Programs and Policies to Assist High School Dropouts in the Transition to Adulthood Programs and Policies to Assist High School Dropouts in the Transition to Adulthood Dan Bloom Summary Dan Bloom of MDRC examines policies and programs designed to help high school dropouts improve their educational attainment and labor market outcomes. So called “second-chance” programs, he says, have long provided some combination of education, training, employment, counseling, and social services. But the research record on their effectiveness is fairly thin, he says, and the results are mixed. Bloom describes eleven employment- or education-focused programs serving high school dropouts that have been rigorously evaluated over the past thirty years. Some relied heavily on paid work experience, while others focused more on job training or education. Some programs, especially those that offered paid work opportunities, generated significant increases in employ- ment or earnings in the short term, but none of the studies that followed participants for more than a couple of years found lasting improvements in economic outcomes. Nevertheless, the findings provide an important foundation on which to build. Because of the high individual and social costs of ignoring high school dropouts, the argument for investing more public funds in services, systems, and research for these young people is strong. The paucity of conclusive evidence, however, makes it hard to know how to direct resources and magnifies the importance of ensuring that all new initiatives provide for rigorous evaluation of their impacts. Bloom concludes with recommendations for policy and research aimed at building on current efforts to expand and improve effective programs for dropouts while simultaneously develop- ing and testing new approaches that might be more effective and strengthening local systems to support vulnerable young people. He stresses the importance of identifying and disseminat- ing strategies to engage young people who are more seriously disconnected and unlikely to join programs. A recurring theme is that providing young people with opportunities for paid work may be useful both as an engagement tool and as a strategy for improving long-term labor market outcomes. www.futureofchildren.org Dan Bloom is co-director of the Health and Barriers to Employment Policy Area at MDRC. VOL. 20 / NO. 1 / SPRING 2010 89 Dan Bloom The transition to adulthood is to expand and improve effective programs likely to be perilous and rocky for dropouts while simultaneously develop- for young people who drop ing and testing new approaches that might out of high school. In fact, be more effective and strengthening local sys- even those who earn a high tems to support vulnerable young people. school diploma or a General Educational Development (GED) certificate face increas- The Magnitude and Consequences ingly long odds of success if they do not go on of the Dropout Problem to get at least some postsecondary education National studies estimate that 3.5 million to or training. Young people from low-income 6 million people between the age of sixteen families are substantially less likely than and twenty-four are high school dropouts— their higher-income peers to move smoothly meaning that they have not earned a high through school, making it much more diffi- school diploma and are not now enrolled in cult for them to earn family-sustaining wages high school.1 and, potentially, to reach other adult mile- stones such as marrying. Dropouts come disproportionately from low-income and minority families. According Through a variety of school reforms begin- to the National Center for Education ning in preschool and running through high Statistics (NCES), the share of sixteen- to school, U.S. educators are working to prevent twenty-four-year-olds who are out of school young people from getting off track. For and lack a diploma or GED is 4 percent in the foreseeable future, however, the nation the highest income quartile and 17 percent in will also need “second-chance” systems and the lowest quartile. Similarly, the dropout programs to re-engage and re-direct young rate is 6 percent for whites, 12 percent for people who leave the public school system. blacks, and 20 percent for Hispanics.2 The research record on the effectiveness of Moreover, the dropout problem is heavily such programs is fairly thin and the results concentrated in a subset of high schools that are mixed, but there are some positive are themselves concentrated in large north- findings on which to build. Moreover, the ern and western cities and in the South.3 individual and social costs of neglecting this problem are potentially enormous. Experts disagree about how to calculate high school graduation rates. Surprisingly, they I begin by describing the magnitude and even disagree about whether the national consequences of the dropout problem, with dropout rate has been rising or falling in