Cooperative Learning: The Evidence, Structures, and Classroom Design

Cooperative learning is one of the best-evidenced ideas in classroom research and one of the most often wasted. The research effect sizes are real; the practice that produces them is demanding. A badly run cooperative lesson is indistinguishable from group work in which one student does the thinking and the others put their names on the product. The distance between the research and the average classroom isn't the theory — it's the design rules. This evidence review walks through what cooperative learning actually means in the research, what the effect sizes are, and the design decisions that separate the structures that produce the EEF's +5 months from the ones that produce nothing at all.

What cooperative learning actually is

Cooperative learning is structured small-group learning organised around two non-negotiable features: positive interdependence, and individual accountability. Positive interdependence means the group succeeds only if every member contributes; individual accountability means every member's contribution is separately visible.

It is not the same as group work. Most classroom "group work" fails one or both of these tests. A poster produced by four students, one of whom did the work, is group work. A problem-solving sequence in which each student has a role, the group submits a joint answer, and each student then sits a short individual check tied to the same material, is cooperative learning. The architecture is the difference.

Where cooperative learning came from

The theoretical roots are in social interdependence theory (Morton Deutsch, and later David and Roger Johnson), cognitive-developmental theory, and behavioural learning theory. School-based cooperative learning was systematised by Johnson and Johnson in the 1970s onward, and by Robert Slavin's Student Teams-Achievement Divisions (STAD) and Teams-Games-Tournaments (TGT) work at Johns Hopkins. Later traditions — Kagan's structures, complex instruction, reciprocal teaching — all build on the same basic architecture.

What the research actually shows

Johnson and colleagues' meta-analytic work found significant positive effects across eight distinct cooperative methods. Kyndt and colleagues' later classroom-only meta-analysis of 65 studies found positive effects on achievement, attitudes, and perceptions, with moderation by subject, age, and culture. A synthesis of the Kyndt evidence reported achievement effects around ES ≈ 0.54 and attitude effects around ES ≈ 0.15. The Education Endowment Foundation's modern estimate is roughly +5 months of additional progress on standardised measures.

Large trials also show that effects depend on context. A 129-school randomised controlled trial of cooperative paired reading reported effect sizes in the region of 0.2 — a real and useful effect for a low-cost intervention, smaller than the classroom-only meta-analytic average because the trial ran in the wild rather than in optimised conditions. Later transfer of the same programme into high schools was not universal, another reminder that cooperative learning effects depend on the specific structure, age phase, and task.

The aggregate verdict: cooperative learning is well-evidenced, with effects at the upper end of the mainstream evidence base when the structures are faithfully implemented. The effects depend heavily on how the group is organised, not on the fact that pupils are sitting together.

The five design principles

Johnson and Johnson's widely used formulation names five features of cooperative learning as necessary for the effect. Missing any one predicts underperformance.

• Positive interdependence. A genuine reason for students to rely on each other. Jigsaw, where each student has a unique piece of the information the group needs, is one clean version.
• Individual accountability. Every student's contribution is visible. A common version: group product plus individual quiz or oral check tied to the same material.
• Face-to-face promotive interaction. Students actually talk, explain, debate. This is the cognitive work.
• Social skills taught explicitly. Turn-taking, explanation, disagreement-without-conflict are not automatic. They are taught with sentence stems, modelled, and practised.
• Group processing. A short reflection at the end: what did we do well, what do we change next time.

When teachers see cooperative learning fail, it is almost always because one of these is absent. Groups without positive interdependence are just shared-table individual work; groups without individual accountability are social loafing machines; groups without taught social skills revert to whoever is confident.

Five structures that hit the design principles cleanly

If the principles sound abstract, the structures make them concrete.

• Think-Pair-Share. Individual thinking time (accountability), then partner discussion (interaction), then class share. The cheapest cooperative structure; the most widely useful.
• Jigsaw. Each group member becomes an "expert" on a distinct sub-topic by studying it first, then teaches it to the home group. No group member can succeed without the others.
• Numbered Heads Together. Students discuss in groups; one student per group is then randomly called to answer for the group. Individual accountability is random.
• Reciprocal roles in reading. Summariser, clarifier, questioner, predictor. Each student must fulfil their role in order for the group to function.
• STAD (Student Teams-Achievement Divisions). Teach content to the whole class, have teams study together, test individually, group reward based on individual improvement. A longer-running structure suited to skill-and-drill subjects.

For working templates, see Cooperative Learning Lesson Plan.

Classroom examples across phases

Primary. Year 4 reading comprehension using reciprocal roles. Groups of four read a short text; one student summarises, one clarifies hard vocabulary, one questions, one predicts what happens next. After the text, each student writes an individual three-sentence response to a comprehension question.

