No. 31 November 2002
Public School Graduation Rates in the United States
Jay P. Greene, Ph.D.
Senior Fellow, Manhattan Institute for Policy Research
Marcus A. Winters
Research Associate, Manhattan Institute for Policy Research
The report’s main findings are the following:
- The national graduation rate for the public school class of 2000 was 69%. The rate for white students was 76%; for Asian students it was 79%; for African-American students it was 55%; for Hispanic students it was 53%; and for Native Americans it was 57%.
- Florida’s public schools had the lowest overall graduation rate in the nation with 55% of students graduating, followed by Georgia, the District of Columbia, and Arizona.
- New Jersey had the highest overall graduation rate with 87%, followed by North Dakota, Utah, and Iowa.
- Wisconsin had the lowest graduation rate among African-American public school students with 41%, followed by Florida, Oregon, and Tennessee. The highest rate of graduation among African-American students was 74% in West Virginia, followed by Arkansas, Massachusetts, and Virginia.
- Mississippi had the lowest graduation rate among Hispanic public school students with 23%, followed by New York, Oregon, and Florida. The highest rate of graduation among Hispanic students was 73% in Louisiana, followed by Wyoming, Hawaii, and Virginia.
- Nebraska had the lowest graduation rate among Native American public school students with 40%, followed by Minnesota, Nevada, and Oregon. The highest rate of graduation among Native American students was 86% in Alabama, followed by Illinois, Oklahoma, and Texas.
- Graduation rates for African-American, Hispanic, and Native American public school students were particularly low in a number of states; for each group there were six different states with graduation rates below 50%.
- Rhode Island had the lowest graduation rate among Asian public school students with 66%, followed by Tennessee, Hawaii, and Massachusetts. The highest rate of graduation among Asian students was 95% in Illinois, followed by Missouri, Oklahoma, and Maryland.
- Florida had the lowest graduation rate among white public school students with 60%, followed by Tennessee, Georgia, and Alaska. The highest rate of graduation among white students was 89% in North Dakota, followed by South Dakota, Nebraska, and Iowa.
- The National Center for Education Statistics (NCES) finds a national high school completion rate of 86.5% for the class of 2000. The discrepancy between the NCES’ finding and this report’s finding of a 69% rate is largely caused by NCES’ counting recipients of General Educational Development (GED) certificates and other alternate credentials as high school graduates, and by its reliance on a methodology that is likely to undercount dropouts.
About the Authors
Jay P. Greene is a Senior Fellow at the Manhattan Institute for Policy Research where he conducts research and writes about education policy. He has conducted evaluations of school choice and accountability programs in Florida, Charlotte, Milwaukee, Cleveland, and San Antonio. He has also investigated the effects of school choice on civic values and integration.
His research was cited four times in the Supreme Court’s opinions in the landmark Zelman v. Simmons-Harris case on school vouchers. His articles have appeared in policy journals, such as The Public Interest, City Journal, and Education Next, in academic journals, such as The Georgetown Public Policy Review, Education and Urban Society, and The British Journal of Political Science, as well as in major newspapers, such as the Wall Street Journal and Christian Science Monitor. Most recently he published a piece on vouchers and school integration in the Wall Street Journal, analyses of problems with special education in Education Week, National Review Online and The Education Gadfly, and a defense of high stakes testing in Education Next.
Greene has been a professor of government at the University of Texas at Austin and the University of Houston. He received his B.A. in history from Tufts University in 1988 and his Ph.D. from the Government Department at Harvard University in 1995. He lives with his wife and three children in Weston, Florida.
Marcus A. Winters is a Research Associate at the Manhattan Institute’s Education Research Office where he studies and writes on education policy. He recently graduated from Ohio University with a B.A. in political science, for which he received departmental honors, and a minor in economics.
We would like to thank Rob Fusco for his research assistance on this project. We would also like to thank Kaleem Caire and the Black Alliance for Education Options who, in their belief in the importance of an accurate picture of graduation rates, helped make this report possible by commissioning and supporting Dr. Greene’s original study of public high school graduation rates for the class of 1998.
