Once upon a time, a town was having a serious health problem. Approximately 30 percent of its children were coming down with typhoid and other diseases because of contaminated drinking water. The town council allocated millions to medical care for the victims, yet some of them died or were permanently disabled. One day, an engineer proposed to the town council that they install a water treatment plant, which would prevent virtually all cases of the disease. “Ridiculous!” fumed the mayor. “We can't afford it!”
The engineer pointed out that they were already paying millions for treatment of a preventable disease.
“But if we bought a water treatment plant,” the mayor responded, “how could we afford to treat the children who already have the disease?”
“Besides,” added a councilman, “most of our children don't get the disease. The money we spend now is targeted to exactly the children who need it!” After a brief debate, the town council rejected the engineer's suggestion.
The town council's decision in this parable is, of course, a foolish one. From a purely economic point of view, the costs of providing medical services to large numbers of children over a long time were greater than the cost of the water treatment plant. More important, children were being permanently damaged by a preventable disease.
In education, we have policies that are all too much like those of the foolish town council. A substantial number of children fail to learn to read adequately in the early grades. Many are retained, assigned to special education, or maintained for many years in remedial programs. The financial costs of providing long-term remedial services after a student has already failed are staggering, but even more tragic are the consequences for individual children who fail so early.
Despite some improvements and a growing acceptance of the idea that prevention and early intervention are preferable to remediation, programs (and funding) for at-risk students overwhelmingly emphasize remediation. The unspoken assumption behind such policies is that substantial numbers of students—due to low IQs, impoverished family backgrounds, or other factors—are unlikely to be able to keep up with their classmates and will therefore need long-term support services to keep them from falling further behind.
If early school failure were, in fact, unavoidable, we might have a rationale for continuing with the policies we have now. But a growing body of evidence refutes the proposition that school failure is inevitable for any but the most retarded children. Further, the programs and practices that, either alone or in combination, have the strongest evidence of effectiveness for preventing school failure for virtually all students are currently available and replicable. None of them is exotic or radical.
In the following pages, we summarize the conclusions of our major, federally funded review (Slavin et al., in press) on the effects of programs intended to prevent early school failure. Our review focused on a variety of indicators of success and failure. Most early intervention programs involving students from birth to age 4 have used IQ, language proficiency, and other measures that predict school success and their outcomes. We reported these outcomes, but placed greater emphasis on measures of actual school success or failure: reading performance, retention, and placement in special education. Whenever possible, we emphasized long-term effects of early interventions.
We reviewed several types of early schooling programs. One important common feature of these programs is that they are expensive, and most are of similar orders of magnitude of cost. For example, reducing class size by half (from 30 to 15, for instance) involves hiring an additional certified teacher for each class. Yet this same teacher could teach a preschool class, could be added to the kindergarten staff to enable a school to have full-day kindergarten, or could tutor about 15 low-achieving 1st graders 20 minutes per day. Retention or provision of extra-year programs for kindergartners or 1st graders adds about $4,000, or one year's per-pupil cost per child. The costs of Writing to Read and other integrated computer-assisted-instruction programs require at least one additional aide per school plus initial and continuing costs roughly comparable to the cost of additional certified teachers. The popularity of all these programs indicates that we are willing to spend money to prevent early school failure, but which investments pay off?
The Reading Link
The consequences of failing to learn to read in the early grades are severe. Longitudinal studies find that disadvantaged 3rd graders who have failed one or more grades and are reading below grade level are extremely unlikely to complete high school (Lloyd 1978, Kelly et al. 1964). Remedial programs such as Chapter 1 have few if any effects on students above the 3rd grade level (see Kennedy et al. 1986). Many children are referred to special education programs largely on the basis of reading failure, and then remain in special education for many years, often for their entire school careers.
