1 Background and Context
Failure to complete high school has been recognized as a social problem in the United States for decades and, as discussed below, the individual and social costs of dropping out are considerable. Social scientists, policy makers, journalists, and the public have pondered questions about why students drop out, how many drop out, what happens to dropouts, and how young people might be kept from dropping out. Currently, many voices are arguing about the effects of standards-based reforms and graduation tests on students' decisions to drop out and about which dropout counts are correct. A significant body of research has examined questions about dropouts, and this section of the report provides an overview of current knowledge about these young people. We begin with a look at the history of school completion.
CHANGING EXPECTATIONS FOR STUDENTS
Expectations for the schooling of adolescents in the United States have changed markedly in the past 100 years. Indeed, the very notion of adolescence as a phase of life distinct from both childhood and adulthood came into common parlance only in the first decades of the twentieth century, at roughly the same time that educators began to develop increasingly ambitious goals for the schooling of students beyond the eighth grade (Education Week, 2000:36). At the turn of the last century, as Sherman Dorn noted in the paper he prepared for the workshop, “fewer than one of every
ten adolescents graduated from high school. Today, roughly three of every four teens can expect to earn a diploma through a regular high school program” (Dorn, 2000:4).
High school in the early part of the century was a growing phenomenon, but it was still made available primarily to middle- and upper-class students and was generally focused on rigorous college preparatory work. At the turn of the century, the lack of a high school diploma did not necessarily deter young people from going on to successful careers in business or politics. As the number of students enrolled in high school grew, from approximately 500,000 in 1900 to 2.4 million in 1920 and then to 6.5 million in 1940, notions of the purpose of postelementary schooling were evolving.
Dorn provided the committee with an overview of trends in graduation rates over the twentieth century, noting three features of the overall trend that stand out: 1 (1) a steady increase in graduation rates throughout the first half of the twentieth century; (2) a decrease around the years during and immediately after the Second World War; (3) a plateau beginning with the cohort of students born during the 1950s. He discussed possible explanations for these changes in school completion rates.
One possible explanation is the influence of changes in the labor market. A number of developments had the effect of excluding increasing numbers of young people from full-time employment in the early decades of the twentieth century, including the mechanization of agriculture, increases in immigration, and the passage of new child labor laws. As teenagers had more difficulty finding work, increasing numbers of them stayed enrolled in school. The dip during the later 1940s is correspondingly explained by the fact that it was not only adult women who moved into the workforce to replace male workers who left employment for military service, but also teenagers of both sexes. The postwar dip and plateau also correlates with the growing availability of part-time employment and other labor opportunities for teenagers, which challenged the perception that completing school was important to financial success.
Dorn describes a pattern in which participation in successive levels of schooling gradually increases until the pressure spills over into the next level. Increasing proportions of the potential student population tend to
1 Dorn based his discussion of the trendlines on the Current Population Survey, census data, and state and district administrative data sources.
participate in schooling to a given level until saturation is reached—that is, until virtually all are enrolled. Expectations regarding participation in the next level then expand, and the pattern is repeated. In the United States, the norm has moved from primary schooling, to the eighth-grade level, and then to high school completion. State laws regarding school enrollment have moved along with these expectations. Currently, most states require that students stay enrolled through the age of 16. The steady increase in high school enrollment during the first half of the century thus reflects the gradual development of the now widely shared conviction that all teenagers should complete high school. Current political discourse reflects a developing expectation that the majority of students will not just complete high school but also participate in some form of higher education.
It was not until the 1960s that dropping out was widely considered a social problem because it was not until midcentury that sufficient percentages of young people were graduating from high school so that those who did not could be viewed as deviating from the norm. Dorn illustrated the views of dropping out that were becoming current in that period with this 1965 quotation from sociologist Lucius Cervantes (quoted in Dorn, 2000:19):
It is from this hard core of dropouts that a high proportion of the gangsters, hoodlums, drug-addicted, government-dependent prone, irresponsible and illegitimate parents of tomorrow will be predictably recruited.