Society puts a great deal of emphasis on graduating from high school . . . and for good reason. In addition to being a prerequisite for college, earning a high school diploma is a reliable indicator of future economic success. Statistics show that on average high school dropouts earn salaries far lower than high school graduates and in general are more likely to place burdens upon society.
Because of the importance of obtaining a high school diploma, communities rightly expect their neighborhood schools to make it one of their main goals to graduate as many of their students as possible. Unfortunately, it is rare for graduation rates to be widely publicized, and frequently those that are disseminated do not measure what communities recognize as a graduation rate.
Official graduation statistics are too often based upon definitions or allow exemptions that prevent the results from conforming to our common-sense understanding of what a graduation rate should be. Most people consider any student that finishes high school with a regular high school diploma to be a graduate and students who fail to do so as dropouts. Too often official graduation statistics fail to meet this criterion. For example, in Washington State only students who have completed the paper work necessary to be officially considered dropouts are reported as such. Students who did not fill out the necessary paperwork but are no longer in school are considered “unknown”, though the state admits that many of them are in fact dropouts.
A report by the Manhattan Institute and the Black Alliance for Educational Options published last year, “High School Graduation Rates in the United States,” addressed these problems by introducing the Greene Method to calculate graduation rates simply and with reasonable accuracy. The technique produces an accurate estimate of the graduation rate by comparing the number of students that enter a high school class to the number of students receiving a regular diploma, with some adjustments for population change.
This report uses a newly refined version of the Greene Method to calculate graduation rates for the public school class of 2000. We also compare these results to the public school class of 1998, as recalculated using the refined method.
Unlike last year’s report, here we have not calculated graduation rates for the nation’s 50 largest school districts. We have only calculated state and national figures in order to limit data collection to a single source, described in the next section. Using a single source provides greater confidence that the numbers we use in the report are, or at least should have been, collected and reported according to the same guidelines.
The method used in this report to calculate graduation rate estimates is essentially the same as the one used in last year’s report, with two refinements. The refinements should improve the precision of the estimates by better calculating the number of students in a given class. This method is not intended to produce pinpoint graduation statistics, but rather to generate a reliable estimate of the ratio of those who entered a high school class to those who satisfactorily completed graduation standards.
The data used for this report came from the Core of Common Data (CCD) at the National Center for Education Statistics (NCES), which is a unit of the U.S. Department of Education. First, we estimated the number of students in the graduating cohort by attempting to find how many students entered a class in the 9th grade in 1996 and should be expected to graduate four years later in 2000. We did this by averaging enrollment for a given class in the 8th, 9th and 10th grade years, 1995–96, 1996–97 and 1997–98 respectively.
This average serves as a “smoothed” estimate of the cohort’s 9th grade enrollment. We did this because the population in this cohort may change between 8th and 9th grades as some students transfer between the public and private sectors during the transition from middle to high school. Ninth grade enrollments are also inflated by the fact that a significant number of students tend to be held back in that grade. And 10th grade enrollments tend to drop following the artificially inflated 9th grade figures and because students often begin dropping out of school between 9th and 10th grades.
We then measured population changes that would affect the cohort’s enrollment in the four years after 9th grade in order to control for changes in enrollment levels caused by students moving in and out of a state rather than by students dropping out of school. We assumed that the change in the population of our cohort class would mirror the change in population of the entire high school population among the relevant population group (i.e., public school students in Wisconsin, African-American students in Alabama, etc.) during the same years. To do this, we took the difference in total enrollment in high school (grades 9–12) between the 9th grade and 12th grade years of the cohort class (1996–97 and 1999–2000 respectively) and divided it by the total enrollment in the 9th grade year in order to estimate the percent change in high school population.
Next we increased or decreased the “smoothed” cohort enrollment by the percent change in population. This gave us a reasonable estimate of how many students should have graduated high school in 2000 if no students had dropped out.
