Whenever and Wherever We Choose: The Replication of Success for All

'Whenever and Wherever We Choose'
The Replication of 'Success for All'
The Authors present data demonstrating that substantially greater success for disadvantaged students can be routinely ensured in schools that are neither exceptional nor extraordinary.
By ROBERT E. SLAVIN, NANCY A. MADDEN, LAWRENCE J. DOLAN BARBARA A. WASIK, STEVEN M. ROSS, AND LANA J. SMITH
ROBERT E. SLAVIN is director of the Elementary School Program at the Center for Research on Effective Schooling for Disadvantaged Students, Johns Hopkins University Baltimore, where NANCY A. MADDEN LAWRENCE J. DOLAN are Principal Research Scientists, and BARBARA A. WASIK is a research scientist. STEVEN M. ROSS and LANA J. SMITH are professors of education at Memphis State University, Memphis, Tenn. An earlier version of this article was presented in April 1993 at the annual meeting of the American Educational Research Association, held in Atlanta. The research was funded by the Office of Educational Research and Improvement, U.S. Department of Education (No. OERI-R-117-R0002), the Carnegie Corporation of New York, the Pew Charitable Trust, the Abell Foundation and the France and Merrick Foundations. However, the opinions expressed are those of the Authors.
FIFTEEN years ago, Ronald Edmonds put forth a proposition that has served as the touchstone of the school effectiveness movement ever since: "We can, whenever and wherever we choose, successfully teach children whose schooling is of interest to us."¹ This proposition has two parts: one is the assertion that every child can learn; the other, that we have the knowledge we need to create schools in which every child will learn.
Can every child learn? Research has demonstrated that at-risk students can achieve at far higher levels than they have in the past. For example, research on early intervention and research on one-one tutoring demonstrate the principle that virtually every child can learn.² Extraordinary teachers and schools prove every day that they are able to produce extraordinary outcomes with disadvantaged and minority children. While it is becoming clear that every child can learn, there is a part of Edmonds' formulation that is more problematic. Is it really true that we can ensure the learning of every child "whenever and wherever" we choose? Demonstrations of outstanding learning achievements in laboratories or in schools with unusual, charismatic principals and teachers are useful, but they do not tell us that success can be replicated on a large scale. Many pilot programs have shown substantial success, but that success has proved difficult to replicate. Clearly, it is not enough to demonstrate success and expect others to do likewise. For a program or strategy to make a difference on a broad scale, it must be robust under many circumstances, it must withstand the test of time, and it must demonstrate effectiveness in new sites that are not under the day-to-day control of the program developers.
In this article we present a summary of data from Success for All, a program that is now in the process of making the transition from local pilot to national model. This is the first report to present data on several sites beyond the original home of Success for All in Baltimore. Earlier evaluations of the program have shown that it can be highly successful in increasing reading achievement among very disadvantaged students.³ Can this success by replicated whenever and wherever we choose?
It is important to state up front what we mean by "we." No one can pretend that researchers, developers, or government agencies can by themselves ensure, the success of all students. The enthusiastic and wholehearted commitment school staffs and district administrators also essential. In the case of Success of All, we work only with school staffs that, have voted at least 80% in favor of participating and districts that have made clear commitment to implement the program.
Success for All began in Baltimore 1986. It was designed in a collaboration between our group at Johns Hopkins University and the Baltimore City Public Schools and piloted in one school in the 1987-1988 school year. Since then Success for All has expanded both within and outside Baltimore and is currently being implemented in a total of 100 schools in 40 school districts in 20 states from coast to coast, including Canada. From the outset, the program has emphasized rigorous evaluation of its result comparison with those of matched control schools in the same districts, preferably through the use of individually administered tests of reading. Due to funding limitations, not all Success for All schools are being assessed in this way, but we currently have high-quality assessment data from 15 schools in seven districts in seven states, a remarkable body of evidence for an innovative program. Three of the districts were evaluated by Johns Hopkins staff members, and four (using identical measures and procedures) were evaluated by an independent group at Memphis State University. This article summarizes the findings from these seven districts, plus two that have done their own evaluations using routinely administered standardized tests.
