Dropping out of high school is related to a number of negative outcomes. For example, the median income of high school dropouts age 18 and over was $12,184 in 2003 (U.S. Census Bureau 2005). By comparison, the median income of those age 18 and over who completed their education with a high school credential (including a General Educational Development (GED) certificate) was $20,431. Dropouts are also less likely to be in the labor force than those with a high school credential or higher, and are more likely to be unemployed if they are in the labor force (Bureau of Labor Statistics 2004). In terms of health, dropouts over the age of 24 tend to report being in worse health than adults who are not dropouts, regardless of income (U.S. Department of Education 2004). Dropouts also make up disproportionately higher percentages of the nation's prison and death row inmates.1
This report builds upon a series of National Center for Education Statistics (NCES) reports on high school dropout and completion rates that began in 1988. It presents estimates of rates in 2002 and 2003, provides data about trends2 in dropout and completion rates over the last three decades, and examines the characteristics of high school dropouts and high school completers in 2002 and 2003. Four rates are presented to provide a broad picture of high school dropouts and completers in the United States, with each contributing unique information: the event dropout rate, the status dropout rate, the status completion rate, and the averaged freshman graduation rate—an indicator new to this report series.
Data presented in this report are drawn from the annual October Current Population Survey (CPS), the annual Common Core of Data (CCD) collections, and the annual GED Testing Service (GEDTS) statistical reports.4 Data in the CPS files are collected through household interviews and are representative of the civilian, noninstitutionalized population in the United States. The CCD data are collected from state education agencies about all public schools and school systems in the United States, and contain administrative record data that are representative of all public school students in this country. The GEDTS data are also built from administrative record data, and contain information about all GED test takers (data presented in this report are only for test takers in the 50 states and the District of Columbia).
As with all data collections, those used in this report are useful for calculating some estimates but are poorly suited for calculating other types of estimates. For example, CPS data are well suited for studying the civilian, noninstitutionalized population in the United States, but do not provide information about military personnel or individuals residing in group quarters such as prison inmates. Data from CCD are appropriate for studying public school students in a given year, but do not provide information on private school students. GEDTS data are helpful for identifying the number of people who take and pass the GED examination in a given year, but do not contain information about schools that GED test takers attended before taking the GED test. In addition, none of the data sets track individual students over time, limiting their usefulness for studying processes and precise timelines associated with graduating or dropping out. Note that the CCD data for high school dropouts in the 2002-03 school year were not available at the time this report was written. However, diploma data for the 2002-03 school year were available, which is why 2002-03 averaged freshman graduation rates are presented, but not state-level public high school event dropout rates.
All changes or differences noted in this report are statistically significant at the p ≤ .05 level. When significance tests fail to meet the p ≤ .05 criterion and the comparison is of substantive interest, terminology such as "no measurable difference was found" is used in this report. This does not necessarily mean that there is no actual difference between the compared estimates. With a larger sample, the difference may have tested significant at the p ≤ .05 level.