Almost all children, regardless of social class or other factors, enter 1st grade full of enthusiasm, motivation, and self-confidence, fully expecting to succeed in school. By the end of 1st grade, many of these students have already discovered that their initial high expectations are not coming true, and they have begun to see school as punishing and demeaning. Trying to remediate reading failure later on is very difficult because students who have failed are likely to be unmotivated, with poor self-concepts as learners. They are anxious about reading, and they hate it. Reform is needed at all levels of education, but no goal of reform is as important as seeing that all children start off their school careers with success, confidence, and a firm foundation in reading.
Success in the early grades does not guarantee success throughout the school years and beyond, but failure in the early grades does virtually guarantee failure in later schooling. If there is a chance to prevent the negative spiral that begins with early reading failure from the start, then it seems necessary to do so. Even very expensive early interventions can be justified on cost-effectiveness grounds alone if they reduce the need for later and continuing remedial and special education services, retentions, and other costs (Barnett and Escobar 1987). While the cost-effectiveness estimates associated with the Perry Preschool Model (Berrueta-Clement et al. 1984) have been criticized as unrealistic by many researchers (see Holden 1990), they have contributed to a widespread acceptance of the idea that early intervention, even if expensive, ultimately pays back its costs. Given, then, that there is growing agreement on the proposition that investments in early intervention are worthwhile, we must turn again to the question of which forms of early intervention are likely to have the greatest impact.
Birth to Age 3 Interventions
Both child-centered and family-centered interventions with at-risk children can make a substantial and, in many cases, lasting difference in their IQ scores (Wasik and Karweit, in press). In child-based interventions, infants and toddlers are placed in stimulating, developmentally appropriate settings for some portion of the day. Family-centered interventions provide parents with training and materials to help them stimulate their children's cognitive development, to help them with discipline and health problems, and to help them with their own vocational and home management skills.
The IQ effects of the birth-to-3 programs were mostly seen immediately after the interventions were implemented, but longer-lasting effects were found in a few cases. The extremely intensive Milwaukee Project (Garber 1988) found the largest long-lasting effects. It provided 35 hours per week of infant stimulation, including one-on-one interaction with trained caregivers followed by high-quality preschool. Parent training and vocational skills training were also included. At age 10, the children (of mildly retarded mothers) had IQs like those of low-risk children, and their IQs were substantially higher than those of a randomly selected control group of at-risk children. As the children reached the 4th grade, they were reading a half year ahead of the control group, and special education referrals were also reduced.
A study of the Gordon Parent Education Program (Jester and Guinagh 1983), which provided impoverished parents with intensive training in child stimulation, found that at age 10, children who had been in the program at least two years still had higher IQs than did a randomly selected control group. Also, they had fewer than half as many special education placements (23 percent vs. 53 percent).
The Carolina Abecedarian Project provided at-risk children with intensive infant stimulation and preschool programs seven hours a day for at least five years, along with services to families. A longitudinal study (Ramey and Campbell 1984) found that K–2 children in the program had higher IQs and fewer retentions than similar control students.
The studies of birth-to-3 interventions demonstrate that IQ is not a fixed attribute. It can be modified by changing a child's environment at home and/or in special center-based programs. Birth-to-3 interventions can also influence special education referrals and retention. It apparently takes intensive intervention over a period of several years to produce lasting effects on measures of cognitive functioning, but even the least intensive models, which often produced strong immediate effects, may be valuable starting points for an integrated combination of age-appropriate preventative approaches over the child's early years.
Impact of Preschool
When compared to similar children who do not attend preschool, those who did attend have been found to have higher IQ and language proficiency scores immediately following the preschool experience, although follow-up assessments typically find that these gains do not last beyond the early elementary years at most (see Karweit, in press a; McKey et al. 1985). In addition, little evidence indicates that preschool experience has any effect on elementary reading performance.
Several studies do show that the most important lasting benefits of preschool are on other outcomes such as retention and placement in special education. Preschool has also been found to have a very long-term impact on dropouts, delinquency, and other behaviors (Berrueta-Clement et al. 1984). It may be that the effects of preschool on outcomes for teenagers are due to the shorter-term effects on retention and special education placements in the elementary grade. Retention and special education placement in elementary school have been found to be strongly related to dropping out of high school (Lloyd 1978).