the a particular focus on the heterogeneity of the past thirty years and whether racial dispari- dropout population. Next, I describe what ties in graduation rates have been declining researchers know about the effectiveness of or growing.4 It seems clear, however, that programs designed to assist young people over this period several developments have who leave school before graduation, focusing amplified the negative consequences of drop- mainly on how the programs affect par- ping out of school. First, well-documented ticipants’ educational attainment and labor changes in the labor market have dramatically market outcomes. I conclude with some rec- reduced the availability of well-paying jobs ommendations for policy and research that for young people, particularly young men, would build on the current evidence base without postsecondary education. Adjusted 90 THE FUTURE OF CHILDREN Programs and Policies to Assist High School Dropouts in the Transition to Adulthood for inflation, the earnings of young men with volume, it is difficult to prove a causal rela- no high school diploma dropped 23 percent tionship between earnings trends and mar- between 1973 and 2006 (the earnings of riage trends. The correlations, however, are young men with only a high school degree striking. In 1970, 68 percent of male dropouts dropped about the same percentage).5 between age twenty-two and thirty-two were married; in 2007, after earnings for dropouts Even before the current recession began, had dropped precipitously, the marriage rate growing numbers of young dropouts were for this group had fallen to 26 percent. One entirely disconnected from both school and study found earnings a key predictor of mar- work. More than half of all sixteen- to nine- riage rates for young men. Similarly, although teen-year-old high school dropouts had no trends in out-of-wedlock births are affected paid employment in 2007. Declining employ- by many factors, having children outside of ment among dropouts is one symptom of a marriage is strongly correlated with educa- broader collapse in the youth labor market. tion. In 2006, a startling 67 percent of births In just eight years—from 1999 to 2007—the to female high school dropouts under age share of all sixteen- to nineteen-year-olds thirty were out-of-wedlock (by contrast, the with no paid employment during the entire out-of-wedlock birth rate was 10 percent for year rose from 44 percent to 59 percent.6 women under thirty with a master’s degree).9 Second, changes in sentencing and other Diverse Population criminal justice policies have sharply High school dropouts are a heterogeneous increased the number of young adults who group. In the first place, they leave high are incarcerated. The rate of incarceration school for many different reasons. In a 2005 in the United States stayed relatively flat survey of more than 400 dropouts, about 47 for most of the twentieth century and then percent reported that a major reason for their exploded beginning in the late 1970s. More decision to drop out was that “classes were than 2 million Americans (most of them not interesting.” Overall, 62 percent said they young men) are now in prison or jail—many were receiving grades of “C’s and above.” At for offenses that would not have led to prison the other end of the spectrum, 35 percent of terms thirty years ago.7 Spending time in respondents identified “failing in school” as prison not only strains family ties but also a major reason why they dropped out. For depresses future earnings.And high school many dropouts, the major reasons for leaving dropouts are much more likely than their school—needing to get a job (32 percent) or more educated peers to become involved to care for a family member (22 percent) or with the justice system. More than two-thirds becoming a parent (26 percent)—were not of state prison inmates have no high school directly related to school itself.10 diploma—though a substantial share has earned a GED while incarcerated.8 Second, dropouts follow different trajectories after leaving school. Most try to continue Trends in labor market conditions and incar- their education. The National Education ceration may have made it harder for high Longitudinal Study tracked a sample of young school dropouts to reach other adult mile- people who entered high school in 1988 and stones. As discussed by Sheldon Danziger were scheduled to graduate in 1992. About and David Ratner in their article in this 20 percent of the sample dropped out of high VOL. 20 / NO. 1 / SPRING 2010 91 Dan Bloom Figure 1. Status in 2000 of 100 People Who Were Eighth Graders in 1988 Lowest socioeconomic status Highest socioeconomic status 100 100 8th graders 8th graders in 1988 in 1988 38 8 dropped out of dropped out of high school high school at least once at least once 62 16 22 1 7 92 did not late high school no high school no high school late high school did not drop out of diploma (3) diploma or GED diploma