ELEMENTS OF SUCCESS FOR ALL
Our basic approach to designing a program to ensure the success of all disadvantaged children begins with two essential principals: prevention and immediate, intensive intervention. That is, learning problems must be prevented by providing children with the best available classroom programs and by engaging parents in support of their children's school success. When learning problems do appear, corrective interventions must be immediate, intensive, and minimally disruptive to students' progress in the regular program. Thus students receive help early on, when their problems ate small. The help is intensive and effective enough to catch students up with their classmates, so that they can profit from their regular classroom instruction.
Instead of letting students fall further and further behind until they need special or remedial education or are retained in grade, Success for All gives them ,whatever they need to keep up the basic skills as soon as they need it. Typically, Success for All does not require significant additional expenditures, but rather shifts in existing Chapter 1, special education, and other dollars from remediation to prevention and early intervention. The elements of Success for All are described below.⁴
Reading tutors.
One of the most important elements of the Success for All model is the use of one-to-one tutoring the most effective form of instruction known-to support students' success in reading.⁵ The tutors are certified teachers with experience teaching Chapter 1, special education, and/or primary-level reading. Tutors work directly with individual students who are having difficulty keeping up with their reading groups. The students are taken from their homeroom classes by tutors for 20-minute sessions during times other than reading or math periods. In general, tutors support students' success in the regular reading curriculum, rather than teaching different objectives. For example, if the regular reading teacher is working on stories with long vowels or is reaching strategies to monitor comprehension, so does the tutor. However, tutors seek to identify learning deficits and use different strategies to teach the same skills.
During daily 90 minute reading periods, tutors serve as additional reading teachers to reduce the size of reading classes. Brief forms carry information on students' specific deficits and needs between reading teachers and tutors, and reading teachers and tutors meet regularly to coordinate their approaches with individual children.
The initial decisions about placements in reading groups and the need for tutoring are made according to the results of informal reading inventories given to each child by the tutors. Subsequent changes in reading group placements and tutoring assignments are made according to outcomes of eight-week assessments, which include teacher judgments as well as more formal assessments.
First graders receive priority for tutoring, on the assumption that the primary function of the tutors is to help all students be successful in reading the first time, before they become remedial readers.
Reading program.
For most of the day, students in grades 1 through 3 are assigned to heterogeneous, age-grouped classes of about 25. But during a regular 90-minute reading period, they are grouped into reading classes of 15 students who are all performing at the same reading level. For example, a first-semester, second grade reading class might contain first, second, and third grade students, all reading at the same level.
Regrouping allows teachers to teach the whole class without having to break the class into reading groups. This greatly reduces the time spent on seatwork and increases the time available for direct instruction. We do not expect reduction in class size to increase reading achievement by itself,⁶ but the small classes ensure that every reading class will be working at one level, which eliminates many of the workbooks, dittos, and other follow-up activities that are needed in classes with multiple reading groups. The regrouping is a form of the Joplin Plan, which has been found to increase reading achievement in the elementary grades.⁷
The reading program itself has been designed to take full advantage of having 90 minutes of direct instruction. The program emphasizes the development of basic language skills and sound and letter-recognition skills in kindergarten. It uses an approach based on sound blending and phonics starting in first grade (although kindergarten students who show readiness may be accelerated into the first grade program if the school chooses). Students in pre-K, kindergarten, and first grade use the Peabody Language Development Kits to help them build language concepts essential to later reading success.
The K-1 reading program uses a series of "shared stories," in which part of the story is written in small type and read by the teacher, while part is written in large type and read by students. The student portion uses phonetically controlled vocabulary. The program emphasizes oral reading to student partners as well as the teacher, instruction in story structure and specific comprehension skills and integration of reading and writing. When they reach the primer reading level, students use a form of Cooperative Integrated Reading and Composition (CIRC) with novels or basals. CIRC uses cooperative learning activities built around story structure, prediction, summarization, vocabulary building, decoding practice, writing and direct instruction in reading comprehension skills.
Eight-week assessments.
Every eight weeks, reading teachers assess students' progress through the program. The results of these assessments ate used to determine who is to receive tutoring, to suggest adaptations in students' programs, and to identify students who need other types of assistance, such as family interventions or vision/hearing screenings.⁸
Preschool and kindergarten.
