This paper describes the requirements for effective reform in education. At the classroom level, successful reform depends on a focus on four factors: quality of instruction, levels of instruction appropriate to student needs, incentives for students, and adequate time for learning. Research on means of enhancing each of these factors is reviewed. Requirements for school effectiveness and school change are reviewed, and these are discussed in the Latin American context. The paper concludes that the key to school improvement in Latin America will be professional development focusing on whole-school change, including adoption of proven methods and materials on a schoolwide basis. Effective change is most likely to be comprehensive, systemic, and sustained, with school staffs taking an active role in choosing from among effective alternatives. At the system level, school-by-school change must begin small and expand rapidly with support from all levels of school administration.

Education has always been a critical requirement for economic development, but as the world economy is changing, education is becoming increasingly important. Natural resources, location, and other factors are declining in importance as the ability of a nation's people to do complex work is becoming the key determinant of national wealth. Nations throughout the world have long been aware of the link between education and prosperity and are seeking to improve their educational systems. Increasing investment in education is one focus in many countries, but there is no guarantee that increasing funding will result in increased student performance. What matters is what schools and school systems do with existing or increased funds to increase the efficacy of teaching and learning.

The most important dynamic in education is the interaction between teacher and child. Every other element of the education system merely provides the context within which teacher-child interaction takes place. A discussion of school reform must begin with a discussion of the teaching behaviors and school characteristics associated with optimum student achievement, and then build up from there a system to support these behaviors and characteristics.

The purpose of this paper is to summarize research on effective teaching and effective schools, and to then relate this research to policies and practices that might be involved in the reform of schools in Latin America and elsewhere.

In the past twenty years, research on teaching has made significant strides in identifying teaching behaviors associated with high student achievement (Brophy and Good 1986; Rosenshine and Stevens 1986). However, effective instruction is not just good teaching. If it were, we could probably find the best lecturers, make video tapes of their lessons, and show them to students (see Slavin 1987a, 1994). Consider why the video teacher would be ineffective. First, the video teacher would have no idea what students already know. A particular lesson might be too advanced for a particular group of students, or it may be that some students already know the material being taught. Some students may be learning the lesson quite well, while others are missing key concepts and falling behind because they lack prerequisite skills for new learning. The video teacher would have no way to know who needed additional help, and would have no way to provide it in any case. There would be no way to question students to find out if they were getting the main points and then to reteach any concepts students were failing to grasp.

Second, the video teacher would have no way to motivate students to pay attention to the lesson or to really try to learn it. If students were failing to pay attention or were misbehaving, the video teacher would have no way of doing anything about it. Finally, the video teacher would never know at the end of the lesson whether or not students actually learned the main concepts or skills.

The case of the video teacher illustrates the point that teachers must be concerned with many elements of instruction in addition to the lesson itself. Teachers must attend to ways of adapting instruction to students' levels of knowledge, motivating students to learn, managing student behavior, grouping students for instruction, and testing and evaluating students. These functions are carried out at two levels. At the school level, the principal and/or central administrators may establish policies concerning grouping of students (e.g., tracking), provision and allocation of special education and remedial resources, and grading, evaluation, and promotion practices. At the classroom level, teachers control the grouping of students within the class, teaching techniques, classroom management methods, informal incentives, frequency and form of quizzes and tests, and so on. These elements of school and classroom organization are at least as important for student achievement as the quality of teachers' lessons.

Slavin (1984) proposed a model of effective instruction which focused on the alterable elements of Carroll's model (1963, 1989) "Model of School Learning." These are the elements that teachers and schools can directly change. The components of this model of alterable elements of effective instruction are as follows:

1. Quality of Instruction: The degree to which information or skills are presented so that students can easily learn them. Quality of instruction is largely a product of the quality of the curriculum and of the lesson presentation itself.

2. Appropriate Levels of Instruction: The degree to which the teacher makes sure that students are ready to learn a new lesson (that is, they have the necessary skills and knowledge to learn it), but have not already learned the lesson. In other words, the level of instruction is appropriate when a lesson is neither too difficult nor too easy for students.

3. Incentive: The degree to which the teacher makes sure that students are motivated to work on instructional tasks and to learn the material being presented.

4. Time: The degree to which students are given enough time to learn the material being taught.

The four elements of this QAIT (Quality, Appropriateness, Incentive, Time) model have one important characteristic: All four must be adequate for instruction to be effective. Again, effective instruction is not just good teaching. No matter how high the quality of instruction, students will not learn a lesson if they lack the necessary prior skills or information, if they lack the motivation, or if they lack the time they need to learn the lesson. On the other hand, if the quality of instruction is low, then it makes no difference how much students know, how motivated they are, or how much time they have. Each of the elements of the QAIT model is like a link in a chain, and the chain is only as strong as its weakest link. In fact, it may be hypothesized that the four elements are multiplicatively related, in that improvements in multiple elements may produce substantially larger learning gains than improvements in any one.

Most of the advances in recent research on teaching have come about as a result of correlational process-product research, in which the practices of instructionally effective teachers have been contrasted with those of less effective teachers, controlling for student inputs. In recent years, the findings of these process-product studies have been incorporated into coherent instructional programs and evaluated in field experiments. Other coherent instructional methods not based on the process-product findings, such as mastery learning, cooperative learning, tutoring, and individualized instruction methods, have also been evaluated in field experiments. Each of these instructional methods is based on its own psychological or educational theories. However, the QAIT model is meant to inform all potential forms of classroom organization. Given a relatively fixed set of resources, every innovation in classroom organization solves some problems but also creates new problems which must themselves be solved. Tradeoffs are always involved. Understanding the terms of these tradeoffs is critical for an understanding of how to build effective models of classroom organization.

