Zakira and Amyra are reading a book together. It is lunchtime and they are sitting on the ledge that surrounds the school fish pond. Zakira, 9, closes the book and Amyra, 5, looks up with a smile. They are buddies.
The peace garden at Archbishop Sumner Church of England Primary School, in South London, is a bright spot with a quietly bubbling pond, benches and bamboo. It is separate from the main playground - the shouts of children playing tag and scrambling over the giant climbing frame can barely be heard.
"Some people just don't want to be reading buddies," Zakira says, with a maturity that perhaps comes from working her way through shelves of books. "They just want to play. But I think it's a good opportunity for me because then I can read the book to the little ones and they tell me how they feel about the book and learn more words. I enjoy doing it."
The girls are one of five pairs in the school's reading recovery scheme. The project matches volunteers from Year 5 with children in Year 1 who teachers feel would benefit from some extra reading time.
It is a neat example of peer-to-peer tutoring and illustrates how, when used well, the method can deliver great results. The younger child receives one-to-one tuition, while the older student boosts their own learning, takes responsibility and develops skills such as communication.
Similarly, Reading Rockets, a national literacy initiative in the US, promotes "partner reading" as an effective classroom strategy. As with the buddy scheme, this method involves pairing students. Teachers determine that one child, the "player", needs help in a specific area, and they are matched with another child who acts as their "coach". Roles are swapped regularly, giving every child the chance to coach others.
The programme is based on studies from Vanderbilt Kennedy Center for Research on Human Development in Tennessee, where Peer-Assisted Learning Strategies (Pals) was developed. Pals was designed to help teachers include children with learning disabilities in whole-class learning. However, it has proven to be effective for students of all abilities, and a meta-analysis from the University of Baltimore has found that the impact of Pals and other paired programmes on struggling readers is similar to that of one-to-one tutoring by teachers.
The usefulness of peer tutoring is not limited to primary school students - or to reading, for that matter. It has been found to be beneficial for older children, too, and in all sorts of subjects. In fact, the Education Endowment Foundation (EEF) - an independent grant-making charity set up in England in 2011 and given more than pound;200 million of government funding to allocate over 15 years - says that peer tutoring can lead to children making six months more progress in a year than expected.
The idea is relatively easy to implement, providing that student tutors are trained properly - for example, they must know the right questions to ask to help their tutees build a real understanding, rather than declaring, "That's wrong, this is the right answer".
Matching the children carefully is also important. Pairing different age groups works better than matching children within a class, and a modest difference in ability is more productive. For example, the most able 10- year-old should be matched with the most able eight-year-old, rather than pairing the most proficient 10-year-old with the least capable eight-year- old, or vice versa.
A 1997 analysis of peer-tutoring studies in California shows that students are more likely to admit that they are having problems with their work to a fellow student than to their teacher. "Disadvantaged students often found their peers more approachable than teachers for extra assistance, perceiving their teachers as too busy," the report says. "Peer tutoring and mentoring have the potential to alter the low-achiever's self- perception as an incompetent learner.
"Working with a tutor or mentor affords the learner a non-threatening way by which to learn how to set and accomplish goals, reason through dilemmas and solve problems."
From theory to practice
So, if a number of studies show that peer tutoring is one of the most effective methods of improving students' results, why are some educators still so sceptical about it?
On the face of it, peer tutoring is the perfect tool. Cheap and easily adaptable, it is included in the EEF's list of the 21 methods that have the biggest impact for the lowest cost. But the problem is getting peer tutoring to work in schools; it is not enough to simply find an effective method, evaluate it and then shout about how good it is.
Kevan Collins, chief executive of the EEF, says that it is not so much what you do, but the way you do it.
"Implementation is noble," he says. "It is not bureaucracy. It's not just `do as you're told'. It is about understanding, reflecting, giving something the best chance you can and evaluating whether it's working. Most things will make a bit of difference. The question is what will make the best use of the resources and time you've got."
At Archbishop Sumner, teachers are clearly doing something right. In summer 2012, 28 students at the primary school took key stage 2 Sats, and 96 per cent achieved the expected level in English for 11-year-olds. But is it possible to find out exactly what the buddy reading initiative contributed to this success?
Nick Timlin, special educational needs coordinator and reading recovery teacher at Archbishop Sumner, is confident that peer tutoring is having an impact. "Reading buddies is (a scheme) I've seen working before. I've seen the effects and done some research into it," he says. "We do not collect data separately for this project, but we do look at children's starting levels in reading, and I talk to the class teacher during catch-up meetings and look at the child's progress."
As well as the reading project, several other peer-to-peer schemes covering subjects from art to social skills are running at Archbishop Sumner, where a third of students receive free school meals. All these interactions may help children such as Amyra to increase their test scores, but they also foster a sense of belonging, encourage responsibility and an awareness of what it is like to be a teacher in the older children, and give younger children something to look forward to in their later years at the school. It is possible that other, immeasurable qualities of peer tutoring, such as friendship or pride, also have an impact on learning.
Trying to unpick Zakira's exact contribution to Amyra's progress in reading is impossible. And pinning down exactly what works or adds value is complicated for the glaringly obvious reason that schools are not research labs.
`We know it works'
Andy Wiggins, from Durham University's Centre for Evaluation and Monitoring, who is principal investigator of a shared maths project in 83 English primary schools in Worcester, Medway, Leeds and Durham, says that schools do not have to do the research because it has already been done. His investigation explores how best to use peer tutoring in maths lessons; the schools are also taking part in research into how best to train large numbers of teachers to use the technique in maths.