Most Success for All schools provide a half-day preschool and/or a full-day kindergarten for all eligible students. The preschool and Kindergarten provide a balanced and developmentally appropriate learning experience for young children. The curriculum emphasizes the development and use of language and provides a mix of academic readiness and music, art, and movement activities. Readiness activities include the use of integrated thematic units, Peabody Language Development Kits, and a program called Story Telling and Retelling (STAR) in which students retell stories read by teachers.
Family support team.
A "family support team" that consists of any social workers, parent liaisons, counselors, and others who work in the school provides parenting education and makes an effort to involve parents in support of their children's success in school. In addition, family support staff members lend assistance when there are indications that students are not working up to their full potential because of problems at home. For example, families of students who are not getting adequate sleep or nutrition, who need glasses, who are not attending school regularly, or who are exhibiting serious behavior problems receive assistance from the family support team. Links with appropriate community service agencies are also forged to provide as much focused service as possible for parents and children.
Program facilitator.
A full-time program facilitator works at each school to oversee (with the principal) the operation of Success for All. The facilitator helps plan the Success for All program, assists the principal with scheduling, and visits classes and tutoring sessions frequently to help teachers and tutors with individual problems. The program facilitator may work with individual children in order to find successful strategies for teaching them and then return them to the tutors or teachers. The facilitator also coordinates the activities of the family support team with those of the instructional staff.
Teachers and teacher training.
The teachers and tutors are regular teachers. They give detailed teacher's manuals, supplemented by two days of inservice training at the beginning of the school year and several inservice training sessions throughout the year on such topics as classroom management, instructional pace, and the implementation of the curriculum.
Special education. Every effort is made to deal with students' learning problems within the context of the regular classroom, as supplemented by tutors. Resource services for special education are still provided for students previously assigned to special education, but no new, assignments to resource services are made for reading problems. The assumption is that the tutoring services available to all students will be more appropriate. Self-contained services are maintained for seriously handicapped students whose needs cannot be met by regular classrooms.
Advisory committee.
An advisory committee composed of the building principal, facilitator, selected teachers and tutors, and parent representatives meets regularly to review the progress of the program and to identify and solve any problems that arise.

EVALATION DESIGN

Success for All uses a common evaluation design, with variations to account for local circumstances. Every Success for All school involved in a formal evaluation is matched with a control school that is similar in poverty level percentage of students qualifying for free lunch, historical achievement level, ethnicity of the student body, and a variety of other factors. Children in the Success for All schools are then matched either by scores on district-administered standardized tests given in kindergarten or (starting in 1991 in several districts) by scores on the Peabody Picture Vocabulary Test (PPVT) given by the project in the fall of kindergarten or first grade. In some cases, analyses of covariance rather than matches of individual children were used, and, in the case of Key School in Philadelphia, the schools were matched but the individual children could not be matched (because the school serves many students with limited proficiency in English who were not tested by the district in kindergarten).
The measures used in the 1992 and 1993 evaluations of Success for All were as follows:
1. Woodcock Reading Mastery Test. Three Woodcock scales -word identification, word attack, and passage comprehension - were administered to the students by trained testers. Word identification assesses recognition of common sight words, word attack assesses phonetic synthesis skills, and passage comprehension assesses comprehension in context.
2. Durrell Analysis of Reading Difficulty. The Durrell oral reading scale was also administered individually. It presents a series of graded reading passages that students read aloud; the passages are followed by comprehension questions.
Except at Philadelphia's Key School, analyses of covariance-with pretests as covariates were used to compare raw scores in all evaluations, and separate analyses were conducted for students in general and for students in the lowest 25% of their grades. At Key School analyses of variance -were used and results were reported separately for Asian (mostly Cambodian) and non-Asian students.
The summaries of student performance are presented here in grade equivalents (adjusted for covariates) and effect sizes (proportion of a standard deviation separating the experimental and control groups), averaging across individual measures. Neither grade equivalents nor averaged scores were used in the analyses, but they are presented here as a useful summary. The outcomes for all students in the Success for All and control schools, as well as those for the students in the lowest 25% of their grades - those who are most at risk - are presented below. In most cases the lowest 25% was determined by scores on the PPVT that was given as a pretest. In Baltimore and in Charleston, South Carolina, however, PPVT protests were not given, and the lowest 25% were chosen according to scores on the posttest.
