Workshop on Understanding and Promoting Knowledge Accumulation in Education:
Tools and Strategies for Education Research

Day 2 – June 30, 2003

Agenda Item: Ways of Taking Stock: Replication, Scaling up, Meta-Analysis,
Professional Consensus-Building - Dr. Robert Slavin

DR. ROBERT SLAVIN: Well, good morning, I’m here to talk with you today about the scaling up of Success For All. I’m going to assume that you’ve heard of it, this is an elementary, I guess now an elementary and middle school reform program operating in about 1500 schools around the U.S. in 48 states that is focused on a very high level of professional development, use of systematic phonics and supports for phonemic awareness and children’s reading development, leading into extensive use of cooperative learning and more inquiry oriented activities in the upper grades, and emphasis on prevention, early intervention, trying to make sure the kids get off to a good start in school, and engaging parents and community actors in support of children’s success in identifying problems early and providing intensive assistance for children who are having difficulties.

Success For All began in 1987 in one school in Baltimore and then over the years has been expanding and therefore we’ve had the experience of having to take an idea from a very early stage and move it along into practice at a scale that matters because our intention from the beginning was to make a difference at the policy level and therefore you have to be working in not just a few places and publish a few articles and so on and be done with it. But if you’re serious about this you’ve got to show how these ideas can play out on a broader scale. But I thought today, breaking ordinary form, I would go even further back before 1987 and talk a little bit about the intellectual history that led to Success For All and then pick up with kind of where we went with that history.

Long before Success for All, going back to my first research grant of $2,000 dollars as a junior in college, there was cooperative learning, and in some ways I spent all of my early career, most of my early career, focusing on cooperative learning strategies in which kids work in small learning teams. And it so happened that about the time that I was getting started in this and working from the beginning with my wife, Nancy Madden, and other people, starting at Johns Hopkins University in the early ‘70’s, there were other groups of people also working on cooperative learning in various places and designing a number of different kinds of cooperative learning strategies. A very exciting time really in terms of whole new ways of thinking about how classrooms could be organized.

The cooperative learning was also quite unique in that here was something that was used from fairly early on, on a pretty broad scale, but it brought forth an enormous flood of studies, of experimental studies, studies that compared experimental and control groups, a small proportion but still a significant number of them using random assignment methods and developed really a pretty strong set of principles about what it takes to make cooperative learning work for academic achievement. And different investigators came to this in different ways but what they generally found was that cooperative learning in which you simply have kids sitting together, helping each other out, working on projects without a great deal of structure was ineffective for improving student achievement. What tended to happen in those groups is that one kid would do all the work and the other kids would watch, or one kid would get the answer and give the answer to the other kids, which was not beneficial either to that kid or to the kids who received the answer. And so the natural tendency, the way in which cooperative learning was often interpreted by individual teachers who’d kind of heard of it and made it up, made up their own strategies, was really shown pretty consistently to be ineffective.

In contrast the forms of cooperative learning that worked in terms of improving student achievement at least were ones in which there was more structure to the groups in which the purpose perceived by the members of the group was to try to make sure that all members of the group had mastered the material being studies. A variety of ways to do this, it could be working on projects where each person had a unique portion of the project that they were responsible for and others could help them but the individual kid still had to do it themselves. Or more frequently that there were cooperative learning strategies in which individual children had to how their own personal knowledge of the material being studied without the help of their peers and then the group benefited in some way from the good performance of their group mates, so that the motivation for the children involved in cooperative learning was to try to tutor and assist one another to learn the material rather than to simply give each other the answer, rather than to simply do the task. So many researchers, many reviewers, came to the similar conclusions that cooperative learning required some kind of a group goal, some kind of individual accountability, some kind of structure to enable children to succeed.

One thing also very interesting in cooperative learning is that with this enormous amount of research and the very rapid increase in use of cooperative learning strategies through the ‘70’s and ‘80’s and into the ‘90’s, there were very few people who didn’t like it, this was something very popular among researchers as well as among practitioners, and so this was not a case in which you had different points of view and there were people who were severely opposing this strategy, but rather there was general feeling that cooperative learning was probably a good thing and again, general consensus about the forms of cooperative learning that were likely to be affected.