A number of scholars have argued that as enrollments have increased, high schools' missions have evolved. Many jurisdictions responded to the arrival of waves of immigrants by making it more difficult for families to avoid enrolling their children in school, arguing that public schools were the best vehicle for assimilating these new citizens and would-be citizens (Education Week, 2000:4). As the children of the lower and middle classes entered high school, however, expectations and graduation standards were lowered. Thus, the postwar plateau might also be explained by the notion that, as Dorn put it, “by the 1960s high schools really had succeeded at becoming the prime custodians for adolescents” (Dorn, 2000:10). If high schools were actually providing little benefit for the students on the lower rungs of the socioeconomic ladder, according to this reasoning, there was little motivation for increasing the graduation rate from 70 or 80 percent to 100 percent.
Another notable trend was the general decrease in gaps between completion rates for whites and nonwhites and other population subgroups.
Observers have noted that this narrowing of the gap relates to the saturation effect described earlier—completion rates for Hispanics and African Americans have moved up while those for whites have remained level (Cameron and Heckman, 1993a:5). At the same time, however, alternative notions of school completion have proliferated (discussed in greater detail below). Dorn called attention to the fact that in Florida six different types of diplomas are available and that other states have adopted similar means of marking differing levels of achievement. The categories of school completion are not fixed and apparently not of equivalent value; it may be that many minority students who have converted statistically from dropouts to school completers have in fact moved to an in-between status that needs to be better understood. This circumstance significantly complicates the task of statisticians and others who attempt to keep track of students' progress through school. It also complicates policy discussions about social goals for young people, expectations of the education system, and possible solutions to the problem of dropouts.
LOOKING AT DROPOUTS
A recent report from the National Center for Education Statistics (NCES) shows that five percent of all young adults who were enrolled in grades 10-12 (519,000 of 10,464,000) dropped out of school between October 1998 and October 1999 (National Center for Education Statistics, 2000:iii). That report provides a wealth of other important information, noting, for example, that Hispanic and African American students are significantly more likely than white students to drop out and that students from poor families are far more likely to drop out than are students from nonpoor families. The report provides information on trends in dropout rates over time and comparisons among students by age, racial and ethnic characteristics, and the like.
The statistical information in this and other reports is valuable, but it provides only a snapshot of the situation across the country. General statistical reports are not designed to reveal the effects of particular policies, programs, and educational approaches on particular groups of students, but variations in the numbers suggest possible sources of more detailed understanding. School completion rates reported by states and districts show wide variation, for example, from 74.5 percent for Nevada to 92.9 percent for Maine. The rates at which students complete school vary over time and are different for different population subgroups, regions, and kinds
of schools, and for students who differ in other ways. (The school completion rate is only one of several ways of measuring dropout behavior; see discussion below). The reported data (from NCES) suggest that particular factors are associated with dropping out, such as single-parent homes, teenage pregnancy, history of academic difficulty, and retention in grade. Other researchers have identified specific school factors that are associated with dropping out, discussed below.
The rates can be calculated in different ways, which means that dropout or school completion rates for the same jurisdiction can look very different, depending on which method is used. Indeed, there is no single dropout measure that can be relied on for analysis; there are many rates based on different definitions and measures, collected by different agents for different purposes. The NCES report, for example, opens by presenting two calculations of dropouts, 5 percent and 11 percent, respectively, for slightly different groups, as well as a percentage of school completers, 85.9 percent (2000:iii).
The confusion about counting dropouts is not surprising when one considers the challenges of counting students in different categories. Numerous decisions can drastically affect the count: At what point in the school year should student enrollment be counted? Should it be done at every grade? How long should a student's absence from school be to count as dropping out? What age ranges should be considered? What about private and charter schools and students who are home-schooled? In most school districts and states, significant numbers of students move into and out of their jurisdictions each year, so school careers are difficult to track. Even within a jurisdiction, many students follow irregular pathways that are also difficult to track—they may drop out of school temporarily, perhaps more than once, before either completing or leaving for good. Different jurisdictions face different statistical challenges, depending on the composition of their student populations. Districts with high immigrant populations may have large numbers of young people who arrive with little documentation of their previous schooling, so that determining which among them have completed school is difficult. What students do after dropping out is also highly variable. Alternative educational and vocational programs, which may or may not be accredited means of completing secondary schooling requirements, have proliferated. A significant number of students take the General Educational Development (GED) Test every year; many (but not all) of them receive school completion credentials from their states.