The QAIT model is designed primarily to clarify the tradeoffs involved in alternative forms of classroom organization. This paper presents a perspective on what is known now about each of the QAIT elements and, more importantly, explores the theoretical and practical ramifications of the interdependence of these elements for effective school and classroom organization.

Quality of instruction refers to the activities we think of first when we think of teaching: lecturing, discussing, calling on students, and so on. When instruction is high in quality, the information being presented makes sense to students, is interesting to them, is easy to remember and apply.

The most important aspect of instructional quality is the degree to which the lesson makes sense to students. For example, teachers must present information in an organized orderly way (Kallison 1986), note transitions to new topics (Smith and Cotton 1980), use clear and simple language (Land 1987), use many vivid images and examples (Hiebert, et al. 1991; Mayer and Gallini 1990), and frequently restate essential principles (Maddox and Hoole 1975). Lessons should be related to students' background knowledge, using such devices as advance organizers (Pressley, et al. 1992) or simply reminding students of previously learned material at relevant points in the lesson. Enthusiasm (Abrami, Leventhal, and Perry 1982) and humor (Kaplan and Pascoe 1977) can also contribute to quality of instruction, as can use of media and other visual representations of concepts (Hiebert, Wearne, and Taber 1991; Kozma 1991).

Clear specification of lesson objectives to students (Melton 1978) and a substantial correlation between what is taught and what is assessed (Cooley and Leinhardt 1980) contribute to instructional quality, as does frequent formal or informal assessment to see that students are mastering what is being taught (Crooks 1988; Kulik and Kulik 1988) and immediate feedback on the correctness of their performances (Barringer and Gholson 1979).

Instructional pace is partly an issue of quality of instruction and partly of appropriate levels of instruction. In general, content coverage is strongly related to student achievement (Dunkin 1978; Barr and Dreeben 1983), so a rapid pace of instruction may contribute to instructional quality. However, there is obviously such a thing as too rapid an instructional pace (see Leighton and Slavin 1988). Frequent assessment of student learning is critical for teachers to establish the most rapid instructional pace consistent with the preparedness and learning rate of all students.

Perhaps the most difficult problem of school and classroom organization is accommodating instruction to the needs of students with different levels of prior knowledge and different learning rates. If a teacher presents a lesson on long division to a heterogeneous class, some students may fail to learn it because they have not mastered such prerequisite skills as subtraction, multiplication, or simple division. At the same time, there may be some students who know how to divide before the lesson begins, or learn to do so very rapidly. If the teacher sets a pace of instruction appropriate to the needs of the students lacking prerequisite skills, then the rapid learners' time will be largely wasted. If the instructional pace is too rapid, the students lacking prerequisite skills will be left behind.

There are many common means of attempting to accommodate instruction to students' diverse needs, but each method has drawbacks that may make the method counterproductive. Various forms of ability grouping seek to reduce the heterogeneity of instructional groups. Special education and remedial programs are a special form of ability grouping designed to provide special resources to accelerate the achievement of students with learning problems. However, between-class ability grouping plans, such as tracking, can create low-ability classes for which teachers have low expectations and maintain a slow pace of instruction, and which many teachers dislike to teach (Good and Marshall 1984; Oakes 1985, 1987; Rowan and Miracle 1983; Slavin 1987b, 1990a). Similar problems make self-contained special education classes of questionable benefit to students with learning handicaps (see Leinhardt and Bickel 1987; Leinhardt and Pallay 1982; Madden and Slavin 1983). Within-class ability grouping, such as the use of reading and mathematics groups, creates problems of managing multiple groups within the classroom, reduces direct instruction to each student, and forces teachers to assign large amounts of unsupervised seatwork to keep students engaged while the teacher is working with a reading or mathematics group (Barr 1992).

Research on assignment of students to ability-grouped classes finds no achievement benefits for this practice at the elementary or secondary levels (see Slavin 1987b, 1990a; Oakes 1985, 1987). However, forms of ability grouping in which elementary students remain in heterogeneous classes most of the day but are regrouped into homogeneous reading or mathematics classes can be instructionally effective if teachers actually adapt their level and pace of instruction to meet the needs of the regrouped classes. In particular, the Joplin Plan, and certain nongraded plans in which elementary students are regrouped for reading or mathematics across grade lines and instructional level, is based on performance level rather than age can be instructionally effective (Slavin 1987b; Gutiérrez and Slavin 1992). Also, research on within-class ability grouping finds this practice to increase student mathematics achievement, particularly when the number of groups used is small and management techniques designed to ensure smooth transitions and high time-on-task during seatwork are used (Slavin 1987b; Slavin and Karweit 1984).

Group-based mastery learning (Bloom 1976; Block and Burns 1976; Guskey and Gates 1985) is an approach to providing levels of instruction that does not use permanent ability groups but rather regroups students after each skill is taught on the basis of their mastery of that skill. Students who attain pre-set criteria (e.g., 80%) on a formative test work on enrichment studies while non-masters receive corrective instruction. In theory, mastery learning should provide appropriate levels of instruction by ensuring that students have mastered prerequisite skills before they receive instruction in subsequent skills. However, within the confines of traditional class periods, the time needed for corrective instruction may slow the pace of instruction for the class as a whole. Studies of group-based mastery learning conducted in elementary and secondary schools over periods of at least four weeks have found few benefits of this approach in comparison to control groups given the same objectives, materials, and time as the mastery learning groups (Slavin 1987c).