Each of the evaluations summarized below compares children who began Success for All in first grade or earlier to children who amended the control school during the same period. Because Success for All is a prevention program, students who begin the program after first grade are not considered to have received the full treatment (although they are of course served -within the schools). At this writing, data for spring 1993 were available from some sites, while 1992 data were used for others. The most recent data available are presented in each case.⁹
Baltimore. The original home of Success for All, Baltimore is the district with the longest longitudinal data base and the largest number of schools being evaluated. As of Spring 1992, the original pilot school had used the program for five years, and four additional schools had used it for four years. The student bodies at all the schools are almost entirely African American; from 75% to 96% of the students in each school are eligible for subsidized lunches.
Students in Success for All are performing significantly better than students in control groups at all grade levels from first to third. The effect sizes for each grade are: grade 1, +.31; grade 2, +.61; grade 3, +.53. As has been typical of Success for All evaluations, the effects have been particularly large for students who are in the lowest 25% of their classes: grade 1, +.77; grade 2, +1.73; grade 3, +.87.
In addition to the effects seen on individually administered reading measures. Success for All has had an impact on several other measures. Among these are the standardized tests routinely administered by the district. On the Comprehensive Tests of Basic Skills (CTBS), no differences were found between Success for All and control students in first grade, but significant differences in favor of Success far All were found on the CTBS reading comprehension and language scales in second through fifth grades. We do not consider standardized test scores to be as reliable, valid, and relevant as the individualized measures, but we track them because of their importance to the district.
Evaluations of Success for All in Baltimore have also found positive effects on attendance. Retention in grade has been reduced from an average of 11% in grades K-3 to near zero. Avoiding retention is a policy of Success for All rather than an outcome of the program, but it is important to show that inner-city schools can get along without retaining large numbers of students. In 1992-93 three additional schools began using Success for All, and we are piloting a program to ease the transition to middle school for students graduating from Success for All elementary schools.¹⁰
Philadelphia. The first school district outside Baltimore to implement Success for All was Philadelphia. Francis Scott Key Elementary School, which serves a population that is majority Southeast Asian (mostly Cambodian) has been using the program since 1988. Three more schools began to implement Success for All in 1991. These are all very poor schools that serve primarily African American children, with nearly 100% of the students qualifying for subsidized lunches.
Asian students at Key School exceeded those in their control school by an average of more than a full grade-equivalent in first grade (ES = +4.69); Asian students in Success for All were reading above grade level, While their counterparts in the control schools were nonreaders, scoring near the bottom of the scale on all tests. Asian second graders at Key School exceeded their controls by more than a grade equivalent (ES = +1.67). Asian third graders (ES = +.47), and fourth graders (ES = +.37) exceeded control students by six and five months respectively. Non-Asian students at Key School out performed their controls by an average of approximately 3.5 months in first grade (ES = + 1.47), three months in second grade (ES = +.26), four months in third grade (ES = +.27), and five months in fourth grade (ES = +.24).
The results for two of the three Philadelphia schools that began Success for All in 1991 (data on the third school were lost) show that first graders in these schools were reading above grade level and two months ahead of their controls (ES = +.31). The lowest-achieving 25% of Success for All students were reading about at grade level and exceeded their controls by 3.5 months (ES = +.62).¹¹
Charleston, South Carolina. Pepperhill Elementary School in Charleston began to use Success for All in the 1990-91 school year. Pepperhill is the only Success for All school being evaluated that is not a Chapter 1 school (although a school in Philadelphia that does not qualify as a Chapter 1 school adopted the program in 1992-93). Despite a poverty rate much higher than the national average (40% of students qualify for subsidized lunches and 60% are African American), Pepperhill does not qualify for Chapter 1 funding within the Charleston district. However, under the South Carolina Educational Improvement Act, it does receive state funds for compensatory education that pay for the cost of Success for All.
Still, because it is not a Chapter 1 school, Pepperhill has less money to implement Success for All than do other schools in the program. As a result, Pepperhill is the only school being evaluated that uses paraprofessionals rather than certified teachers as tutors. The paraprofessionals are of high quality: one is certified to teach in another state, and another has a four-year degree. Moreover, all received significantly more training than is given to tutors who are certified to teach. Pepperhill provides us with the first opportunity to evaluate Success for All in a school that has fewer dollars to spend and a relatively less needy population.