Tracking dropout behavior is clearly messy. In response, statisticians have devised a variety of ways of measuring the behavior: status dropout rates, event dropout rates, school completion rates, and more. Unfortunately, the many measures often lead to confusion or misunderstanding among people trying to use or understand the data. A later section of this report addresses in greater detail some of the reasons why measuring this aspect of student behavior is complicated and describe what is meant by some of the different measures that are available. First, however, it is worth summarizing the general picture of high school dropouts that has emerged from accumulated research. These general observations describe trends that are evident regardless of the method by which dropouts are counted.
WHO DROPS OUT
The overall rate at which students drop out of school has declined gradually in recent decades, but is currently stable. A number of student characteristics have been consistently correlated with dropping out over the past few decades. 2 First and most important, dropping out is significantly more prevalent among Hispanic and African American students, among students in poverty, among students in urban schools, among English-language learners, and among students with disabilities than among those who do not have these characteristics. The characteristics of the students most likely to drop out illustrate one of the keys to understanding the phenomenon: that dropping out is a process that may begin in the early years of elementary school, not an isolated event that occurs during the last few years of high school. The process has been described as one of gradual disengagement from school. The particular stages and influences vary widely, but the discernible pattern is an interaction among characteristics of the family and home environment and characteristics of a student's experience in school.
Family and Home Characteristics
Income In general, students at low income levels are more likely to drop out of school than are those at higher levels. NCES reports that in
2 Data in this section are taken from National Center for Education Statistics (1996, 2000), which are based on the Current Population Survey. The numbers are event dropout rates.
1999 the dropout rate for students whose families were in the lowest 20 percent of income distribution was 11 percent; for students whose families fall in the middle 60 percent it was 5 percent; and for students from families in the top 20 percent it was 2 percent.
Race/Ethnicity Both Hispanic and African American students are more likely to drop out than are white students, with the rate for Hispanic students being consistently the highest. In 1999, 28.6 percent of Hispanic students dropped out of school, compared with 12.6 percent of black students and 7.3 percent of white students. It is important to note that among Hispanic youths, the dropout rate is significantly higher for those who were not born in the United States (44.2%) than for those who were (16.1%). Two important issues relate to this last point: first, a significant number of foreign-born Hispanic young people have never been enrolled in a U.S. school. Second, the majority of those who were never enrolled have been reported as speaking English “not well” or “not at all.” The status of Hispanic young people offers an illustration of the complexities of counting dropouts. Young people who have never been enrolled in a U.S. school but have no diploma typically show up in measures of status dropout rates (people of a certain age who have no diploma) but not in measures of event dropout rates (students enrolled in one grade but not the next who have not received a diploma or been otherwise accounted for). This issue is addressed in greater detail below.
Family Structure Research has shown an increased risk of academic difficulty or dropping out for students who live in single-parent families, those from large families, and those, especially girls, who have become parents themselves. Other factors have been noted as well, such as having parents who have completed fewer years of schooling or who report providing little support for their children's education, such as providing a specific place to study and reading materials.
History of Poor Academic Performance Not surprisingly, poor grades and test scores are associated with an increased likeliness to drop out, as is enrollment in remedial courses.
Educational Engagement Researchers have used several measures of stu-
dents' educational engagement, including hours of television watched, hours spent on homework, hours spent at paid employment, and frequency of attending class without books and other necessary materials. Each of these factors has been associated with increased likeliness to encounter academic difficulties and to drop out. That is, the more time a student spends at a job or watching television, the more likely he or she is to drop out. Students who spend relatively little time on homework and who are more likely to attend school unprepared are similarly at increased risk of dropping out.
Academic Delay Students who are older than the normal range for the grade in which they are enrolled are significantly more likely to drop out of school than are those who are not. Similarly, students who have received fewer than the required number of academic credits for their grade are more likely to drop out than other students are.
Risk factors tend to cluster together and to have cumulative effects. The children of families in poverty, for example, have a greater risk of academic difficulty than do other children, and they are also at greater risk for poor health, early and unwanted pregnancies, and criminal behavior, each of which is associated with an increased risk of dropping out (National Center for Education Statistics, 1996:11). Urban schools and districts consistently report the highest dropout rates; the annual rate for all urban districts currently averages 10 percent, and in many urban districts it is much higher (Balfanz and Legters, 2001:22). Student populations in these districts are affected by the risk factors associated with dropping out, particularly poverty, in greater numbers than are students in other districts.
