Accurately Assessing High School Graduation Rates

Abstract

Recent studies have argued that there is a dropout crisis in the U.S. But Mr. Mishel and Mr. Roy find that this contention is based on flawed analyses of inadequate data. Using a much wider range of sources, they conclude that the situation is not nearly so dire as is commonly believed. ********** CONVENTIONAL wisdom asserts that there is a newly discovered dropout crisis in our high schools. As the National Governors Association states, know that about a third of our students are not graduating from high school.... About three-fourths of white students graduate from high school, but only of African American and Hispanic students do. (1) Jay Greene of the Manhattan Institute and Christopher Swanson, formerly with the Urban Institute, are primarily responsible for this claim. They have published reports showing these extraordinarily high dropout rates nationwide, for the states, and for many specific school districts. They also claim that graduation rates have been flat for several decades. (2) Greene's and Swanson's advocacy organizations, such as the Alliance for Excellent Education, the Education Trust, and the Harvard Civil Rights Project. (3) We have come to different conclusions. Both Greene and Swanson calculate dropout rates from enrollment and diploma data reported by school districts, collected by the states and forwarded to the federal government's Common Core of Data, which is managed by the National Center for Education Statistics. Using different data sources, including the U.S. Census and several high-quality longitudinal surveys, we have found that graduation rates have improved for decades, particularly for minorities, though less so in recent years. Instead of half of black students graduating with diplomas, as Greene and Swanson maintain, we conclude that three-fourths of black students obtain regular diplomas and, of the 25% who do not graduate, fully go on to obtain a GED that allows them to enroll in college, enlist in the military, or take advantage of other postsecondary training opportunities. It is a similar story for Hispanics. Rather than indicating stagnant graduation rates, census data show that the high school completion rate for young black adults (aged 25 to 29) was 42% in 1962 and roughly doubled by 2000. (4) There remains a substantial racial gap in high school completion--about 10 percentage points--but it is a far smaller gap than existed 20, 30, or 40 years ago. Trends in Hispanic graduation rates are more difficult to measure over a long period, but it is clear that graduation rates have been steadily improving over the last 20 years: the percentage of Hispanics (aged 20 to 22) who have a high school diploma (excluding GEDs) increased from 62% in 1984 to 76% in 2002. (5) In short, Greene's and Swanson's estimates are more pessimistic than ours. Ordinarily, only experts would have to make technical judgments about the relative quality of data sources and methods of analysis. But in this case, important policy implications flow from whether you believe, for example, that the black dropout rate is 50% and stagnant or 25% and falling. Neither rate is acceptable, but education policy makers would approach the problem very differently depending on which story was true. REALITY CHECK When natural or social scientists examine the reliability of data, one of their first challenges is to see whether results are consistent with other facts that they know. If data are implausible, we should be cautious about accepting them until we have carefully checked how they were derived. If, for example, we were suddenly presented with data showing a big drop in manufacturing imports and a jump in exports, we would immediately be skeptical if these data were inconsistent with unemployment figures showing large layoffs of manufacturing workers. Of course, we would still have to investigate why the import/export data were flawed, but until we did so we would be hesitant to accept them. …