Finally, we divided the number of students who actually received a regular diploma in 2000 by the estimated number of public school students who should have graduated if none had dropped out, producing the state’s estimated graduation rate.
To illustrate the method, let us look at how we calculated the national public school graduation rate. First we averaged the enrollments in the 8th, 9th and 10th grades during the 1995–96, 1996–97 and 1997–98 school years, respectively. This gave us a smoothed 9th grade enrollment estimate of 3,386,591 for the 1996–97 school year.
Next, we found the enrollment change within the nation for the years the cohort was in high school. To do this, we subtracted the 1996–97 total grade 9–12 enrollment from the 1999–2000 total grade 9–12 enrollment to get an increase in enrollment of 534,278. We divided the total change in enrollment by the total 1996–97 enrollment (12,177,494) to get a change in population of 4.387%.
We multiplied our smoothed 9th grade enrollment (3,386,591) by the change in population (.04387), giving us an estimated 9th grade change based on a change in population of 148,569 students. We then added this number to the original smoothed 9th grade enrollment in 1996–97 (3,386,591) to get the number of students that we should expect to graduate in 1999–2000, 3,535,175.
Finally, we divided the number of students who received a regular diploma in 1999–2000 (2,456,116) by the number of students we projected should have graduated (3,535,175) to get a national graduation rate of 69%.
Now let us follow this procedure when calculating a racial breakdown graduation rate for a state. To do this we shall illustrate how we found the graduation rate for African-Americans in Wisconsin public schools.
First, to find the smoothed 9th grade enrollment for 1996–97 we averaged the 8th, 9th and 10th grade enrollments in 1995–96, 1996–97 and 1997–98 respectively. This produced a smoothed estimated 9th grade enrollment of 6,171.
Next, we subtracted the combined grade 9–12 African-American enrollment in 1996–97 (20,006) from the combined grade 9–12 African-American enrollment in 1999–2000 (20,583) to get an increase of 577 students. We then divided this number by the combined grade 9–12 enrollment in 1996–97 (20,006) to get a population change of 2.884%. We multiplied our smoothed 9th grade enrollment estimate (6,171) by our population change (.02884) and added that value (178) to our smoothed 9th grade enrollment estimate (6,171). This gave us an estimate of 6,349 students that should be expected to graduate in 1999–2000.
Finally, we took the number of African-American public school students in Wisconsin who actually received a regular diploma in 1999–2000 (2,573) and divided it by the number of students we estimated should graduate (6,349). This gave us a graduation rate of 41%.
We produced results for each state, both total and broken down by racial category. We also used the same revised version of the Greene Method to recalculate the 1998 high school graduation rates, for the sake of comparison. The two refinements of the Greene Method since the previous report are that we “smooth” 8th, 9th, and 10th grade enrollments instead of simply using the 8th grade enrollment and we adjust for population change using only change in high school enrollments and not change in total student membership. We believe that “smoothing” the enrollment figures improves the accuracy of our estimate of the number of students in the cohort entering high school. And adjusting the cohort to mirror the change in high school population is a more precise adjustment than adjusting by changes in the total school enrollment.
Though the Greene Method’s required adjustments of enrollment data are effective in a large cohort, its estimates are vulnerable in cases where particular instances can have large effects on the outcome. Cases with particularly small cohorts and those with exceptionally high population changes are more susceptible to unique events for which the adjustments cannot correct. Cohorts containing these irregularities distorted the results of our enrollment adjustments and often produced results that were implausible.
Because of the sensitivity of our estimates to enrollment anomalies we developed rules for eliminating graduation estimates from our analysis. We eliminated any cohort for which the smoothed 9th grade enrollment estimate before adjusting for population change was fewer than 200 students as well as any cohort for which there was a greater than 30% population change. Furthermore, if a cohort had a smoothed 9th grade enrollment estimate of fewer than 2,000 students we eliminated it if it had a population change greater than 20%. These rules allow us to focus exclusively on the cohorts for which we have the greatest confidence and eliminate those where anomalies within the population or a limited cohort enrollment were more likely to taint the results.