"The whole problem with educational research is that enthusiasts develop something and, of course, it works because it's their enthusiasm that makes it work," Wiggins says. "What's really difficult is how to transfer that into a model that schools can pick up and use."
There is a tension between wanting everyone to do something in a way that has been tested and shown to work, and giving teachers the confidence to adapt it for their students. "Schools don't need to collect data on peer tutoring," Wiggins says, "because we know it works. We can compare 3,200 children with a control group, which gives you a really reliable comparison. But if you, as an individual teacher, looked at your class on a rainy Wednesday afternoon, who would you compare them with? You could test before and after and see if they make progress, but they could make progress anyway. How do you know what's making the difference? Hence we do it on a research basis and that gives us an idea of the difference it makes to the average child.
"Of course, the average child doesn't exist. Some children will make much better progress and some will make negative progress, but it's about shifting the whole herd. Changes in individuals are much more difficult to measure." The roots of peer tutoring can be traced back to 1795, when Scotsman Andrew Bell published an account of the educational system he had seen in India, where children taught each other the alphabet by drawing in the sand.
The monitorial system, where teachers taught older children and older children taught younger ones, became popular in the 19th century because it was cheap, but it fell out of favour as the parents of monitors objected to their children losing learning time. The method has moved in and out of fashion over the decades, but in the 1960s peer tutoring and peer mentoring really took hold, and have remained popular ever since.
Keith Topping, professor of educational and social research at the University of Dundee, became interested in peer tutoring when he worked as an educational psychologist in Kirklees, West Yorkshire. He looked at ways to involve parents in helping their children with reading. Some of the methods were also suitable for peer tutoring and he found that after the national curriculum was introduced in 1988, teachers' interest in parental involvement waned but peer tutoring hung on. Later, Topping moved into academia and has since published scores of papers and books on the subject. His work in Fife, undertaken with colleagues at Durham University, has contributed to peer tutoring's current popularity. In a large-scale study led by Topping, 129 of the 145 primary schools in the Scottish county introduced peer tutoring, and 12 variations of peer tutoring were compared. Children were peer tutored either once a week or three times a week, in reading, maths or reading and maths. Half were tutored by classmates and the rest by older children.
The study ran from 2006 to 2008 and found that cross-age tutoring had more impact than same-age tutoring; holding sessions three times a week had no greater impact than doing so once a week; and using the method in both reading and maths improved performance slightly more than using it in only one of the subjects.
"The thing to look for in current research is to (ask to) what extent does it add value?" Topping says. "It has to be answering a question that has not already been answered. The question `Does peer tutoring work?' has long since been answered. The answer is yes. What research is doing now is trying to unpick the conditions under which certain kinds of peer tutoring work."
If the research shows that peer tutoring works, why not simply invest lots of money in getting everyone to do it? The question can be answered in three words: Assessment for Learning. AfL remains one of the most effective ways of improving students' results yet conceived. However, in England - where an AfL programme was rolled out as a priority under the now defunct National Strategies for literacy and numeracy - the result has been a "tragedy", according to one of its foremost proponents.
Dylan Wiliam, emeritus professor of educational assessment at the University of London's Institute of Education, told TES last year that the principles behind AfL had not been properly understood. Under the roll- out, it turned from a way of encouraging students to "own" their learning, through self- and peer-assessment, into a way of measuring progress in fine detail. The fear is that peer tutoring could also falter because of the difficulties of conveying a deep understanding of the proven theory to all the teachers who will be implementing it.
So what is the best way to find a balance between insisting that everyone follows strict guidance without necessarily comprehending it (meaning that peer tutoring may not work very well) and giving people enough freedom to reach a deeper understanding by tinkering with the idea (which may also affect how well it works)? Researchers' enthusiasm can overcome this difficulty when they are working directly with schools, but how many teachers can a researcher realistically meet?
"Our feeling is that you do have to keep fidelity (to the programme being implemented) fairly high," Wiggins says. "We've adopted a model where we can't go into schools to make sure they are doing it right.
"Instead we have four consultants in place, one in each (education) authority, who are experts not in peer tutoring but in supporting and training teachers. They are the relevant people if we want to make things happen on the ground, and that's what we're testing."
Although teachers have every right to be sceptical about the research, they should be wary of dismissing peer tutoring as just the latest "fad". On the other hand, they are also perfectly placed to ask awkward questions that could prompt research to move in a new direction.
As the EEF's Collins says: "When a teacher says, `It didn't work with my children', I would challenge them. I would say, `Have you given it the best chance of being successful?'
"If they have met that professional obligation but say, `It didn't work for our kids, therefore we're trying something else', that seems to me to be a very professional journey."
Everything you need to know about peer learning from TES Connect
TES Connect has produced an online module of resources, research videos, articles and teaching materials to help you get the best results from peer-to-peer learning.
Everything you need to know has been collected in one place, from basic advice on how to get started to detailed explanations and information on more advanced techniques in paired reading and paired maths.
We have also recruited two globally significant experts in the field to share their content and their expertise.
Keith Topping is professor of educational and social research at the University of Dundee, Scotland. His main research focus is peer learning (including peer tutoring, cooperative learning and peer assessment).
Allen Thurston is professor of education and director of the Centre for Effective Education at Queen's University Belfast, Northern Ireland. His work focuses on how to promote more positive educational outcomes for students from disadvantaged backgrounds and how to promote social inclusion for all children.
We have also worked with grant-making charity the Education Endowment Foundation, which has concluded from research that peer-to-peer learning is one of the most cost-effective interventions available to teachers.