how states calculate high school graduation rates.

We close this section by suggesting the need to consider current graduation rates in the context of historical trends of this measure and the need to consider both trends and current rates in the context of the current global economy. Our interpretation of the research at hand is that graduation rates have certainly not been increasingly steadily since 1960, as table 1 would suggest. Neither have they been in a steady decline. Rather, the evidence from Heckman and LaFontaine suggests a 2- to 3-percentage-point fluctuation around a relatively flat forty-year trend line centered at about 77 percent. Thus, schools are apparently doing about as well now as they were forty years ago in terms of graduating students. The problem is that just as the competitive pressures associated with an increasingly global economy have increased, the importance of education in determining personal and national well-being has also grown. "Steady as she goes," then, is an alarming rather than comforting reality when it comes to how well the nation is getting students successfully through high school.