High school graduation and dropout rates have long been used as a central indicator of education system productivity and effectiveness and of social and economic well-being. Today, interest in the accuracy and usefulness of these statistics is particularly acute owing to a confluence of circumstances, including changing demographics, new legislative mandates, and heightened political pressures to reduce the incidence of dropping out. The population of American school-age children is shifting from native whites toward minorities and immigrants, populations that have a higher risk of dropping out; the new regime of educational accountability, especially the movement toward testing for promotion and graduation, has raised fears of a secondary effect on school dropout rates. In other words, students who are unable to pass these assessments may simply leave school before graduating. In addition, the No Child Left Behind (NCLB) Act of 2002 specifically requires an indicator of educational progress other than test scores at the high school level. Timely high school graduation appears to be the indicator of choice.
HIGH SCHOOL DROPOUT AND GRADUATION RATES
Despite the strong need for sound and reliable measures of high school dropout and completion, there has been widespread disagreement among researchers, statisticians, and policy analysts about the “true” rates, how they are best measured, and what trends are evident over time. Recently, a number of analysts have argued that the growing importance of alternative high school credentials, combined with various technical problems and political pressures,
has led to serious overreporting of “official” high school graduation rates. Their analyses produce national graduation rates of about 70 percent overall and 50 percent for minorities, numbers that are lower than those reported on the basis of official government sources (e.g., Education Week, 2009; Greene and Winters, 2002; Warren, 2004). Some researchers also contend that this problem of overreporting the graduation rate has been getting worse over time (Heckman and LaFontaine, 2008, 2010). Others (e.g., Mishel and Roy, 2006) counter that these analyses are incorrect and that the graduation rate, while still unacceptably low, has been accurately reported in national government surveys and has not changed appreciably over the past 20 years. Similar discrepancies, depending on data sources and the analyses conducted, exist in dropout and graduation estimates at state and local levels. At a time when policy makers are vitally interested in tracking the incidence of dropping out of school, they are faced with choosing among substantially discrepant estimates that would lead them to different conclusions regarding both the size of the dropout problem and how it has changed in recent years.
DATA SOURCES USED FOR ESTIMATES
Estimates of these rates are derived from a variety of sources using a variety of procedures. National estimates are derived from both cross-sectional and longitudinal sample surveys. The Current Population Survey (CPS) conducted by the U.S. Census Bureau is a nationally representative cross-sectional household survey that asks detailed questions about educational enrollment and experiences in October of each year. The National Center for Education Statistics (NCES) and the Bureau of Labor Statistics periodically conduct longitudinal surveys that track representative samples of youth through the usual high school years and beyond.
School administrative records on enrollments, dropouts, and diplomas have typically been used by states and school districts for reporting these rates. These data are reported annually to NCES as part of the Common Core of Data (CCD) collection of information on public schools in the country and have also been used to generate national, state, and district estimates of dropout and completion rates. Many states and school districts now have longitudinal unit-record administrative data systems that allow them to track the progress of individual students over time. However, decisions about ways to handle specific groups of students (e.g., students who transfer or who leave school but obtain a high school equivalency credential, like the General Educational Development [GED]) can affect the statistics that are calculated, even when the same formulas are used to calculate the rates.
Each data source brings with it a unique set of issues that can substantially affect the quality and usefulness of dropout rate statistics. Rates derived from sample-based surveys (both cross-sectional and longitudinal) have
been criticized because they rely on respondent self-reports (Heckman and LaFontaine, 2008, 2010), and some have questioned the degree to which longitudinal data accurately track disadvantaged populations (see National Research Council, 2010). Rates estimated from aggregated counts in administrative data systems have been questioned when adjustments are not made to control for repeating ninth graders or to account for transfer students (Warren, 2005). The ways that states and local school districts classify students as dropouts, graduates, or completers can significantly affect the rates that are calculated.
Whatever the data source, there are also major questions in defining both an appropriate numerator and a denominator in calculating these rates. For example, should it include private school enrollees? Recent émigrés enrolled in U.S. schools but who spent most of their education outside the U.S. education system? GED recipients? Special education students? “On-time” graduates only? Obviously, these choices should be driven by the policy questions being addressed as well as the availability of the desired data. However, until recently, no standard conventions for data inclusion or exclusion have been widely accepted in the education research and policy community. Efforts by the National Governors Association represent some progress toward standardizing methods for estimating graduation rates (National Governors Association Task Force on State High School Graduation Data, 2005). Nevertheless, there remains a lack of understanding about which calculation methods and which data are most appropriate for different policy questions, and often the best data sources may not be available for the calculations.
The Committee for Improved Measurement of High School Dropout and Completion Rates was formed to convene a workshop and to make recommendations about these issues. Specifically, the steering committee was asked to address the following questions:
What are the available measures of dropout and completion rates, how are they determined, and what are their strengths and limitations?
To what extent do current and proposed measures attain the necessary levels of accuracy, given the types of policy decision that they inform?
What is the state of the art with respect to constructing longitudinal student accounting systems for measuring dropout and completion rates? What is the feasibility and desirability of moving to such systems? What are some of the issues that need to be considered when designing these data systems?