The total and racial category graduation rate estimates for the nation and each state, as calculated using the Greene Method, are listed in Table 1. For the public school class of 2000, we found a national graduation rate of 69%, a 1% increase from 1998. When we broke down the national cohort into racial categories we found that Hispanic students posted the lowest graduation rate at 53%, followed by African-Americans with 55% and Native Americans with 57%. Whites and Asians faired better, with graduation rates of 76% and 79% respectively. These results showed a 2% increase for African-American students, a 1% improvement among Hispanics and whites, and no change for Native Americans and Asians since 1998. The gains made by the racial groups and the nation as a whole are not large enough to justify the conclusion that any significant change has actually taken place since 1998.
Table 2 ranks the states according to their total graduation rate estimate. New Jersey ranked first among the states with a graduation rate of 87% in 1999–2000. Closely following New Jersey were North Dakota and Utah with 86%, and Iowa with 85%. Florida ranked last among the states with a 55% graduation rate. Other states at the bottom of the rankings were Georgia with 56%, Washington, D.C. with 58%, and Arizona and South Carolina with 59%.
As mentioned, we eliminated cohorts in which the smoothed pre-adjusted 9th grade enrollment estimate was too small or the population change was too large. This led us to exclude a significant number of states from our racial breakdown. We were also limited in this respect by the number of states that reported statistics broken down by race. 12 states did not supply CCD with sufficient data for us to calculate graduation rates for their racial categories for 2000. This is, however, an improvement from the 15 states that did not supply sufficient information for the 1998 school year.
Because many states are not included in the racial breakdown the rankings in this respect are less reliable as a comparison of the states. They do, however, remain of interest when they show large disparities with the total rankings. Tables 3–7 rank the states according to their estimated graduation rate for each racial category in our analysis: white, African-American, Hispanic, Asian and Native American, respectively.
Our racial breakdown showed that some states with high overall graduation rates performed significantly worse at graduating minority students. Nebraska, which ranked 5th among the states in overall graduation rate with 84% in 1999–2000, ranked 24th among the 31 states reporting enough information for our analysis at graduating African-American students with 53%. Nebraska also ranked last among the 17 states in our analysis at graduating Native Americans with a graduation rate of 40%. Iowa, whose total graduation rate of 85% ranked 4th among the states, graduated only 58% of its African-American students. By contrast, Washington D.C. ranked near the bottom in overall graduation percentage, but was ranked 10th at graduating Hispanics with 55% and 5th at graduating African-Americans with 64%.
The low graduation rates among minorities are perhaps the most disturbing results produced by this report. The highest state graduation rates among Hispanics and African-Americans are comparable to the lowest rates among whites and Asians. In fact, the lowest state graduation rate among whites in 1999–2000, 60% in Florida, would have ranked 12th among African-American students and 6th among Hispanics.
Another disturbing result we found was the often-large disparity between our estimate and the national graduation rates as reported by the National Center for Education Statistics (NCES). According to a report by NCES, the national high school completion rate in 2000 was 86.5%. The national graduation rate according to our analysis is 69%. Much of the difference between our result and NCES can be attributed to their counting recipients of high school equivalency certificates, such as the GED, as graduates. The NCES national high school completion rate is also subject to inflation because it is derived from a survey that relies upon accurate self-reporting by respondents.
People who received any certificate other than a regular diploma or above are not counted as graduates in this report. There are several reasons that we exclude high school equivalency recipients in our calculations. First, the purpose of a graduation rate is to evaluate schools, not the dropouts themselves. Though it may be beneficial for an individual to acquire a GED, those students cannot be said to have graduated from any particular high school. A student who may have received an equivalency certificate from a community college, from prison, or several years after he has left high school cannot be credited to his past high school as a graduate.