WHY STUDENTS DROP OUT
Students who have dropped out of school have given three common reasons (ERIC Digest, 1987:1):
- A dislike of school and a view that school is boring and not relevant to their needs;
- Low academic achievement, poor grades, or academic failure; and
- A need for money and a desire to work full-time.
These responses in no way contradict the statistical portrait of students who drop out in the United States, but they offer a somewhat different perspective from which to consider the many factors that influence students' decisions about school and work. Shifts in the labor market can have profound effects on students' behavior that are evident in national statistics, particularly those that track changes over many years. Scholars have also identified socioeconomic factors that correlate with the likelihood of a student's dropping out. However, each student whose life is captured in dropout statistics is an individual reacting to a unique set of circumstances. The circumstances that cause a particular student to separate from school before completing the requirements for a diploma can rarely be summed up easily, and rarely involve only one factor. Nevertheless, educators and policy makers alike see that dropping out of school diminishes young people's life chances in significant ways, and look for ways to understand both why they do it and how they might be prevented from doing it.
Dropping Out as a Process
Rumberger summarizes a key message from the research on the factors associated with dropping out:
Although dropping out is generally considered a status or educational outcome that can readily be measured at a particular point in time, it is more appropriately viewed as a process of disengagement that occurs over time. And warning signs for students at risk of dropping out often appear in elementary school, providing ample time to intervene (Rumberger, 2000:25).
Beginning with some points that can be difficult to discern in the complex statistics about dropping out, Rumberger noted that the percentage of young people who complete high school through an alternative to the traditional course requirements and diploma (through the GED or a vocational or other alternative) has grown: 4 percent used an alternative means in 1988 while 10 percent did so in 1998—though the calculated school completion rate among 18- to 24-year-olds remained constant at about 85 percent (Rumberger, 2000:7). Several longitudinal studies show that a much larger percentage of students than are captured in event or status dropout calculations drop out of school temporarily for one or more periods during high school. Doing so is associated with later dropping out for good, with a decreased likelihood of enrolling in postsecondary schooling, and with an increased likelihood of unemployment.
Focusing on the process that leads to the ultimate decision to drop out, Rumberger stresses the importance of interaction among a variety of contributing factors: “if many factors contribute to this phenomenon over a long period of time, it is virtually impossible to demonstrate a causal connection between any single factor and the decision to quit school” (Rumberger, 2001:4). Instead, researchers have looked for ways to organize the factors that seem to be predictive of dropping out in ways that can be useful in efforts to intervene and prevent that outcome. As noted above, two basic categories are characteristics of students, their families and their home circumstances, and characteristics of their schooling.
Rumberger pays particular attention to the concept of engagement with school. Absenteeism and discipline problems are strong predictors of dropping out, even for students not experiencing academic difficulties. More subtle indicators of disengagement from school, such as moving from school to school, negative attitude toward school, and minor discipline problems can show up as early as elementary and middle school as predictors of a subsequent decision to drop out. The role of retention in grade is very important in this context:
. . . students who were retained in grades 1 to 8 were four times more likely to drop out between grades 8 and 10 than students who were not retained, even after controlling for socioeconomic status, 8th grade school performance, and a host of background and school factors (Rumberger, 2000:15).
Rumberger's work confirms other research on family characteristics that are associated with dropping out, particularly the finding that belonging to families lower in socioeconomic status and those headed by a single parent are both risk factors for students. He also looked at research on the role that less concrete factors may play. Stronger relationships between parents and children seem to reduce the risk of dropping out, as does being the child of parents who “monitor and regulate [the child's] activities, provide emotional support, encourage decision-making . . . and are generally more involved in [the child's] schooling” (Rumberger, 2000:17).
At the workshop, David Grissmer touched on some other factors that don't make their way into national statistics but that could play a significant role for many young people. He pointed to studies of hyperactivity and attention-deficit disorder that indicate that while the percentage of all young people affected is small, roughly 5 percent, the percentage of high school dropouts affected is much larger—perhaps as much as 40 percent. He noted that dyslexia, depression, and other cognitive or mental health
problems can have significant effects on students' capacity to learn and flourish in the school environment, but that these situations are often overlooked in statistical analyses.