The most extreme form of accommodation to individual differences short of one-to-one tutoring is individualized instruction, in which students work entirely at their own level and rate. Individualized instruction certainly solves the problem of providing appropriate levels of instruction, but it creates serious problems of classroom management, often depriving students of adequate direct instruction. Research on individualized instruction has not generally found positive effects on student achievement (Harley 1977; Horak 1981). However, Team Assisted Individualization, a form of individualized instruction which also incorporates the use of cooperative learning groups, has been found to consistently increase student achievement in mathematics (Slavin, Leavey and Madden 1984; Slavin, Madden, and Leavey 1984; Slavin and Karweit 1985; Slavin 1985).

Thomas Edison once wrote that "genius is one percent inspiration and ninety-nine percent perspiration." The same could probably be said for learning. Learning is work. This is not to say that learning must be drudgery, but it is certainly the case that students must exert themselves to pay attention, to study, and to conscientiously perform the tasks assigned to them, and they must somehow be motivated to do these things. This motivation may come from the intrinsic interest value of the material being learned, or may be created through the use of extrinsic incentives, such as praise, grades, stars, and so on (see Stipek 1993).

If students want to know something, they will be more likely to exert the necessary effort to learn it. This is why there are students who can rattle off the names and statistics relating to every player on their favorite sports team, but do not know their multiplication facts. Teachers can create intrinsic interest in material to be taught by arousing student curiosity, for example by using surprising demonstrations, by relating topics to students' personal lives, or by allowing students to discover information for themselves (Brophy 1987; Malone and Lepper 1988).

However, not every subject can be made intrinsically interesting to every student at all times. Most students need some sort of extrinsic incentive to exert an adequate level of effort on most school tasks. For example, studies of graded versus pass-fail college courses find substantially higher achievement in classes that give grades (Gold, Reilly, Silberman, and Lehr 1971; Hales, Bain, and Rand 1971). At the elementary level, informal incentives, such as praise and feedback, may be more important than the formal grading system (see Brophy 1987). One critical principle of effective use of classroom incentives is that students should be held accountable for everything they do. For example, homework that is checked has been found to contribute more to student achievement than homework that is assigned but not checked (Cooper 1989). Also, questioning strategies that communicate high expectations for students, such as waiting for them to respond (Rowe 1974) and following up with students who do not initially give full responses (Brophy and Evertson 1974), have been found to be associated with high achievement (see Good 1987).

Several methods of providing formal incentives for learning have been found to be instructionally effective. One practical and effective method of rewarding students for appropriate, learning-oriented behavior is home-based reinforcement (Barth 1979) through provision of daily or weekly reports to parents on student behavior. Another is group contingencies (Dolan, et al. 1992; Hayes 1976), in which the entire class or groups within the class are rewarded on the basis of the behavior of the entire group.

Cooperative learning methods (Slavin 1990b) involve students working in small learning groups to master academic material. Forms of cooperative learning that have consistently increased student achievement have provided rewards to heterogeneous groups based on the learning of their members. This incentive system motivates students to encourage and help one another to achieve. Rewarding students based on improvement over their own past performance has also been found to be an effective incentive system (Natriello 1987; Slavin 1980).

In addition to being a product of specific strategies designed to increase student motivation, incentive is also influenced by quality of instruction and appropriate levels of instruction. Students will be more motivated to learn about a topic that is presented in an interesting way, that makes sense to them, that they feel capable of learning. Further, a student's motivation to exert maximum effort will be influenced by their perception of the difference between their probability of success if they do exert themselves and their probability of success if they do not (Atkinson and Birch 1978; Slavin 1977, 1994). That is, if a student feels sure of success or, alternatively, of failure, regardless of his or her efforts, then incentive will be very low. This is likely to be the case if a lesson is presented at a level much too easy or too difficult for the student. Incentive is high when the level of instruction is appropriate for a student, so that the student perceives that with effort the material can be mastered, so that the payoff for effort is perceived to be great.

Instruction takes time. More time spent teaching a subject does not always translate into additional learning, but if instructional quality, appropriateness of instruction, and incentives for learning are all high, then more time on instruction is likely to pay off in greater learning.

The amount of time available for learning depends largely on two factors: Allocated time and engaged time. Allocated time is the time scheduled by the teacher for a particular lesson or subject and then actually used for instructional activities. Allocated time is mostly under the direct control of the school and teacher. In contrast, engaged time, the time students actually engage in learning tasks, is not under the direct control of the school or the teacher. Engaged time, or time-on-task, is largely a product of quality of instruction, student motivation, and allocated time. Thus, allocated time is an alterable element of instruction (like quality, appropriateness, and incentive), but engaged time is a mediating variable linking alterable variables with student achievement.

While allocated time must be an essential element in any model of classroom organization, research on this variable has found few consistent effects on student achievement. For example, research on hours in the school day and days in the school year has found few relationships between these time variables and student achievement (Frederick and Walberg 1980; Karweit 1989). The Beginning Teacher Evaluation Study found that allocated time in specific subjects had no effect on student achievement in those subjects when time was measured at the class level (Marliave, Fisher, and Dishaw 1978). On the other hand, research on engaged time generally finds positive relationships between the time students are on task and their achievement, but even with this variable, results are inconsistent (see Karweit 1989).