In what ways can the analysis of data from current and proposed systems for measuring dropout and completion rates be used to help understand changes in the rates?
How can this information be used to improve practice at the local level and improve public policies at the state and national levels?
In response to the charge, the committee organized a workshop designed to explore the strengths and weaknesses of various kinds of rates, the policy decisions based on them, and the kinds of data required to inform those policy decisions. The committee began this task by conducting a review of the literature. The topic addressed by this project—dropping out of high school—is one that has been studied in great depth, and the literature base is quite expansive. A review of the entire literature base was beyond the scope and resources of this study. The committee therefore focused its review on research explicitly related to its charge: studies on the calculation of dropout and completion rates, the information needed to calculate them, and the policy uses of these rates. The committee also conducted a limited review of research on the relationships between education attainment and social and economic outcomes. This review was designed to provide context for the work and to document the value of reporting dropout and completion rates, but it was not intended to be an exhaustive review of the literature on this topic. Based on this review, the committee identified the researchers who have been actively pursuing this line of study and invited a subset of them to participate in the workshop. The committee also recruited a set of policy makers, practitioners, and stakeholders to discuss these issues during the workshop.
The workshop was held on October 23 and 24, 2008, and consisted of four panels of speakers. The first panel focused on policy uses of these rates, and panelists represented different administrative levels of the education system in this country (i.e., national, state, district, and school). The second panel made presentations about methods for calculating the rates, including discussion of the decisions required and the strengths and weaknesses of the methods. The third panel focused on development of longitudinal databases and included representatives from state and local school districts, who talked about their work to develop these systems. The final panel addressed the issue of how these data systems can be used to improve policy and practice. This panel focused specifically on early indicators of students at risk of dropping out and how this research could be used to better inform policy and practice. The workshop agenda appears in Appendix A, along with a list of workshop participants and guests. The papers and presentations from this workshop, the research that the presenters referenced, and the information that the committee gathered as part of its own literature review served as the basis for this report and the committee’s recommendations.
Throughout this report, we use several terms that warrant clarification. We use the term “graduate” to refer to a student who earns a regular high school diploma and “graduation rate” as an indicator of the percentage of students in a given population who earned a regular high school diploma. We note, however, that the definition of “regular diploma” may vary as well as the time allowed to complete it. We use the term “completer” as the all-encompassing term to refer to a student who finished high school via one of multiple ways, such as by earning a regular high school diploma, a GED, or another type of certificate (a certificate of attendance, certificate of completion, etc.). Likewise, “completion rate” indicates the percentage of students in a given population who finished high school in any of these ways. We use the term “dropout” to refer to a student who did not complete high school and “dropout rate” as an indicator of the percentage of students in a given population who did not complete high school. Dropouts may include those who earn a GED or an alternative credential (depending on the specific indicator or the purpose of the indicator), but the group does not include students still enrolled in school after they were expected to complete. There are a number of policy definitions of these terms that further specify them (e.g., NCLB specifies that the graduation rate should include only on-time diploma earners, and it classifies GED recipients with dropouts). Unless otherwise specified in the report, we use the terms in their most general sense.
There are four general categories of dropout/completion indicators, which are defined below.
Individual cohort rate: a rate derived from longitudinal data on a population of individuals who share a common characteristic at one point in time, such as entering high school. The rate is based on tracking the students over the 4 years of high school or more to determine which of them graduated and which of them dropped out.
Aggregate cohort rate: a rate designed to approximate an individual cohort rate when longitudinal data are not available by using aggregate counts of students (e.g., number of ninth graders in a given year, number of graduates in a given year). For instance, an aggregate cohort rate might compare the number of students who graduate in one year with the number of students who entered high school 4 years earlier.
Status rate: a rate that represents the fraction of a population that falls into a certain category at a given point of time (e.g., the percentage of the total U.S. population that does not have a high school diploma).
Event rate: a rate that is the fraction of a population that experiences a particular event over a given time interval. For instance, the event dropout rate indicates the percentage of students who exit school during a specific academic year without having earned a diploma.
ORGANIZATION OF THE REPORT
This report summarizes the proceedings from the workshop. Following this introduction, Chapter 2 draws on the presentations from the first panel and explains why these rates are important and how they are used for policy purposes. Based on information presented during the second panel discussion, Chapter 3 discusses the decisions that must be made in calculating these rates, and Chapter 4 explores the different types of rates and their uses. An important use of dropout and completion rates is to identify which students are likely to drop out and when they are most at risk in order to implement programs and/or interventions aimed at keeping students in school. Chapter 5 draws from several of the workshop presentations and discusses the research on early indicators of dropping out as well as on building data systems that incorporate these indicators to enable early identification of at-risk students. Chapter 6 continues the discussion of database development and summarizes the presentations made by state and district representatives participating in the third panel. Chapter 7 lays out ways the data systems can be used to improve policy and practice. The committee’s conclusions and recommendations are presented at the end of each chapter and are summarized in Chapter 8. The workshop agenda appears in Appendix A, along with a list of workshop participants and guests. Biographical sketches of committee members and staff appear in Appendix B.