Schools also play a role in outcomes for students. Rumberger presented data showing that when results are controlled for students' background characteristics, dropout rates for schools still vary widely. Rumberger's (2000) review of the literature on school effects identifies several key findings:
- The social composition of the student body seems to influence student achievement—and affect the dropout rate. That is, students who attend schools with high concentrations of students with characteristics that increase their likelihood of dropping out, but who don't have those characteristics themselves, are nevertheless more likely to drop out. This finding relates to the fact that dropout rates are consistently significantly higher for urban schools and districts than for others (Balfanz and Legters, 2001:1).
- Some studies suggest that school resources can influence the dropout rate through the student-teacher ratio and possibly through teacher quality.
- The climate, policies, and practices of a school may have effects on dropping out. Indicators of the school climate, such as attendance rates and numbers of students enrolled in advanced courses, may be predictive of dropping out. There is some evidence that other factors, such as school size, structure, and governance, may also have effects.
A variety of different kinds of evidence point to the importance of early attention to the problems that are associated with subsequent dropping out. The correspondence between the many risk factors that have been enumerated is not, however, either linear or foolproof. Dynarski (2000) notes that despite strong associations between a variety of characteristics and dropping out, using individual risk factors as predictors is tricky: research that has evaluated the predictive value of risk factors has shown that the one “that was best able to predict whether middle school students were dropouts—high absenteeism—correctly identified dropouts only 16 percent of the time” (Dynarski, 2000:9).
A quantitative look at the effectiveness of dropout prevention pro-
grams can seem sobering, but it is important to bear in mind that even a perfectly successful program—one that kept every potential dropout in school—would affect only a small fraction of students. Any program that is an attempt to intervene in time to prevent dropping out must begin with a group of students who share defined risk factors, but of whom only a fraction would actually have dropped out. That is, even among groups of students with many risk factors, the dropout rate rarely goes over approximately 15 percent, and it is only these 15 of 100 students who receive an intervention whose fates could potentially be changed. When resources are limited, correctly identifying the students who will benefit most from intervention (those who are most likely to drop out) is clearly important. However, since many different kinds of factors affect dropout behavior, using them as predictors is not easy. This point is also relevant to Rumberger's point that if numerous factors contribute to a multiyear process of dropping out, isolating a cause or an effective predictor would logically be very difficult.
Though the quantitative evidence of effectiveness is not overwhelming, Dynarski (2000) used the results of a Department of Education study of the effectiveness of dropout prevention programs to provide a description of some of the strategies that seem to work best. Providing individual-level counseling to students emerged as a key tool for changing students' thinking about their education. Another tool was creating smaller school settings, even within a large school, if necessary. Students are more likely to become alienated and disengaged from school in larger settings, and are likely to receive less individualized attention from teachers and staff. 3 Not surprisingly, providing counseling and creating smaller school settings requires more staff, and, in turn, the expenditure of more resources per pupil (Dynarski, 2000).
Others who have explored the effectiveness of dropout prevention programs have come to conclusions that amplify and support Dynarski's findings. McPartland and Jordan (2001) advocate, among other things, that high schools be restructured to provide smaller school settings and to both increase student engagement with school and strengthen students' relationships with school staff. McPartland has also suggested specific supports for students who enter high school unprepared for challenging academic work,
3 The work of Lee and Burkam (2001), Fine (1987), and others on the structure of high schools is relevant to this point.
including extra time to complete courses and remediation outside of school hours.
In summary, the committee finds several important messages in the research on dropout behavior:
- A number of school-related factors, such as high concentrations of low-achieving students, and less-qualified teachers, for example, are associated with higher dropout rates. Other factors, such as small school settings and individualized attention, are associated with lower dropout rates.
- Many aspects of home life and socioeconomic status are associated with dropout behavior.
- Typically, contributing factors interact in a gradual process of disengagement from school over many years.
Conclusion: The committee concludes that identifying students with risk factors early in their careers (preschool through elementary school) and providing them with ongoing support, remediation, and counseling are likely to be the most promising means of encouraging them to stay in school. Using individual risk factors to identify likely dropouts with whom to intervene, particularly among students at the ninth-grade level and beyond, is difficult. Evidence about interventions done at this stage suggests that their effectiveness is limited.