The outcomes for first graders at Pepperhill show a substantial positive effect of the program on student achievement. This school has the highest mean reading level of any Success for All school (a grade equivalent of 2.45), four months ahead of its control school. Effects were also quite positive for students in the lowest 25% of their grade.¹²
Memphis State evaluations. The evaluations of Success for All in Baltimore, Philadelphia, and Charleston are being conducted directly by our group at Johns Hopkins. Several additional evaluations are being conducted by an independent evaluation team led by Steven Ross and Lana Smith and working out of Memphis State University. These include evaluations in Memphis, in Fort Wayne, Indiana, in Montgomery, Alabama, and in Caldwell, Idaho.
The evaluations in three of the four districts structured by the Memphis State team found substantial positive effects of Success for All. In Memphis, first graders in an inner-city school were reading well above grade level (a grade equivalent of 2.1) and three months ahead of their matched controls (ES = +.38). Effects in second grade were substantial for the general student population (ES = +.51) and even stronger for those in the lowest 25% (ES = +2.66). Standardized test scores also favored Success for All students in reading comprehension.¹³
A study of two schools in Fort Wayne, Indiana, found strong positive effects of Success for All on student achievement. First graders were reading well above grade level (a grade equivalent of 2.45) and more than three months ahead of controls (ES = +.51). The lowest 25% of students were also reading above grade level and 4.5 months ahead of their control group (ES = +.79). Second graders were reading four months ahead of their controls (ES = +.44), and the lowest 25% of second graders were almost on grade level and more than four months ahead of their counterparts (ES = +.79).¹⁴
Two schools in Montgomery, Alabama, had the largest program effects of any first grade evaluation. Success for All first graders were reading more than five months ahead of their peers in the control group (ES = +1.32). Among the lowest 25%, control students were not reading at all, while Success for All students posted an average grade equivalent score of 1.46 (ES = +2.86).¹⁵
The only school that failed to show positive effects of Success for All is a school in rural Caldwell, Idaho. First graders in Caldwell performed no better than their control group. We find these results puzzling. Observations of program implementation indicated that the school was implementing Success for All with spirit and fidelity; in fact, this school was rated by our facilitators as one of the best implementations.
It is possible that the reason for the failure to find differences may have to do with the control school, not the Success for All school. Observers found that the matched control school was an extraordinary place - a new, facility with an exceptionally able principal and staff. First graders at the Success for All school in Caldwell had some of the highest scores of any of our schools, with an average grade equivalent in reading of 2.22. In this case, however, the control first graders also had high scores (a grade equivalent of 2.1 8).¹⁶
DISTRICT EVALUATIONS AND POOLED OUTCOMES

Some of the school districts implementing Success for All have sent us standardized test data. A school in Charleston, West Virginia, showed substantial gains in test scores and attendance and reduced its retentions to zero. A school in Wichita Falls, Texas, saw, the percentage of third graders meeting minimum expectations on a state test increase from 48% to 70% in reading and from 8% to 53% in writing, while the rest of the district stayed at about the same level on both scales. Two schools in Modesto, California, made extraordinary gains in grades 1 through 3 on the Comprehensive Tests of Basic Skills.
In addition to analyzing the site-by-site results, we also used a method for Combining the experimental/control comparisons over many replications and over time.¹⁷ This technique - called a multisite replicated experiment - considers each successive group of students and each pair of experimental and control schools a replication. For example, across the 15 Success for All schools we have structured, 37 separate groups of first graders have experienced the program and have been assessed on individually administered reading measures. Twenty-one groups of second graders and 13 groups of third graders have been in the program since first grade. This pooling procedure is a minor variation on the kind of meta-analysis commonly used in medical research, and it allows us over time to build up a sample large enough to conduct school-level analyses of program effects.
Figure 1 shows the mean grade equivalents and effect sizes for all 15 Success for All schools and their control schools. For students in general, effect sizes averaged more than half a standard deviation each year, which means that the average Success for All child performs better than 70% of the control students. Analyses at the school level indicate that the effect sizes are significantly different from zero at all three grade levels (p<.001). because standard deviations increase with each year in school, a constant effect size implies a growing difference in absolute or grade equivalent terms; Success for All students exceed matched control students by about three months in first grade but by almost seven months in third grade.