or GED diploma (1) drop out of high school or GED (13) by 2000 by 2000 or GED (6) high school by 2000 by 2000 Source: Authors’ calculations based on Cheryl Almeida, Cassius Johnson, and Adria Steinberg, Making Good on a Promise: What Policymakers Can Do to Support the Educational Persistence of Dropouts (Boston: Jobs for the Future, 2006). school at least once. By 2000, eight years after Third, young people who leave school and their scheduled graduation date, nearly two then become disconnected face a variety thirds (63 percent) had earned a high school of personal and situational obstacles. For diploma or, much more commonly, a GED example, a recent study focusing on New certificate, and 43 percent had attended a York City identified five overlapping groups postsecondary institution. Presumably, more of young people who are at particularly high than 63 percent of the dropouts attempted to risk of leaving school, not returning, and then continue their education.11 ending up unemployed or out of the labor force: immigrant youth, young people with As figure 1 shows, however, the above data disabilities (learning disabilities or emotional mask huge differences by socioeconomic and behavioral issues), young people involved status. As noted, higher-income students in the justice system, youth aging out of fos- are much less likely to drop out in the first ter care, and young mothers.13 place. And almost all of the higher-income dropouts earned a GED or diploma by Programming for Dropouts 2000. By comparison, less than half of the Second-chance programs have long offered low-income dropouts had a credential by opportunities for young people who leave the 2000. Similarly, among those in the highest K–12 education system without earning a socioeconomic group, 67 percent of those diploma. Ranging from large national pro- who had a diploma or GED had enrolled in grams or networks like the Job Corps (more college, compared with 29 percent of those in than 100 sites nationwide) and YouthBuild the lowest group (numbers not shown in the (more than 200 programs) to small indepen- figure).12 Young people from higher-income dent programs run by churches or communi- families who drop out of school are often ty-based organizations, these programs able to get back on track, while their lower- typically provide some combination of income peers are more likely to flounder. education, training, employment, counseling, 92 THE FUTURE OF CHILDREN Programs and Policies to Assist High School Dropouts in the Transition to Adulthood and social services. Some, like the Job Corps, network of programs known for serving drop- have dedicated streams of federal funding, outs and targeting the GED, now includes while others piece together funding from the twenty-nine diploma-granting schools. federal Workforce Investment Act of 1998 (WIA) and other state and local sources. Moreover, although second-chance programs Many target specific subsets of youth, such as once viewed the GED credential as the those with disabilities or those in the foster ultimate goal, their aim now is increasingly care or juvenile justice systems, reflecting the to help former dropouts obtain postsecond- availability of targeted funding for those ary education, which has become a virtual groups. High school dropouts are typically prerequisite for admission to the middle overrepresented among these vulnerable class. The Gateway to College program, populations, which are discussed by D. developed at Portland (Oregon) Community Wayne Osgood, E. Michael Foster, and Mark College and now operating in twenty-three E. Courtney in their article in this volume. other community colleges across the country, (Other programs for dropouts have broader gives high school dropouts a chance to attend eligibility criteria, but end up serving many high school and college simultaneously. young people from these same vulnerable Gateway students begin as a group during groups.) the Foundation Term, during which they strengthen basic academic skills and adjust to Not so long ago, second-chance programs college. Those who succeed in these courses that helped dropouts earn the GED creden- gradually move into the regular college cur- tial were fairly clearly differentiated from riculum. Gateway relies on “average daily traditional high schools. Today the landscape attendance” funding from the K–12 system is far more varied. On the one hand, many even though students are enrolled at a com- school districts are developing “multiple munity college. pathways” initiatives that offer a wider range of high school options in an effort to pre- In addition, the Bill and Melinda Gates vent young people from leaving the K–12 Foundation has given grants to YouthBuild system. For example, New York City’s Office and other community-based youth employ- of Multiple Pathways has created Transfer ment programs to help them build pathways Schools (small schools specifically designed to