Clearly, attendance at a high-quality preschool program has long-term benefits for children, but it is equally clear that preschool experience is not enough to prevent early school failure, particularly because we find little evidence to support preschool effects on student reading performance. Preschool experiences for 4-year-olds should be part of a comprehensive approach to prevention and early intervention, but a one-year program, whatever its quality, cannot be expected to solve all the problems of at-risk children.
The Kindergarten Question
Since the great majority of children now attend kindergarten or other structured programs for 5-year-olds, the main questions about kindergarten in recent years have focused on full-day vs. half-day programs and on effects of particular instructional models. Research comparing full- and half-day programs generally finds positive effects of full-day programs on end-of-year measures of reading readiness, language, and other objectives. However, the few studies that have examined maintenance of full-day kindergarten effects have failed to find evidence of maintenance even at the end of 1st grade (see Karweit, in press b).
Several specific kindergarten models were found to be effective on end-of-kindergarten assessments. Among these were Alphaphonics, Early Prevention of School Failure, and TALK. These are all structured, sequenced approaches to building pre-reading and language skills, which are thought to be important predictors of success in 1st grade. However, only Alphaphonics presented evidence of long-term effects on student reading performance (Karweit, in press b). IBM's Writing to Read computer program has had small positive effects on end-of-kindergarten measures, but longitudinal studies have failed to show any carryover to 1st or 2nd grade reading (Freyd and Lytle 1990, Slavin 1991).
Retention, Developmental Kindergarten, and Transitional 1st Grade
Many schools attempt in one form or another to identify young children who are at risk for school failure and give them an additional year before 2nd grade to catch up with grade-level expectations. Students who perform poorly in kindergarten or 1st grade may simply be retained and recycled through the same grade. Alternatively, students who appear to be developmentally immature may be assigned to a two-year “developmental kindergarten” or “junior kindergarten” sequence before entering 1st grade. Many schools have a “transitional 1st grade” or “pre-1st” program to provide a year between kindergarten and 1st grade for children who appear to be at risk.
Interpreting studies of retention and early extra-year programs is difficult. Among other problems, it is unclear what the appropriate comparison group should be. Should a student who attended 1st grade twice be compared to 2nd graders (his or her original classmates) or 1st graders (his or her new classmates)?
Studies comparing students who experienced an extra year of school before 2nd grade have generally found that these students appear to gain on achievement tests in comparison to their same-grade classmates but not in comparison to their agemates. Further, any positive effects of extra-year programs seen in the year following the retention or program participation consistently wash out in later years (Karweit and Wasik, in press; Shepard and Smith 1989). Clearly, the experience of spending another year in school before 2nd grade has no long-term benefits. In contrast, studies of students who have been retained before 3rd grade find that controlling for their achievement, such students are far more likely than similar nonretained students to drop out of school (Lloyd 1978).
Class Size and Instructional Aides
A popular policy in recent years has been to markedly reduce class size in the early elementary grades. Because it is so politically popular and straightforward (albeit expensive) to implement, class size reduction should in a sense be the standard against which all similarly expensive innovations should be judged.
Decades of research on class size have established that small reductions in class size (for example, from 25 to 20) have few if any effects on student achievement. However, research has held out the possibility that larger reductions (for example, from 25 to 15) may have a meaningful impact (see Slavin, in press).
The largest and best-controlled study ever done on this question was a recent statewide evaluation in Tennessee (Word et al. 1990). Kindergartners were randomly assigned to classes of 15, 25 with an aide, or 25 with no aide, and they maintained the same configurations through the 3rd grade. This study found moderate effects in favor of the small classes as of the 3rd grade. A year after the study, this difference was still positive but very small (Nye et al. 1991). Other statewide studies of class size reduction in the 1st grade in South Carolina (Johnson and Garcia-Quintana 1978) and Indiana (Farr et al. 1987) found even smaller effects of substantial reductions in class size.