Effects for students in the lowest 25% of their groups are consistently larger than for students in general, averaging an effect size of +1.04 in grade 1, +1.47 in grade 2, and +1.49 in grade 3. By the end of third grade these low-achieving students are performing better than 93% of matched control students. These effect sizes are significantly different from zero (p<.001 in grades I and 2; p<.004 in grade 3). Larger effect sizes for low-achieving students than for students in general have been found in almost every evaluation of Success for All. This is primarily a result of the tutoring, family support, and other services that are principally provided to the lowest-achieving students. A major goal of Success for All is to build a floor under the achievement of all students, and the large gains made by the lowest achievers are evidence that this is occurring.
PROGRAM EFFECTS OVER TIME

Data on the effects of Success for All in successive years of implementation are summarized in Figure 2.¹⁸ These data dearly show that the longer a school is in the program, the better the effects on the reading performance of students in all grades. For example, in the first implementation year, Success for All first graders exceed their control counterparts by an effect size of +.34; this rises to +.57 for the second group of first graders, +.78 for the third, and +.82 for the fourth.
There are two likely explanations for the growth in improvement in successive years. One is that the schools simply get better with practice. because Success for All involves so many changes, implementing all of them to a high degree of quality can take more than a single year. The second factor to consider is the fact that only in the second year have first graders participated in Success for All since kindergarten, and only in the third year have they participated since prekindergarten.
REPLICABILITY AND IMPACT

The results of evaluations of 15 Success for All schools in seven states clearly show, that the program improves student performance in reading. In all but one school, Success for All students learned significantly more than matched control students. Significant effects were not seen on every measure at every grade level, but the consistent direction and magnitude of the effects shows unequivocal benefits for Success for All students. Evaluations for previous years have shown that Success for All has been effective in its original districts, Baltimore and Philadelphia. These new evaluations add evidence that the program can be replicated far from its original home with similar results.
The Success for All evaluations have used reliable and valid measures, consisting of individually administered tests that are sensitive to all aspects of reading, including comprehension, fluency, word attack, and word identification. Moreover, the performance of Success for All students has been compared to that of matched students in matched control schools. Replication of such high-quality experiments in such a wide variety of schools and districts is extremely unusual.
Although the outcomes on individually administered measures have the greatest validity and scientific importance, a number of other indicators point to the replicability and practical impact of Success for All. One indicator is standardized test scores, which were found to increase significantly in Baltimore, in Charleston, West Virginia, in Wichita Falls, Texas, and in Modesto, California.
Another measure is the fact that most of the districts being evaluated (and many others as well) have expanded the program to additional sites after the initial pilots. An important indicator of the to business of Success for All is the fact that, of the more than 70 schools that have used the program for periods of up to six years, only three have dropped out (in all cases because the principal changed). However, many other Success for All schools have survived changes of superintendents, principals, facilitators, and other key staff members.
There is nothing magic about Success for All. None of its components are completely new or unique. The first-year results in Caldwell, Idaho, which seem so puzzling on the surface, support the commonsense observation that schools serving disadvantaged students can have great success without a special program if they have an outstanding staff. Other prevention/early intervention models, such as Reading Recovery¹⁹ and James Comer's School Development Program²⁰ have also shown themselves to be effective with disadvantaged children. The main value of the Success for All research is not in validating a particular model or in demonstrating that disadvantaged students can learn. Rather, the findings' greatest importance is in demonstrating that substantially greater success for disadvantaged students can be routinely ensured in schools that are neither exceptional nor extraordinary schools that were not producing great success before the program was introduced.
We cannot ensure that every school has a charismatic principal or that every student has a charismatic teacher. Nevertheless, we can ensure that every child, regardless of family background, has an opportunity to succeed in school.
If Success for All were markedly more expensive than traditional programs, that would be a major practical limitation on its replicability. However, this is usually not the case. Additional funding to hire tutors does help ensure the success of every child. But, while two of the original Success for All schools did receive significant additional funds, none of the rest have. Schools with high enrollments of poor children usually have sufficient Chapter 1 funds to implement a credible form of the model, and these funds are often supplemented by reallocated funds or personnel from special education, state compensatory education, desegregation settlements, bilingual education, and other sources. Only one of the 15 schools in the longitudinal evaluation receives significantly more funding than its control school.