Studies of means of increasing student time on task generally go under the heading of classroom management research. Process-product studies (see, for example, Brophy and Good 1986) have established that teachers' use of effective management strategies is associated with high student achievement. However, several experimental studies focusing on increasing time on-task have found that it is possible to increase engaged time and still have no significant effect on student achievement (Slavin 1986; Stallings and Krasavage 1986; Wilson and Nachtigal 1986).

The QAIT model, whose elements were described in the previous sections, can also be conceptualized in terms of intermediate effects on time-related variables. Figure 1 depicts a model of how alterable elements of instruction might affect student achievement.

INSERT FIGURE 1

In Figure 1, two types of independent variables are presented: Student inputs and alterable variables. Student inputs refer to factors over which the school has little control in the short run: Student ability and those aspects of motivation to learn that students bring from home (as distinct from the motivation created by classroom practices). The alterable variables are the QAIT elements discussed earlier. Of course, student inputs are not immutable, but can be affected by classroom practices. For example, student aptitude to learn a specific lesson may be strongly influenced by background knowledge resulting from earlier instruction, by specific training in thinking, problem solving, or study skills, or by general intellectual stimulation or learning skills provided by the school. Student motivation to learn is also largely a product of past experiences in school. However, in the context of any given lesson, the student inputs can be considered fixed, while the alterable variables can be directly manipulated by the school or teacher.

The effects of the alterable variables on student achievement are held to be mediated by two time-related variables: Instructional efficiency and engaged time, or time-on-task. Instructional efficiency can be conceptualized as the amount of learning per time. For example, students will learn more in a ten-minute lesson high in instructional efficiency than in a lesson of similar length low in instructional efficiency. Engaged time is the amount of time students are actually participating in relevant learning activities, such as paying attention to lectures and doing assignments. Instructional efficiency is simply the inverse of Carroll's "time needed to learn," and engaged time is essentially his "time available for learning." Instructional efficiency and engaged time are multiplicatively related to student achievement; obviously, if either is zero, then learning is zero.

The QAIT model can be easily related to instructional efficiency and engaged time. Instructional efficiency is a product of the quality of instruction (e.g., organization and presentation quality of the lesson), appropriate levels of instruction (students have prerequisite skills but have not already learned the lesson), and incentive (students are interested in learning the lesson). Of course, aptitude and motivation also contribute to instructional efficiency for any given student. Engaged time is primarily a product of allocated time and incentive.

The relationship between improvements in each of the four alterable elements and effects on student achievement is held to be multiplicative. If any of the elements is at zero, learning will be zero. Above zero, the argument that the effects of the four elements are multiplicative rests in part on an assumption that effects of increasing each element are greatest at low levels and ultimately reach maximum or asymptotal levels (within a fixed amount of allocated time). For example, motivation to learn will reach a maximum in terms of affecting student achievement at some point. Effects of quality and appropriateness of instruction are similarly likely to reach a point of diminishing returns. Time on-task not only cannot be increased beyond 100% of time allocated, but it is doubtful whether increases beyond, say, 90% produce significant increases in learning. This may explain why several studies which produced substantial gains in time on-task have produced minimal effects on student achievement (see Emmer and Aussiker 1990; Slavin 1986).

The substantive implication of a multiplicative relationship among the QAIT elements is that it may be more effective to design instruction to produce moderate gains in two or more elements than maximize gains in only one. To increase a plant's growth, moderate increases in light, water, and fertilizer are likely to be more productive than large increases in only one of these elements. By the same token, substantial increases in any one element of the QAIT model, leaving all others unaffected, is likely to be less effective than more moderate, across-the-board improvements.

Another implication of the assumption that there is a point of diminishing returns in the achievement effect of each of the QAIT elements is that different types of programs will work differently in different settings depending on pre-existing levels of each. For example, a program focused on increasing time on-task is likely to be more effective in classrooms low on this variable than in those beginning at 80-90% levels. Highly motivated students may profit more from programs focusing on providing appropriate levels of instruction than from motivationally focused programs, and so on. Put another way, the relationship between each QAIT element and achievement is multiplicative, but the multiplier associated with improvements depends on where students began on each element. If, for example, quality of instruction is high but the level of instruction is inappropriate for many students, then the multiplier for increasing quality will be lower than that for increasing appropriateness.

The contention that the relationships between the QAIT variables, instructional efficiency, engaged time, and student achievement are multiplicative is pivotal to the model proposed here. In addition to implying that achievement will be zero if any of the alterable variables are zero, it also implies that while improving any one of the variables is likely to increase achievement arithmetically (up to a point of diminishing returns), improving more than one is likely to increase achievement geometrically. Since there are many random or uncontrolled factors in student achievement, and since achievement in any particular skill is so much a function of prior knowledge, ability, and motivation, it may be that for any new program to have a measurable effect on student mean achievement, it must improve multiple elements of instruction and therefore have a geometric effect on learning.

Effective teaching behaviors are critical to student achievement, but there are also key factors at the school level that also contribute to instructional effectiveness. There is a body of research on school effectiveness that has identified characteristics associated with outstanding achievement gains. Some of these are simply aggregations of the characteristics of effective teachers. For example, schools that have an orderly environment, high time on task, and frequent monitoring of student progress tend to produce higher achievement than other schools (Mortimore et al. 1988; Purkey and Smith 1983; Teddlie and Stringfield 1993). However, there are also many elements of effective schooling that go beyond what happens in individual classrooms. For example, more effective schools are ones that have a clear academic mission and focus. They have principals who are strong instructional leaders, are proactive initiators of change, and encourage participation of the whole staff in making change (Teddlie and Stringfield 1993).