postsecondary education for their partici- for students who have fallen behind for their pants. These college-focused efforts appear age or who have dropped out) and Young to be growing, though they are likely to serve Adult Borough Centers (evening programs a subset of dropouts with stronger academic for older students operated by partnerships skills. between schools and community organiza- tions), along with new GED programs, all Beyond specific programs, some cities are of which are supplemented by Learning to working to develop coherent youth “systems” Work, a program that provides job readiness to improve coordination among the many instruction and paid internships. On the other programs that serve specific subsets of dis- hand, some community-based programs with advantaged youth or provide a narrow range experience serving dropouts now operate of services using separate, targeted funding charter schools or alternative high schools. streams. Without a single agency or entity to For example, YouthBuild, a large national take responsibility for ensuring that young VOL. 20 / NO. 1 / SPRING 2010 93 Dan Bloom people with few family supports make a suc- Although the programs and studies can be cessful transition to adulthood, many youth categorized in many ways, table 1 groups will fall through the cracks. The National them according to their primary service League of Cities has profiled efforts by sev- approach. The first three programs—the eral cities to build collaboration across many National Supported Work Demonstration, public systems, including law enforcement, the Youth Incentive Entitlement Pilot education, workforce development, and child program (YIEPP), and the Conservation welfare, to better serve disconnected youth.14 and Youth Service Corps—relied heavily on paid work experience, while the next Program Effectiveness six—JOBSTART, the National Job Training Most programs that target high school Partnership Act, New Chance, the Center dropouts have never been formally evaluated for Employment and Training (CET), for effectiveness. Moreover, because the Job Corps, and National Guard Youth programs are often run by small community- ChalleNGe—focused more on job training or based organizations, the most rigorous education. The last two—the Teenage Parent evaluation methods are probably not feasible Demonstration and the Learning, Earning, or appropriate in many cases. The result is a and Parenting program (LEAP)—were gap between the strongly held views of mandatory, welfare-based programs that practitioners who believe they know what encouraged, supported, or required teenage constitutes “best practice” in youth program- mothers to work or go to school. ming, on the one hand, and the knowledge base researchers have built from rigorous This classification scheme is useful in evaluations, on the other. understanding the broad patterns of pro- gram effects, but it is far from perfect. For Table 1 describes eleven rigorous evalua- example, two of the work programs include tions of employment- or education-focused a strong emphasis on education, and some of programs serving high school dropouts that the training programs provide work experi- have been conducted over the past thirty ences of some kind. More important, the years (a few of the programs served both simple categorization does not capture criti- dropouts and in-school youth).15 The table cal factors such as the program atmosphere focuses on major studies that used random- or the types of ancillary services, supports, assignment designs, in which eligible youth and activities provided to participants. were placed, through a lottery-like process, either in a program group that had access Overall, the evaluations tell a mixed story. In to the program being studied or in a con- several, young people in the program group trol group that did not.16 The table does not were substantially more likely than their include some specialized programs that were control group counterparts to earn a GED or rigorously evaluated and may serve some another credential. For example, in the Job dropouts, such as Multi-Systemic Therapy Corps evaluation, 42 percent of the program (a treatment approach for youth with serious group earned a GED within four years after behavior problems), or Structured Training entering the study, compared with 27 percent and Employment Transitional Services (an of the control group. Similarly, 38 percent employment model for young adults with of the program group earned a vocational developmental disabilities).17 or trade certificate, compared with only 15 94 THE FUTURE OF CHILDREN Programs and Policies to Assist High School Dropouts in the Transition to Adulthood Table 1. Selected Rigorous Evaluations of Programs for High School Dropouts Sample size (number of Evaluation (dates) Target group Program model sites) Summary of results Work programs National Supported Work 17- to 20-year-old high Paid work experience, with 