Second, the GED is not equivalent to a high school diploma. The effort and knowledge necessary to obtain a high school equivalency certificate is not the same as is required to graduate from high school with a regular diploma. Most importantly the future prospects of those receiving a GED are more closely related to dropouts than graduates. Some researchers find moderate benefits from obtaining a GED,  while research by Stephen Cameron and Nobel Prize winning economist James Heckman finds that there is no difference between the outcomes for a dropout and a recipient of an equivalency certificate. Though there may be disagreement over the degree of difference between GED recipients and dropouts, no study that we are aware of claims that outcomes for a recipient of a GED are equivalent to those who receive a regular diploma.
NCES calculates a 4-year completion rate in another less publicized report, which we feel better represents a true graduation statistic, though it too is limited.  To calculate a 4-year completion rate, the NCES collects information on dropouts, enrollment and completers from the many states. The states are told to report as drop-outs any students that were enrolled in school at some time during a particular year, did not transfer to another school, miss school because of a suspension or school-excused illness, or die, and were not enrolled at the beginning of the next year. This is a reasonable definition of a dropout. NCES then divided the number of students that received a diploma in the 1999-2000 school year (their year 4) by the sum of the dropouts in that cohort’s 9th, 10th, 11th and 12th grade years (years 1,2,3, and 4 respectively) and the number of students that received a diploma in the 12th grade year (year 4). The formula used by NCES is:
High School Completers Year 4
Drop Outs (Grade 9 Year 1 + Grade 10 Year 2 + Grade 11 Year 3 +
Grade 12 Year 4) + High School Completers Year 4
In the report, NCES provides high school completion rates for 33 states, but did not report a national high school completion rate. The study reports a completion rate that includes equivalency degree recipients and one that only includes those who achieved a regular diploma.
Though this second NCES method for calculating graduation rates seems reasonable, in practice it produces distorted numbers. The reason for this is that to calculate its graduation rate NCES depends upon the correct reporting of dropouts by the many states. Dropout numbers are susceptible to state misclassification and misreporting (see the Washington state example cited in the introduction). Unfortunately many states neither have the resources nor the incentives to track the whereabouts of individual students accurately. Where information is ambiguous school and state officials may have incentives to offer the most benign explanations possible and thereby reduce the number of students classified as dropouts.
Using dropouts rather than the number of students enrolled in a cohort allows the states, through improper reporting, to provide information that would cause NCES to overestimate completion rates. In its report, NCES points to this problem, writing,
“. . . State and local policies and data collection administration may have profound effects on the count of dropouts and completers reported by a state. . . . Although state CCD Coordinators verify each year that they have followed the CCD dropout definition, states vary in their ability to track students who move in and out of districts, and it is probable that some students have been misclassified.”
Unlike the NCES method, the Greene Method is not susceptible to the limitations of dropout reporting by the states. Because the Greene Method relies solely on enrollment data, which are more reliable than dropout numbers, it is able to more accurately estimate a graduation percentage. Students that may have been lost in the NCES report because of improper reporting by the states would show up in the enrollment data used to estimate graduation rates using the Greene Method. Using enrollment data rather than reported dropouts allows us to better eliminate the influence states can have on their own reported graduation rate.
NCES reported graduation rates for 33 states. Of these states, seven differ from our numbers by at least ten percentage points. This causes a great disparity in the rankings of the states. According to the NCES numbers, Wisconsin ranks first of 33 in total graduation rate at 89.3%, while with a graduation estimate of 81% it ranked 9th according to our calculations. With graduation percentages of 73%, Maine and Massachusetts ranked 24th and 25th respectively using the Greene Method, while NCES ranked Maine 6th with a completion rate of 86.1% and Massachusetts 7th with a completion rate of 85.5%. Of the 18 states for which NCES did not report graduation rates, only 3 ranked in the top half of our analysis, so disparities at the top of the list were primarily due to differences in the data rather than the absences of these states from the NCES report. Table 8 compares this NCES report’s graduation rates with our estimates.