Improvement in student achievement rarely comes about as a result of teachers and principals reading lists of effective teacher behaviors or school characteristics. On the contrary, effective school change usually comes about because schools and teachers adopt specific programs and practices that provide well validated teacher and student materials, staff development procedures, assessments, and so on (Slavin 1990c).

There are hundreds of well developed, replicable programs designed for specific subjects and grade levels (such as elementary math or high school biology). A listing of more than five hundred such programs is found in Educational Programs that Work (1987), a publication of the National Diffusion Network (NDN). The NDN is a program of the U.S. Department of Education designed to identify and help disseminate promising and replicable programs. The NDN programs are supposed to have evidence of effectiveness, and while the quality of evidence typically presented is often poor (see Klein 1992), there is at least some reason to believe that the programs are effective, and each is backed up by manuals, materials, and training services that make them relatively easy to replicate.

A few types of programs have particularly impressive evidence of effectiveness. One is cooperative learning methods, in which students work in small groups to help each other master academic material (Slavin 1990b). Cooperative methods incorporating group goals and individual accountability have been consistently more effective than traditional methods. Another is writing process methods (Hillocks 1984), in which students work in small groups to help one another develop as writers. Both cooperative learning and writing process methods have been used at all grade levels. Research has also supported the use of one-to-one tutoring for first graders who are at risk for school failure (Wasik and Slavin 1993). In particular, a program called Reading Recovery (Pinnell 1989) has been well researched and is widely used in the U.S. and elsewhere.

Recently, there has been a substantial focus in the U.S. on comprehensive school change programs. The most thoroughly evaluated and effective of these is a program called Success for All (Slavin et al. 1992, in press; Madden et al. 1993), a comprehensive restructuring of elementary schools to ensure that all children are successful from the early grades onward. Success for All uses research-based curriculum and instruction from prekindergarten to sixth grade, with an emphasis on cooperative learning. First graders who are at risk of reading failure may receive one-to-one tutoring by specially trained tutors. A comprehensive family support program builds positive relationships with parents and deals with such problems as truancy, vision and hearing problems, and behavior problems. Success for All may be particularly appropriate for Latin America because a version of its reading program exists in Spanish. The School Development Program (Comer 1988) is another comprehensive school change model that focuses primarily on building strong links between parents and schools.

The most important lesson of research on effective teaching, effective schools, and effective programs is that to make a substantial difference in student achievement, change must be comprehensive and broad-based and must affect what teachers do with their children on a day-to-day basis. Reforms that have attempted to intervene in only one aspect of curriculum, instruction, or school organization are rarely as effective as those that deal with many critical elements at once. The Success for All program demonstrates the extreme case of this principle; Success for All affects every aspect of elementary education, and this model has been considerably more effective than programs affecting only one aspect of instruction or school organization.

Change in fundamental school practices is not easy or quick. Effective change strategies must include extensive inservice, followup, and internal assessment. It must incorporate effective materials and have adequate resources and infrastructure.

The unit of change is the school and classroom; changes above this level may support local changes but do not guarantee improved outcomes. Change must take place one school at a time, yet national or regional policies can create the conditions under which school-by-school change can be promoted and maintained.

The situation of education in Latin America is highly diverse, both within and between Latin American nations. The prospects for effective changes are obviously quite different in schools with adequate basic resources and well-trained teachers than in schools lacking rudimentary facilities and qualified staff. Yet a few principles derived from research in the U.S. may apply.

First, the school must be the primary unit of change. Change will be most effective and cost-effective to the degree that it gives the professionals already in the schools effective programs and tools to do a better job. This implies a substantial investment in top-quality inservice. Preservice teacher training programs are important, of course, but schools must also have the capacity to train new staff members in the effective methods used in the school.

Second, change must be comprehensive. Change strategies must affect many aspects of instruction and school organization at once in order to magnify the effects.

Third, a continuing process of change must be set in motion. For example, school systems might introduce several proven or promising programs among a relatively small set of schools with a long-term plan of piloting, assessment, and expansion. A change process must start small and, once success is established, expand rapidly. Starting on too broad a scale risks failure from the outset.

Finally, the change process must respect local professionals and local conditions. No program, no matter how effective, will work in the same way everywhere. School staffs must have an opportunity to review possible programs and vote to adopt them and then to adapt them to their own local needs and resources.

This is an exciting time for educational reform throughout the world. Research is increasingly identifying replicable strategies for improving student achievement and other outcomes, and identifying means of introducing these strategies to schools. This paper has discussed some of the principles and practices of effective instruction and school organization with an eye toward informing policy makers concerned with improving schooling in Latin America, but the same basic principles would apply anywhere there is a commitment to educational reform.

* Robert Slavin is currently Director of the Early and Elementary School Program at the Center for Research on Effective Schooling for Disadvantaged Students at Johns Hopkins University. He received his B.A. in Psychology from Reed College in 1972, and his Ph.D. in Social Relations in 1975 from Johns Hopkins University. Dr. Slavin has authored or co-authored more than 140 articles and 14 books, including Educational Psychology: Theory into Practice (Allyn and Bacon 1986, 1988, 1991, 1994), School and Classroom Organization (Erlbaum 1989), Effective Programs for Students at Risk (Allyn and Bacon 1989), Cooperative Learning: Theory, Research, and Practice (Prentice-Hall 1990), and Preventing Early School Failure (Allyn and Bacon, forthcoming). He received the American Educational Research Association's Raymond B. Cattell Early Career Award for Programmatic Research in 1986, and the Palmer O. Johnson award for the best article in an AERA journal in 1988.