Demonstrating that an effective program can be replicated successfully removes one more excuse for the continuing low achievement of disadvantaged children. When Ronald Edmonds stated that we can successfully teach all children "whenever and wherever we choose," he may indeed have been right in principle. However, practical demonstrations of this principle are still essential. To ensure the success of disadvantaged students, -we must have the political commitment to do so, along with the funds and policies to back up the commitment. We must also have methods known to be effective in their original sites and replicable and effective in other sites.
Success for All requires a serious commitment to restructure elementary schools and to reconfigure the uses of Chapter 1, special education, and other funds to emphasize prevention and early intervention rather than remediation. It also requires a vote of at least 80% of teachers in favor of implementing the program. However, when a school makes such a commitment, it can succeed. The evidence summarized in this article offers practical proof that, "whenever and wherever we choose," we can successfully teach all children.

1. Ronald Edmonds, "Effective Schools for the Urban Poor," Educational Leadership, October 1979, p. 23.
2. Robert E. Slavin, Nancy L. Karweit and Barbara A. Wasik, "Preventing Early School Failure : What Works," Educational Leadership, 1992/ January 1993, pp. 10-18; idem, eds., Preventing Early School Failure (Boston: Allyn and Bacon, 1994); and Barbara A. Wasik and Robert E. Slavin. "Preventing Early Reading Failure with One-to-One Tutoring: A Review of Five Programs," Reading Research Quarterly, vol. 28, 1993, pp. 178-200.
3. Nancy A. Madden et al., "Success for All: Longitudinal Effects of a restructuring program for Inner-City Elementary Schools," American Educational Research Journal vol. 30, 1993, pp. 12348; and Robert E. Slavin et tal., Success for All A Relentless approach to Prevention and Early Intervention in Elementary Schools, VA, Educational Research Service , 1992).
4. For more detail on the elements of Success for All, see Slavin et al., op. cit.
5. Wasik and Slavin, op. cit.
6. See Robert E. Slavin, 'School and Classroom Organization in Beginning Reading: Class Size, Aides, and instructional Grouping.' In Slavin, Karweit, and Wasik, Preventing Early School Failure.
7. Robert E. Slavin, 'Ability Grouping and Student Achievement in Elementary Schools: A Best-Evidence Synthesis,' Review of Educational Research, vol. 57, 1987, pp. 293-336.
8. Robert J. Stevens et al., "Cooperative Integrated Reading and Composition: Two Field Experiments," Reading Research Quarterly, vol. 22, 1987, pp. 433-54.
9. For details on evaluation methods and findings, see Slavin et al., op. cit., and the final reports for individual sites (noted below).
10. For more on the Baltimore evaluations, see Madden et al., op. cit., and Slavin et al., op. cit.
11. For more on the Philadelphia evaluations, see Robert E. Slavin, Success for All in the Philadelphia Public Schools: 1991-1992 Evaluation Reprint (Baltimore: Center for Research on Effective Schooling for Disadvantaged Students, Johns Hopkins University, 1993).
12. Barbara A. Wasik and Robert E. Slavin, Success for All at Pepperhill Elementary School- 1992 Evaluation (Baltimore: Center for Research on Effective Schooling for Disadvantaged Students, Johns Hopkins University, 1993).
13. Steven M. Ross and Lana J. Smith, Final Report: 1992-93 Success for All Program in Memphis, Tennessee (Memphis State University, 1993).
14. Lana J. Smith and Stevens M. Ross, Final Report:1992-93 Success for All program in Ft. Wayne, Indiana (Memphis: Memphis State University, 1993).
15. Steven M. Ross et al., Final Report: 1992-93 Success for All program in Montgomery, Alabama (Memphis: Memphis State University, 1993).
16. Steven M. Ross, Lana J. Smith, and Jason Casey, Final Report: 1991-92 Success for All in Caldwell Idaho (Memphis: Memphis State University, 1992).
17. Robert E. Slavin and Nancy A. Madden, "Multi-Site Replicated Experiments: An Application to Success for All," paper presented at the annual meeting of the American Educational Research Association, Atlanta, April, 1992.
18. Ibid.
19. Gay Su Pinnell, "Reading Recovery: Helping At-Risk to Read," Elementary School Journal, vol. 90,1989, pp. 161-82.
20. James Comer, "Educating Poor Minority Children," Scientific American, vol. 259,1988, pp. 42-48.
NOTE: This article was reprinted from Phi Delta Kappan, April 1994, pages 639-647.