This is not to say that our estimates are necessarily the correct graduation rates for these states. Nevertheless, such drastic differences between estimates arrived at through the Greene Method and the NCES numbers should raise questions about what happened to those extra students, and whether states are reporting dropout information correctly. Further adding to this puzzle is the fact that in all 15 cases where the difference between results reported by the NCES and those calculated by the Greene Method were greater than 5%, the Greene Method produced a lower graduation rate. For example, according to the NCES report the total graduation rate in Massachusetts is 85.5%, significantly larger than our estimate of 73%. This difference of 12.5% should raise a red flag that something could be wrong with the reporting process in Massachusetts.
This report’s utilization of the Greene Method for calculating public school graduation statistics is a useful check on state’s officially reported graduation rates. Our utilization of public enrollment and diploma data can be used to find anomalies in state reports. By limiting our analysis only to those generally acknowledged as graduates, we offer the public a clearer interpretation of what it understands as a graduation rate. Also, though we are not able to follow and count individual students over time because of privacy restrictions, in the cases where our estimate varies considerably from official graduation rates we should be concerned with whether states are providing accurate information on this very important topic.
This report gives us a reliable and straightforward estimate of the graduation rate in the nation and for the states individually. The graduation rates as reported in this study tell us that fewer students are graduating from high school than our society recognizes and far fewer than it desires. When more than 3 in 10 students in the nation choose a path, dropping out of high school, that can seriously diminish their future outcomes we are right as a society to have major concerns. Where we see severe problems we should be more open to new ideas for how to revitalize our schools and improve those situations.
- See http://ferret.bls.census.gov/macro/032000/perinc/new03_001.htm.
- Phillip Kaufman, Jin Y. Kwon, and Steve Klein, “Dropout Rates in the United States: 1999,” National Center for Education Statistics, Statistical Analysis Report, November 2000, p.1.
- See State Nonfiscal Public Elementary/Secondary Education Survey Data http://nces.ed.gov/ccd/stnfis.asp.
- The illustration may produce different results than if followed straight through due to rounding error.
- There were some anomalies in the NCES data that we attempted to correct. In Rhode Island, NCES had used its combined 9–12 enrollment in 1999–2000 for white students as the total 9–12 total enrollment as well. To fix this error we simply added the racial breakdown enrollments in grades 9–12 together to achieve a total enrollment. We corrected a similar problem in the calculation of Iowa’s 1998 graduation rate. NCES had reported the same number of white students who had received a regular diploma in 1998–99 as the total number who received a diploma in that year. Again, we added together the racial breakdown numbers to achieve a total number of students who received a regular diploma in 1998–99.
We also found irregularities in the enrollment numbers for Michigan and Ohio. We then contacted the department of education in these states. We were informed that the enrollment data for Michigan were incorrect because it did not include students from some major cities, such as Detroit and Lansing, which tainted the results for the racial breakdowns and the total graduation rate. Ohio informed us that the racial breakdown numbers it reported to NCES were incorrect because they did not include information from Cleveland, but this should not hinder the calculation of the total graduation rate. We therefore deleted all of the graduation estimates for Michigan and the racial breakdown graduation estimates for Ohio.
- Phillip Kauffman, Martha Naomi Alt, and Christopher Chapman, “Dropout Rates in the United States: 2000,” National Center for Education Statistics, Statistical Analysis Report, November 2001, Table 4, p. 20. http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2002114.
- See for example, Richard J. Murnane, John B. Willett, and Kathryn Parker Boudett “Do High School Dropouts Benefit from Obtaining a GED?” Educational Evaluation and Policy Analysis, 17(2), 1995, p.133–147.
- Stephen Cameron and James Heckman, “The Nonequivalence of High School Equivalents,” Journal of Labor Economics, volume 11, number 1, 1993, p. 1.
- Beth Young, “Public High School Dropouts and Completers from the Core of Common Data: School Years 1998–99 and 1999–2000” National Center for Education Statistics, August 2002. Table 2, p. 6. http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2002382.
- Beth Young, “Public High School Dropouts and Completers from the Core of Common Data: School Years 1998–99 and 1999–2000” National Center for Education Statistics, August 2002, p. 1.