And so you could say well, what a terrific story of research influencing practice on a broad scale. If you said that you would be a fool. Because the reality that we began to see and people doing good quality qualitative research and quantitative observation research began to see was that while the research being done by researchers on cooperative learning would tend to provide more structure and study the conditions under which cooperative learning was effective and see that those conditions were replicated, in the more normal course of things cooperative learning meant kids sitting at tables, helping each other from time to time, working on projects in which one kid did the work and the other kids watched, or having one kid get the answer and tell the other kids the answers. In other words the things that from the very early stages, by the late ‘70’s, it was really known among researchers in this field that that wasn’t going to work, by the mid ‘80’s it was really quite well accepted that that wasn’t going to work, and yet if you go out today, or in the mid ‘80’s or in the mid ‘90’s or whenever you like, what you see teachers doing was nothing like what the research was saying was essential. And so here you had a dilemma, and really kind of an interesting observation because here’s a situation in which this very attractive idea had been pushed to quite extensive degree and yet the fundamentals were not there.

This is really the place that our group at Johns Hopkins got to in the mid ‘80’s was great, happiness with what could be achieved using cooperative learning strategies, and other strategies as well, but a feeling about how can you make this happen on a broad scale in a form that is true to what the research says works. And that’s why we began Success For All in many fundamental ways, there’s other parts of this story that are interesting, I mean obviously we for a long time had a very strong concern about at risk children and schools serving at risk children and a lot of work was in the inner city or other high poverty schools, and we knew that it was going to be necessary to do a lot more than just cooperative learning to see that children were going to be successful. We knew that we had to have a focus on prevention and early intervention to see that kids weren’t so far behind in third grade or whatever grade you pick before you would start implementing effective practices.

But a lot of the intellectual background to Success For All came from this experience with cooperative learning where we felt as though it was going to be necessary to design programs that had a great deal of imbedded professional development that was going to have to provide materials for children, manuals for teachers, videos showing how this stuff was going to work, extensive, not only training, but follow-up and coaching to help teachers implement complex strategies. Something that was going to enable teachers to genuinely change their behavior and keep it changed over a long period of time. This forced us in following along this line of thinking, it forced us to some hard decisions in terms of designing a program that could be replicated from the outset, doing careful evaluations of that program, but also being willing to build a staff and build an operation to support quality implementation of a very complex program. And so we did that bit by bit, working within Johns Hopkins University until 1998, and then got too big and split off a separate non-profit foundation called the Success For All Foundation, to provide this level of training, materials, follow-up, and so on as well as continuing the research.

There’s a great deal that we’ve learned in this process and what we’ve learned is really quite daunting if you will about the difficulties of doing scale up and maintaining quality at each level of scale up. I think we’ve been very successful in doing this both in the quality side and in the scale up side, but it isn’t easy. One of the things we’ve learned is it’s very difficult to rely on partnerships, for example, you almost have to do this yourself, it’s hard to rely on part time or stringers, a lot of people think well you just take the best teachers out of school during the summer and they can do training for others. Not so. To do a quality job of implementation requires a great deal of training and monitoring of your own staff and so the concept that you could have people who are just working a few days a year and then provide them adequate training and follow-up is just not the case.

There’s got to be a constant focus on quality of implementation. The problem in education is that no matter how good the research base is things will slip away from outcomes towards process. Educators will proceduralize anything and you’ve got to watch, your own people will proceduralize things if you’re not watching carefully and move away from the purpose of what you’re doing toward the outward appearance of what you’re trying to do. And this is an endless process, one that we’re still engaged in to this day, is trying to find a balance between specificity and prescription about what’s going to happen in schools and accommodation to the unique characteristics and needs of each individual school when you’re working at a very large scale, not wanting to be so flexible that you wind up with implementation of nothing, and not wanting to be so structured that you know that you get a bad reputation as the SFA police, which we hear.

But to engage this issue is essential, you can’t just say well I don’t want to be the FSA police so I’m just going to come in and design something from scratch in every school according to its local needs because what you will do is design, not to put too fine a point on it, the coalition of essential schools, which study after study after study found nothing to be implemented in. But the essential concept is that scale up is possible but it’s not for the faint of heart.

Thank you very much.