861 youth Large increases in employment Demonstration (1976–81) school dropouts (one of graduated stress (5 sites) initially, but no lasting impacts for four target groups) youth target group Youth Incentive Entitlement 16- to 19-year-olds from Guaranteed part-time and 82,000 youth Large, short-term increases in Pilot Projects (1977–81) low-income families summer jobs conditioned (17 sites) employment; no impacts on who had not graduated on school attendance school outcomes from high school American Conservation Mostly 18- to 25-year- Paid work experience in 1,009 youth Increases in employment and and Youth Service Corps old out-of-school youth community service projects; (4 sites) decreases in arrests, particularly (1993–96) education and training; sup- for African American males; short port services follow-up Education and training programs JOBSTART (1985–93) 17- to 21-year-old high Education, training, support 2,300 youth Increases in GED receipt; few school dropouts with services, job placement (13 sites) impacts on labor market out- low reading levels assistance comes (except in CET site) National Job Training Disadvantaged 16- to Education, job skills train- 5,690 youth No earnings impacts for females Partnership Act (out-of- 21-year-old out-of- ing, job placement, on-the- (16 sites) or male non-arrestees, possibly school youth analysis) school youth job training, and support negative impacts for male (1987–94) services arrestees New Chance (1989–92) 16- to 22-year-old teen- Wide range of education, 2,000 youth Increases in GED receipt; age mothers who were employment, and family (16 sites) no impacts on labor market high school dropouts services outcomes Center for Employment Disadvantaged, out-of- Education and vocational 1,500 youth Few impacts on employment and Training (CET) Replication school youth, ages 16 training (12 sites) earnings overall; some impacts (1995–99) to 21 for younger youth Job Corps (1994–2003) Disadvantaged youth, Employment, education, 15,386 youth Earnings and employment ages 16 to 24 and training in a (mostly) (nationwide) impacts in years 3–4 of study residential setting period; impacts faded after year 4 according to administrative data. Results appear stronger for older youth (20 to 24 years old) National Guard Youth High school dropouts, Education, service to 3,000 youth Early results show large increases ChalleNGe (2005–present) ages 16 to 18, who community, and other (10 sites in diploma or GED receipt and are drug free and not components in a quasi- nationwide) smaller gains in employment, heavily involved with the military residential setting; college enrollment, and other justice system 12-month post-residential outcomes mentoring program Mandatory welfare-based programs Teenage Parent Teenage parents receiv- Mandatory education, 6,000 youth One of three programs increased Demonstration ing welfare training, and employment- (3 sites) high school graduation; increases (1987–91) related services; support in employment and earnings services (case manage- ment, workshops, etc.) Ohio Learning, Earning, Teen mothers under age Financial incentives and 7,017 teens Increases in GED receipt and and Parenting Program 20 who are on welfare sanctions based on school (12 Ohio some earnings gains for initially (LEAP)(1989–97) and do not have a GED enrollment and attendance counties) enrolled teens or high school diploma Sources: Rebecca Maynard, The Impact of Supported Work on Young School Dropouts (New York: MDRC, 1980); Judith Gueron, Lessons from a Job Guarantee: The Youth Incentive Entitlement Pilot Projects (New York: MDRC, 1984); JoAnn Jastrzab and others, Impacts of Service: Final Report on the Evaluation of the American Conservation and Youth Service Corps (Cambridge, Mass.: Abt Associates, 1996); George Cave and others, JOBSTART: Final Report on a Program for School Dropouts (New York: MDRC, 1993); Larry Orr and oth- ers, Does Training for the Disadvantaged Work? Evidence from the National JTPA Study (Cambridge, Mass.: Abt Associates, 1997); Janet Quint, Johannes Bos, and Denise Polit, New Chance: Final Report on a Comprehensive Program for Young Mothers in Poverty and Their Children (New York: MDRC, 1997); Cynthia Miller and others, The Challenge of Repeating Success in a Changing World: Final Report on the Center for Employment Training Replication Sites (New York: MDRC, 2005); Peter Schochet, John Burghardt, and Sheena McConnell, “Does Job Corps Work? Impact Findings from the National Job Corps Study,” American Economic Review 98, no. 5 (December 2008); Dan Bloom and Megan Millenky, 21-Month Results from the National Guard Youth ChalleNGe Program Evaluation (New York: MDRC, forthcoming); Ellen Eliason Kisker and others, Moving Teenage Parents to Self-Sufficiency (Princeton: Mathematica Policy Research, 1998); Johannes Bos and Veronica Fellerath, Final Report on Ohio’s Welfare Initiative to Improve School Attendance Among Teenage Parents (New York: MDRC, 1997). VOL. 20 / NO. 1 / SPRING 2010 95 Dan Bloom Table 2. Interim Results from the National Guard Youth ChalleNGe Evaluation Program group Control group Outcome (percentage with outcome) (percentage with outcome) Difference Educational attainment High school diploma or GED 60.5 36.4 24.1*** High school diploma 22.0 16.3 5.7*** GED credential 48.3 21.9 26.5*** Any college credit 24.8 9.6 15.1*** Current activities Attending high school or GED prep 16.1 26.0 –9.9*** Taking college courses 11.6 7.0 4.6*** Working for pay 55.0 50.1 4.9* Serving in military 10.9 6.2 4.7*** High school diploma or GED and in college, 45.5 23.1 22.4*** training, work, or the military Source: MDRC analysis of the National Guard Youth Challenge Evaluation 21-month survey. Asterisks indicate differences that are statis- tically significant, meaning that they are very unlikely to arise by chance. Differences marked with three asterisks are significant at the 1 percent level, those marked with two asterisks are significant at the 5 percent level, and those marked with one asterisk are significant at the 10 percent level. The 1 percent level denotes the highest degree of confidence that the program actually had an impact. percent of the control group. Interim results agreed to attend school, employed 76,000 from the National Guard Youth ChalleNGe youth and virtually erased the large gap evaluation show that about 61 percent of the between the unemployment rates for white program group and 36 percent of the con- and black youth. The Conservation and Youth trol group earned a GED or diploma within Service Corps also provided subsidized jobs twenty-one months after study enrollment and generated some statistically significant (see table 2). The JOBSTART and New increases in employment outcomes, par- Chance studies reported similar findings. ticularly for African American males, over a relatively short follow-up period. Some of the programs, especially those that offered paid work opportunities, also The Job Corps program did not rely on generated significant increases in employ- subsidized jobs but still managed to increase ment or earnings in the short term. For employment and earnings in the third and example, in the National Supported Work fourth years of the study period—and even Demonstration, which provided subsidized longer for older participants (aged twenty to (paid) jobs for up to twelve to eighteen twenty-four at enrollment). Similarly, as shown months to dropouts aged seventeen to in table 2, the National Guard ChalleNGe twenty, the difference in employment rates evaluation found that program group mem- between the program and control groups bers were modestly more likely than their was as high as 68 percentage points early in control group counterparts to be employed the follow-up period. Similarly, the Youth twenty-one months after entering the study. Entitlement project, which guaranteed part- time and summer jobs to all disadvantaged The gains in credentials and short-term earn- young people in certain geographic areas who ings are notable, but none of the studies that 96 THE FUTURE OF CHILDREN Programs and Policies to Assist High School Dropouts in the Transition to Adulthood followed participants for more than a couple for African American study participants. For of years found lasting improvements in eco- example, African American females were less nomic outcomes. Some of the studies (YIEPP likely to become pregnant and African and Conservation and Youth Service Corps) American males improved in measures of did not report or collect long-term data or personal and social responsibility. Few of the are still ongoing (ChalleNGe). In other cases, other programs generated impacts on these early effects faded over time. For example, non-economic measures. the Job Corps evaluation found that increases in employment and earnings faded by year Overall, these findings do not support the five and did not reappear (though, as noted, common perception that “nothing works” for earnings gains persisted for study participants high school dropouts. Many of the positive who were aged twenty to twenty-four when effects produced by the programs, however, they enrolled).18 were modest or relatively short-lived. Moreover, the studies suggest that even some JOBSTART, which operated in thirteen sites, of the relatively successful programs may showed no significant earnings gains overall have difficulty meeting a strict benefit-cost during a four-year follow-up period, but the test. The authors of the Job Corps evaluation study measured large impacts in one site, concluded that the benefits produced by the the Center for Employment and Training in program probably exceeded its costs (about San Jose, California. However, as shown in $16,500 per participant) for older partici- table 1, when CET was replicated in twelve pants, but not for the full study sample. sites during the 1990s, an evaluation found no Nevertheless, the