Abrami, P. C., L. Leventhal, and R. P. Perry. 1982. "Educational Seduction." Review of Educational Research 52: 446-462.

Anderson, L. M., N. L. Brubaker, J. Alleman-Brooks, and G. G. Duffy. 1985. "A Qualitative Study of Seatwork in First-grade Classrooms." Elementary School Journal 86: 123-140.

Anderson, L. M., C. Evertson, and J. Brophy. 1979. "An Experimental Study of Effective Teaching in First-Grade Reading-Groups." Elementary School Journal 79: 193-223.

Atkinson, J. W., and D. Birch. 1978. Introduction to Motivation. 2nd ed. New York: Van Nostrand.

Barr, R. 1992. "Teachers, Materials, and Group Composition in Literacy Instruction." Elementary School Literacy: Critical Issues. Eds. M. J. Dreher and W. H. Slater. Norwood, MA: Christopher-Gordon.

Barr, R., and R. Dreeben. 1983. How Schools Work. Chicago: University of Chicago Press.

Barringer, C., and B. Gholson. 1979. "Effects of Type and Combination of Feedback Upon Conceptual Learning by Children: Implications for Research in Academic Learning." Review of Educational Research 49: 459-478.

Barth, R. 1979. "Home-Based Reinforcement of School Behavior: A Review and Analysis." Review of Educational Research 49: 436-458.

Block, J. H., and R. B. Burns. 1976. "Mastery Learning." Review of Research in Education (Vol. 4). Ed. L. S. Shulman. Itasca, IL: F. E. Peacock.

Bloom, B. S. 1976. Human Characteristics and School Learning. New York: McGraw-Hill.

Brophy, J. E. 1979. "Teacher Behavior and Its Effects." Journal of Educational Psychology 71: 733-750.

---. 1981. "Teacher Praise: A Functional Analysis." Review of Educational Research 51: 5-32.

---. 1987. "Synthesis of Research on Strategies for Motivating Students to Learn." Educational Leadership 45: 40-48.

Brophy, J. E., and C. M. Everston. 1974. Process-Product Correlations in the Texas Teacher Effectiveness Study: Final Report (Research Report No. 74-4). Austin: Research and Development Center for Teacher Education, University of Texas.

Brophy, J. E., and T. L. Good. 1986. "Teacher Behavior and Student Achievement." Handbook of Research on Teaching. Ed. M. C. Wittrock. 3rd ed. New York: McMillan.

Carroll, J. B. 1963. "A Model of School Learning." Teachers College Record 64: 723-733.

---. 1989. "The Carroll Model: A 25-Year Retrospective and Prospective View." Educational Research 18: 26-31.

Cavanaugh, B. R. 1984. Effects of Interdependent Group Contingencies on the Achievement of Elementary School Children. Unpublished doctoral dissertation, University of Maryland.

Clark, C. M. 1987. "The Carroll Model." International Encyclopedia of Teaching and Teacher Education. Ed. M. J. Dunkin. New York: Pergamon.

Comer, J. 1988. "Educating Poor Minority Children." Scientific American 259: 42-48.

Cooley, W. W., and G. Leinhardt. 1980. "The Instructural Dimensions Study." Educational Evaluation and Policy Analysis 2: 7-35.

Cooper, H. 1989. "Synthesis of Research on Homework." Educational Leadership 47.3: 85-91.

Crooks, T. J. 1988. "The Impact of Classroom Evaluation Practices on Students." Review of Educational Research 58: 438-481.

Dolan, L. J., S. G. Kellam, C. H. Brown, L. Werthamer-Larsson, G. W. Rebok, L. S. Mayer, J. Laudolff, J. S. Turkkan, C. Ford, and L. Wheeler. 1992. "The Short-Term Impact of Two Classroom-Based Preventive Interventions on Aggressive and Shy Behaviors and Poor Achievement. Journal of Applied Developmental Psychology.

Dunkin, M. 1978. "Student Characteristics, Classroom Processes, and Student Achievement." Journal of Educational Psychology 70: 998-1009.

Emmer, E. T., and A. Aussiker. 1990. "School and Classroom Discipline Programs: How Well Do They Work?" Students Discipline Strategies. Ed. O. C. Moles. Albany: State University of New York Press.

Evertson, C. M., J. K. Folger, C. Breda, and C. Randolph. 1990. "How Does Inservice Training Affect Teachers' Instruction in Small, Regular, and Regular/Aide Classes?" Paper presented at the annual convention of the American Educational Research Association, Boston.

Frederick, W., and H. Walberg. 1980. "Learning as a Function of Time." Journal of Educational Research 73: 183-194.

Gold, R. M., A. Reilly, R. Silberman, and R. Lehr. 1971. "Academic Achievement Declines Under Pass-Fail Grading." Journal of Experimental Education 39: 17-21.

Good, T. 1987. "Teacher Expectations." Talks to Teachers. Eds. D. Berliner and B. Rosenshine. New York: Random House. 159-200.

Good, T., and S. Marshall. 1984. "Do Students Learn More in Heterogeneous or Homogenous Groups?" The Social Context of Instruction: Group Organization and Group Processes. Eds. P. Peterson, L. C. Wilkinson, and M. Hallihan. New York: Academic Press. 15-38.