The Tennessee class size study also evaluated the effects of providing instructional aides to classes of 25 in grades K–3. The effects of the aides were near zero in all years (Folger and Breda 1990). This is consistent with the conclusions of an earlier review by Schuetz (1980). However, there is evidence, cited below, that aides can be effective in providing one-to-one tutoring to at-risk 1st graders. Reducing class size may be part of an overall strategy for getting students off to a good start in school, but it is clearly not an adequate intervention in itself.
Nongraded Primary Programs
The nongraded primary is a form of school organization in which students are flexibly regrouped according to skill levels across grade lines and proceed through a hierarchy of skills at their own pace (Goodlad and Anderson 1963). This was an innovation of the 1950s and '60s that is making a comeback in the 1990s.
Research from the first wave of implementation of nongraded primary schools supports the use of simple forms of this strategy but not complex ones. In simple forms, students are regrouped across grade lines for instruction (especially in reading and mathematics) and are taught in groups. Such programs primarily allow teachers to accommodate individual needs without requiring students to do a great deal of seatwork (as is necessary in traditional reading groups, for example). In contrast, complex forms of the nongraded primary—which make extensive use of individualized instruction, learning stations, and open space—are generally ineffective in increasing student achievement (Gutiérrez and Slavin 1992).
One-to-One Tutoring
Of all the strategies reviewed in this article, the most effective by far for preventing early reading failure are approaches incorporating one-to-one tutoring of at-risk 1st graders. Wasik and Slavin (1990) reviewed research on five specific tutoring models. One of these, the model used in Success for All, is discussed below. In addition, Reading Recovery (Pinnell et al. 1988) and Prevention of Learning Disabilities (Silver and Hagin 1990) use certified teachers as tutors. The Wallach tutoring program (Wallach and Wallach 1976) and Programmed Tutorial Reading (Ellison et al. 1968) use paraprofessionals and are correspondingly much more prescribed and scripted.
The immediate reading outcomes for all forms of tutoring are very positive, but the largest and longest-lasting effects have been found for the three programs that use teachers as tutors. Reading Recovery is a highly structured model requiring a year of training and feedback. It emphasizes direct teaching of metacognitive strategies, “learning to read by reading,” teaching of phonics in the context of students' reading, and integration of reading and writing. Two follow-up studies of this program have found that strong positive effects seen at the end of 1st grade are maintained into 2nd and 3rd grade. Effects on reducing retentions were found in 2nd grade in one study, but these effects had mostly washed out by 3rd grade.
Prevention of Learning Disabilities focuses on remediating specific perceptual deficits as well as improving reading skill. It usually operates for two school years (whereas Reading Recovery rarely goes beyond 1st grade). Reading effects of this program were substantial in two of three studies at the end of the program and remained very large as of the end of 3rd grade in one follow-up study.
Improving Curriculum and Instruction
One strategy for enhancing early reading performance is, of course, improving curriculum and instruction in the early grades. All of the tutoring programs cited above used a particular curriculum and set of instructional methods, and it is therefore impossible to separate the unique effects of tutoring from those of the materials and procedures used. Further, any comprehensive approach to prevention and early intervention must include an effective approach to curriculum and instruction in beginning reading.
We do not intend in this article to take on the current controversy about appropriate instruction in beginning reading. We generally agree with the conclusions reached by Adams (1990, p. 416) in a comprehensive, federally mandated review on the topic: In summary, deep and thorough knowledge of letters, spelling patterns, and words, and of the phonological translations of all three, are of inescapable importance to both skillful reading and its acquisition. By extension, instruction designed to develop children's sensitivity to spellings and their relations to pronunciations should be of paramount importance in the development of reading skills. This is, of course, precisely what is intended of good phonic instruction.
Adams goes on to define “good phonic instruction” as instruction that teaches word attack skills in the context of meaning, not in isolation from real reading.