findings provide an impor- increases in earnings over a fifty-four-month tant foundation on which to build. follow-up period (women at the program sites that most faithfully implemented the model One important study is not included in table made shorter-term earnings gains, but these 1 because it targeted in-school youth, but the gains faded after year three). findings may be relevant to the topic dis- cussed here. A random-assignment evaluation Several of the studies measured non-economic of Career Academies, a high school-based outcomes such as crime involvement, drug model, found that it produced statistically sig- use, health, and psychosocial development— nificant increases in earnings over an eight- again, with mixed results. Partway through year follow-up period. Men in the program the evaluation’s follow-up period, the group earned about $30,000 more than their National Guard ChalleNGe program has control group counterparts over the eight produced modest decreases in crime convic- years even though they were no more likely tions and improvements in some measures of to graduate from high school or go to college. psychosocial development. The Job Corps The researchers suggest that the program’s significantly reduced arrests, convictions, and use of “career awareness and development time spent incarcerated over the first four activities,” including job shadowing and work- years of the study period (these outcomes based learning activities, may have contrib- were not measured after the four-year point). uted to the earnings gains. The Conservation and Youth Service Corps reduced arrests overall and had a range of Perhaps most interesting, the Career positive effects on non-economic outcomes Academies produced significant effects on VOL. 20 / NO. 1 / SPRING 2010 97 Dan Bloom several adult transition milestones. At the as inoculations, whose effects may last for- end of the follow-up period, program group ever, or as vitamins, whose effects wear off if members were more likely to be living they are not taken consistently. independently with children and a spouse or partner, and young men in the program group were more likely to be married and to Some experts have raised be custodial parents.19 These findings sug- gest that improving young people’s economic the question of whether it is prospects may ease their transition into other more appropriate to think adult roles. of time-limited programs for What Conclusions Can Be Drawn? dropouts as inoculations, It is difficult to draw cross-cutting lessons from the evaluations in table 1 because there whose effects may last forever, are many programs and not many unambig- or as vitamins, whose effects uously positive results. For example, the data do not support clear conclusions about wear off if they are not taken whether paid work, a residential structure, consistently. or other program design elements are associated with more positive results in random-assignment studies. It is possible, however, to make a few general points. Second, it is important to note that almost all the programs (and the control groups as First, although sustained positive effects well) involved youth who had volunteered to would obviously be preferable, short-term participate—and who thus had at least some effects are not unimportant. When programs motivation to change their lives. In fact, some achieve short-term increases in earnings of the programs extensively screened appli- or other outcomes, those effects are not cants and accepted only those who demon- erased if the program and control groups strated strong motivation and commitment. have similar outcomes later. Although many Thus the young people who ended up in the programs assert that they can alter the long- control groups likely sought out other pro- term trajectories of their participants, it is grams in the community and received some worth considering whether it is reasonable of the same kinds of services that program to expect even the strongest youth programs group youth received. The study results could to produce effects that can still be measured thus be interpreted to mean that the tested many years later. Results like those achieved programs did not do much better than other in the Career Academies evaluation, where programs in their communities, but that all earnings gains persisted eight years after of the programs were relatively effective for students had completed high school, are very motivated participants. That said, most of the rare—and the Academies that were tested evaluations also found that outcomes were did not serve a highly disadvantaged group of relatively poor for both research groups. For young people. Some experts have raised the example, in the Job Corps study, the aver- question of whether it is more appropriate to age employed sample member earned only think of time-limited programs for dropouts about $10,000 a year during the later years 98 THE FUTURE OF CHILDREN