Guskey, T. R., and S. L. Gates. 1985. "A Synthesis of Research on Group-Based Mastery Learning Programs." Paper presented at the annual convention of the American Educational Research Association, Chicago.

Gutiérrez, R., and R. E. Slavin. 1992. "Achievement Effects of the Nongraded Elementary School: A Best-Evidence Synthesis." Review of Educational Research 62: 333-376.

Hales, L. W., P. T. Bain, and L. P. Rand. 1971. "An Investigation of Some Aspects of the Pass-Fail Grading System." Paper presented at the annual meeting of the American Educational Research Association, New York.

Hartley, S. S. 1977. "Meta-Analysis of the Effects of Individually Paced Instruction in Mathematics." Dissertation Abstracts International 38: 4003A. (University Microfilms No. 77-29, 926).

Hayes, L. 1976. "The Use of Group Contingencies for Behavioral Control: A Review." Psychological Bulletin 83: 628-648.

Hiebert, J., D. Wearne, and S. Taber. 1991. "Fourth Graders' Gradual Construction of Decimal Fractions During Instruction Using Different Physical Representations." Elementary School Journal 91: 321-341.

Hillocks, G. 1984. "What Works in Teaching Composition: A Meta-Analysis of Experimental Treatment Studies." American Journal of Education 93: 133-170.

Horak, V. M. 1981. "A Meta-Analysis of Research Findings on Individualized Instruction in Mathematics." Journal of Educational Research 74: 249-253.

Kallison, J. M. 1986. "Effects of Lesson Organization on Achievement." American Educational Research Journal 23: 337-347.

Kaplan, R. M., and G. C. Pascoe. 1977. "Humorous Lectures and Humorous Examples: Some Effects Upon Comprehension and Retention." Journal of Educational Psychology 69: 61-65.

Karweit, N. L. 1989. "Time and Learning: A Review." School and Classroom Organization. Ed. R. E. Slavin. Hillsdale, NJ: Erlbaum.

Klein, S. S. 1992. "A Framework for Redesigning an R and D-Based National Education Dissemination System in the United States." Knowledge: Creation, Diffusion, Utilization 13.3: 256-286.

Kozma, R. 1991. "Learning With Media." Review of Educational Research 61.2: 179-211.

Kulik, J. A., and C. L. Kulik. 1988. "Timing of Feedback and Verbal Learning." Review of Educational Research 58: 79-97.

Land, M. L. 1987. "Vagueness and Clarity." International Encyclopedia of Teaching and Teacher Education. Ed. M. J. Dunkin. New York: Pergamon.

Leighton, M. S., and R. E. Slavin. 1988. "Achievement Effects of Instructional Pace and Systematic Instruction in Elementary Mathematics." Paper presented at the annual convention of the American Educational Research Association, New Orleans.

Leinhardt, G., and W. Bickel. 1987. "Instruction's the Thing Wherein to Catch the Mind That Falls Behind." Educational Psychologist 22: 177-207.

Leinhardt, G., and A. Pallay. 1982. "Restrictive Educational Settings: Exile or Haven?" Review of Educational Research 52: 557-578.

Madden, N. A., and R. E. Slavin. 1983. "Mainstreaming Students With Mild Academic Handicaps: Academic and Social Outcomes." Review of Educational Research 84: 131-138.

Madden, N. A., R. E. Slavin, N. L. Karweit, L. J. Dolan, and B. A. Wasik. 1993. "Success for All: Longitudinal Effects of a Restructuring Program for Inner-City Elementary Schools." American Educational Research Journal 30: 123-148.

Maddox, H., and E. Hoole. 1975. "Performance Decrement in the Lecture." Educational Review 28: 17-30.

Malone, T., and M. Lepper. 1988. "Making Learning Fun: A Taxonomy of Intrinsic Motivation for Learning." Aptitude Learning and Instruction, Vol. III: Cognitive and Affective Process Analysis. Eds. R. Snow and M. Farr. Hillsdale, NJ: Erlbaum.

Marliave, R., C. Fisher, and M. Dishaw. 1978. Academic Learning Time and Student Achievement in the B-C Period. Far West Laboratory for Educational Research and Development. Technical Note V-29.

Mayer, R. E., and J. K. Gallini. 1990. "When Is an Illustration Worth Ten Thousand Words?" Journal of Educational Psychology 82: 715-726.

Melton, R. F. 1978. "Resolution of Conflicting Claims Concerning the Effect of Behavioral Objectives on Student Learning." Review of Educational Research 48: 291-302.

Mortimore, P., P. Sammons, L. Stoll, D. Lewis, and R. Scob. 1988. School Matters: The Junior Years. Somerset, England: Open Books.

National Diffusion Network. 1987. Educational Programs That Work. 13th ed. Longmont, CO: Sopris West.

Natriello, G. 1987. "The Impact of Evaluation Processes on Students." Educational Psychologist 22: 155-175.

Oakes, J. 1985. Keeping Track: How Schools Structure Inequality. New Haven, CT: Yale University Press.

---. 1987. "Tracking in Secondary Schools: A Contextual Perspective." Educational Psychologist 22: 129-153.

Pinnell, G. S. 1989. "Reading Recovery: Helping At-Risk Children Learn to Read." Elementary School Journal 90: 161-182.

Pressley, M., E. Wood, V. E. Woloshyn, V. Martin, A. King, and D. Menke. 1992. "Encouraging Mindful Use of Prior Knowledge: Attempting to Construct Explanatory Answers Facilitates Learning." Educational Psychologist 27: 91-109.