The practice and theory of beginning reading are changing so rapidly at present that this is a poor time to make recommendations about appropriate practice. At the moment, very little evidence supports any of the new “whole language” approaches in 1st grade beginning reading (see, for example, Stahl and Miller 1989), but such evidence may develop as these programs gain in sophistication and use.
Success for All
Each of the strategies presented above has focused on one slice of the at-risk child's life. While the birth to age 3 and preschool programs have often integrated services to children with services to parents, the programs for older youngsters often focus only on academics and, in most cases, only one aspect of the academic program such as class size, length of day, grouping, or tutoring in reading.
How much could school failure be prevented if at-risk children were provided with a coordinated set of interventions over the years designed to prevent learning problems from developing in the first place and intervening intensively and effectively when they do occur? This is the question posed in research on Success for All (Madden et al. 1991), which is designed to provide children with whatever programs and resources they need to succeed throughout their elementary years.
Success for All emphasizes prevention and early intervention. Prevention includes the provision of high-quality preschool and/or full-day kindergarten programs; research-based curriculum and instructional methods in all grades, preschool to grade 5; reduced class size and nongraded organization in reading; activities to build positive relationships and involvement with parents; and other elements. Early intervention includes one-to-one tutoring in reading from certified teachers for students who are beginning to fall behind in 1st grade and family support programs to solve truancy, behavior problems, emotional difficulties, or health or social service challenges. In essence, Success for All combines the most effective interventions identified in this article and adds to them extensive staff development in curriculum and instruction and a school organizational plan to flexibly use resources to see that students read, stay out of special education, and are promoted each year.
Research on Success for All has found substantial positive effects on the reading performance of all students in grades 1–3, and on reductions in retentions and special education placements (Slavin et al. 1992). The lasting effects of Success for All into 3rd grade are the largest of any of the strategies reviewed in this article, but they cannot be interpreted as maintenance assessments, as the program continues through the elementary grades. However, with few exceptions, the program beyond the 1st grade consists of improved curriculum, instruction, and family support services, not continued tutoring.
Consistent Patterns
We see a consistent pattern in most of the programs and practices in our review. Whatever their nature, preventative programs tend to have their greatest impacts on outcomes closely aligned with the intervention and in the years immediately following the intervention period. The long-term research on effects of preschool on dropout and related variables is one exception to this, but on measures of IQ, reading, special education placements, and retention, preschool effects were like those of other time-limited interventions. The positive effects seen on these variables were strongest immediately after the program and then faded over time.
Some might take the observation that effects of early interventions often fade in later years as an indication that early intervention is ultimately futile. Such a conclusion would be too broad. What research on early intervention suggests is that there is no magic bullet, no program that, administered for one or two years, will ensure the success of at-risk children throughout their school careers and beyond. However, it is equally clear that children must successfully negotiate key developmental hurdles in their first decade of life, and that we know how to ensure that virtually all of them do so.
The first hurdle, for children from birth to age 5, is development of the cognitive, linguistic, social, and psychological bases on which later success depends. Second, by the end of 1st grade, students should be well on the way to reading. Each year afterward, students need to make adequate progress in basic and advanced skills. Their progress should enable them to avoid any need for remedial or special education and to be promoted each year.
Research on birth-to-3, preschool, and kindergarten programs shows that we know how to ensure that children enter 1st grade with good language skills, cognitive skills, and self-concepts, no matter what their family backgrounds or personal characteristics. Research on tutoring and on instruction, curriculum, and organization of early grades education shows that we know how to ensure that children enter 4th grade reading, regardless of their family and personal backgrounds. We have focused on early interventions, but it is important to note that many programs and practices show strong evidence of effectiveness for at-risk students throughout the grades (see Slavin et al. 1989). Rather than expecting short-term interventions to have long-term effects, we need to provide at-risk children with the services they need at a particular age or developmental stage.