Purkey, S. C., and M. S. Smith. 1983. "Effective Schools: A Review." Elementary School Journal 83: 427-452.

Rosenshine, B. V., and R. J. Stevens. 1986. "Teaching Functions." Third Handbook of Research on Teaching. Ed. M. C. Wittrock. Chicago: Rand McNally.

Rowan, B., and A. Miracle. 1983. "Systems of Ability Grouping and the Stratification of Achievement in Elementary Schools." Sociology of Education 56: 133-144.

Rowe, M. B. 1974. "Wait-Time and Rewards as Instructional Variables, Their Influence on Language, Logic, and Fate Control: Part One-Wait-Time." Journal of Research in Science Teaching 11: 81-94.

Slavin, R. E. 1977. "A New Model of Classroom Motivation." Paper presented at the annual convention of the American Educational Research Association, New York.

---. 1980. "Effects of Individual Learning Expectations on Student Achievement." Journal of Educational Psychology 72: 520-524.

---. 1983a. Cooperative Learning. New York: Longman.

---. 1983b. "When Does Cooperative Learning Increase Student Achievement?" Psychological Bulletin 94: 429-445.

---. 1984. "Component Building: A Strategy for Research-Based Instructional Improvement." Elementary School Journal 84: 255-269.

---. 1985. "Team-Assisted Individualization: Combining Cooperative Learning and Individualized Instruction in Mathematics." Learning to Cooperate, Cooperating to Learn. Eds. R. E. Slavin, S. Sharan, R. Hertz-Lazarowitz, C. Webb, and R. Schmuck. New York: Plenum. 177-209.

---. 1986. "The Napa Evaluation of Madeline Hunter's ITIP: Lessons Learned." Elementary School Journal 87: 165-171.

---. 1987a. "A Theory of School and Classroom Organization." Educational Psychologist 22: 89-108.

---. 1987b. "Ability Grouping and Student Achievement in Elementary Schools: A Best-Evidence Synthesis." Review of Educational Research 57: 347-350.

---. 1987c. "Mastery Learning Reconsidered." Review of Educational Research 57: 175-213.

---. 1989. "Achievement Effects of Substantial Reductions in Class Size." School and Classroom Organization. Ed. R. E. Slavin. Hillsdale, NJ: Erlbaum. 247-257.

---. 1990a. "Achievement Effects of Ability Grouping in Secondary Schools: A Best Evidence Synthesis." Review of Educational Research 60.3: 471-499.

---. 1990b. Cooperative Learning: Theory, Research and Practice. Englewood Cliffs, NJ: Prentice-Hall.

---. 1990c. "On Making a Difference." Educational Researcher 19.3: 30-34.

---. 1994. Educational Psychology: Theory into Practice. 4th ed. Boston: Allyn and Bacon.

Slavin, R. E., and N. L. Karweit. 1985. "Effects of Whole-Class, Ability-Grouped, and Individualized Instruction on Mathematics Achievement." American Educational Research Journal 22: 351-367.

Slavin, R. E., M. Leavey, and N. A. Madden. 1984. "Combining Cooperative Learning and Individualized Instruction: Effects on Student Mathematics Achievement, Attitudes, and Behaviors." Elementary School Journal 84: 409-422.

Slavin, R. E., N. A. Madden, L. J. Dolan, B. A. Wasik, S. Ross, and L. Smith. In press. "Whenever and wherever we choose...: The repliation of Success for All." Phi Delta Kappa.

Slavin, R. E, N. A. Madden, N. L. Karweit, L. Dolan, and B. A. Wasik. 1992. Success for All: A Relentless Approach to Prevention and Early Intervention in Elementary Schools. Arlington, VA: Educational Research Service.

Slavin, R. E., N. A. Madden, and M. Leavey. 1984. "Effects of Team Assisted Individualization on the Mathematics Achievement of Academically Handicapped and Non-Handicapped Students." Journal of Educational Psychology 76: 813-819.

Smith, L. R., and M. L. Cotten. 1980. "Effect of Lesson Vagueness and Discontinuity on Student Achievement and Attitudes." Journal of Educational Psychology 72: 670-675.

Stallings, J. A., and D. Kaskowitz. 1974. Follow-Through Classroom Observation Evaluation 1972-73. Menlo Park, CA: Stanford Research Institute.

Stallings, J., and E. M. Krasavage. 1986. "Program Implementation and Student Achievement in a Four-Year Madeline Hunter Follow Through Project." Elementary School Journal 87: 117-138.

Stipek, D. J. 1993. Motivation to Learn: From Theory to Practice. 2nd ed. Boston: Allyn and Bacon.

Teddlie, C., and S. Stringfield. 1993. Schools Make a Difference: On Lessons Learned From a 10-Year Study of School Effects. New York: Teachers College Press.

Wasik, B. A., and R. E. Slavin. 1993. "Preventing Early Reading Failure With One-To-One Tutoring: A Review of Five Programs." Reading Research Quarterly 28: 178-200.

Wilson, R. R., and P. Nachtigal. 1986. Final Report: Cotopaxi/Westcliffe Follow Through Project. Office of Educational Research and Improvement, Grant No. 400-81-0039.

Word, E., J. Johnston, H. P. Bain, B. D. Fulton, J. B. Zaharias, M. N. Lintz, C. M. Achilles, J. Floger, and C. Breda. 1990. Student/Teacher Achievement Ratio (STaR): Tennessee's k-3 Class Size Study, Final Report. Nashville: Tennessee State Department of Education.