Does this mean that we need to provide intensive (and therefore expensive) “preventative” services to at-risk students forever? Perhaps we do need this for a very small portion of students now served in special education. But for the great majority of students, including nearly all of those currently served in compensatory education programs and most of those now called “learning disabled,” we believe that intensive intervention will only be needed for a brief period, primarily one-to-one tutoring in 1st grade. After these students are well launched in reading, they still need high-quality instruction and other services in the later elementary grades to continue to build on their strong base. Improving instruction is relatively inexpensive.
If a cook puts a high flame under a stew, brings it to a boil, and then turns it off, the stew will not cook. If the cook puts a stew on simmer without first bringing it to a boil, the stew will not cook. Only by bringing the stew to a boil and then simmering will the stew cook. By the same token, intensive early intervention for at-risk children with no follow-up in improved instruction is unlikely to produce lasting gains, and mild interventions over extended periods may also fail to bring low achievers into the educational mainstream. Yet intensive early intervention followed by long-term (inexpensive) improvements in instruction and other services can produce substantial and lasting gains.
The best evidence for this perspective comes from research on Success for All. This program usually begins with 4-year-olds, giving them high-quality preschool and kindergarten experiences. These are enough for most children, but the program provides one-to-one tutoring, primarily in 1st grade, for those who have serious reading problems. After that, improvements in curriculum and instruction, plus long-term family support services, are intended to maintain and build on the substantial gains students make in tutoring. The program's findings have shown the effectiveness of this approach; not only do at-risk students perform far better than matched control students at the end of 1st grade, but their advantage continues to grow in 2nd, 3rd, and 4th grades. This is not to say that the particular elements implemented in Success for All are all optimal or essential. Other preschool or kindergarten models, reading models, or tutoring models could be more effective, and outcomes for the most at-risk children could probably be enhanced by intervening before age 4. What is important here is the idea that linking prevention, early intervention, and continuing instructional improvement can prevent school failure for nearly all students.
How Many Students Can Succeed At What Cost?
Our research summary shows that virtually every child can succeed in the early grades in principle. The number who will succeed in fact depends on the resources we are willing to devote to ensuring success for all and to our willingness to reconfigure the resources we already devote to remedial and special education and related services.
We have evidence (particularly from the Success for All research) to suggest that we can ensure the school success of the majority of disadvantaged, at-risk students using the local and Chapter 1 funds already allocated to these schools in different ways (primarily to improve curriculum, instruction, and classroom management in the regular classroom). However, to ensure the success of all at-risk students takes a greater investment. There is a large category of students who would fail to learn to read without intervention but would succeed with good preschool and kindergarten experiences; improved reading curriculum and instruction; and perhaps brief tutoring at a critical juncture, eyeglasses, family support, or other relatively inexpensive assistance. A much smaller group of students might require extended tutoring, more intensive family services, and so on. A still smaller group would need intensive intervention before preschool as well as improved early childhood education, tutoring, and other services to make it in school. One could imagine that any child who is not seriously retarded could succeed in school if he or she had some combination of the intensive birth-to-3 services used in the Milwaukee project; the high-quality preschool programs used in the High/Scope model; the tutoring provided by Reading Recovery or other models; and the improvements in curriculum, instruction, family support, and other services (along with tutoring) provided throughout the elementary grades by Success far All.
The cost of ensuring the success of these extremely at-risk children would, of course, be enormous. Yet a multi-risk child (such as a child from an impoverished and disorganized home with low IQ and poor behavior) will, without effective intervention, cost schools and society an equally enormous amount. Even in the mid-term, excess costs for special or remedial education over the elementary years are themselves staggering. This leaves aside the likely long-term costs of dropping out, delinquency, early pregnancy, and so on (see Barnett and Escobar 1987). The key issue for at-risk students is not if additional costs will be necessary, but when they should be provided. By every standard of evidence, logic, and compassion, dollars used preventively make more sense than the same dollars used remedially.
The good news in research on prevention and early intervention is that early school failure is fundamentally preventable. The implications of this should be revolutionary. At the policy level, it means we can choose to eradicate school failure, or we can allow it to continue. What we cannot do is pretend that we do not have a choice.