Secondary. Year 10 physics jigsaw on energy transfer. Home groups of four split into expert groups, one per energy-transfer mode (mechanical, thermal, electrical, radiative). Experts return to their home group and teach. Each student completes an individual exit quiz covering all four modes.

Tertiary. First-year biochemistry problem set. Four students each take one of four assigned roles (leader, recorder, sceptic, presenter) on an enzyme-kinetics problem. Group produces a worked solution; each student submits an individual short-answer question on a different enzyme before leaving the seminar.

Where cooperative learning fails

The failure modes are consistent and avoidable.

• Social loafing. One student does the work, others sign the poster. Mitigation: individual accountability tied to the same material, random oral checks (Numbered Heads), or group reward based on individual improvement (STAD).
• Unequal participation. The confident student dominates. Mitigation: assigned roles, sentence stems, and talk-time norms taught and displayed.
• Divide and conquer. Students split the task into parallel individual sub-tasks, never actually thinking together. Mitigation: tasks that demand joint reasoning — genuine jigsaw, shared problem-solving, peer teaching — not tasks that decompose cleanly.
• Weak task design. Cooperative structures on worksheet tasks that are better done individually waste everyone's time. Cooperative learning is strongest where the task genuinely benefits from joint reasoning.
• Behaviour norms not established. In a classroom where listening, turn-taking, and respectful disagreement aren't already normal, cooperative learning amplifies rather than solves the problem.

Best fit and poor fit

Best fit: upper primary through tertiary, particularly in science practicals, mathematical reasoning, structured text discussion, languages, and design/technology. Anywhere students can usefully explain and be challenged.

Poor fit: simple worksheet completion, heavily speeded individual test preparation, or classrooms where behaviour and talk norms have not been established. Cooperative learning is also a poor fit as a first resort with very young pupils who have not yet been explicitly taught the social-cognitive routines.

Teacher requirements, assessment, and resources

Cooperative learning is relatively low cost in resources. The investment is task design, behaviour routines, and monitoring talk quality. The teacher is not idle during cooperative work; they are moving, listening to group talk, and making targeted interventions.

Assess both group products and individual understanding. Include peer or self-assessment of participation, but keep it brief — the research doesn't support overlong self-reporting as a mechanism in its own right.

How TAyumira supports cooperative learning

TAyumira supports cooperative learning as one of its ten research-backed teaching methods. When you pick it, the generator produces:

• A lesson with a genuine positive interdependence task (not just group seating)
• Assigned roles or a structured cooperative protocol (Jigsaw, Reciprocal Teaching, STAD, or Think-Pair-Share, matched to the content)
• Sentence stems, talk norms, and role cards
• An individual accountability check: exit ticket, short quiz, or random oral check
• A group processing reflection prompt

Start for free — the Free tier covers the full lesson generation workflow.

FAQ

What is the effect size of cooperative learning?

Kyndt and colleagues' classroom-only meta-analysis of 65 studies reports achievement effects of roughly ES ≈ 0.54. The EEF's modern estimate is roughly +5 months of additional progress. A 129-school paired-reading RCT produced effects in the region of 0.2 — smaller because it ran under real conditions rather than optimised research designs.

What is positive interdependence?

Positive interdependence is a task design in which the group succeeds only if every member contributes. Jigsaw is a clean example: each student holds a unique piece of the information the group needs. Without positive interdependence, cooperative work collapses into one student doing the thinking.

How is cooperative learning different from group work?

Group work is any arrangement of students in groups. Cooperative learning is group work designed around positive interdependence, individual accountability, face-to-face interaction, taught social skills, and group processing. Most classroom group work fails on interdependence and accountability, which is why its effects are so variable.

How do I prevent social loafing?

Three design choices reduce loafing sharply: individual accountability (every student's contribution is visible and assessed), random oral checks where any student might be asked to speak for the group, and group reward based on individual improvement rather than a single group product.

What age is cooperative learning appropriate for?

The Kyndt evidence supports use from upper primary through tertiary. For younger pupils, teach the social and talk routines explicitly before attempting multi-role structures like Jigsaw; Think-Pair-Share is a safer entry point for primary classes.

Related evidence reviews

• Explicit Instruction Evidence
• Formative Assessment Evidence
• Project-Based Learning Evidence
• Problem-Based Learning Evidence

Sources

• Johnson, D. W., Johnson, R. T., & Stanne, M. B. (2000). Cooperative Learning Methods: A Meta-Analysis.
• Kyndt, E., Raes, E., Lismont, B., Timmers, F., Cascallar, E., & Dochy, F. (2013). Cooperative Learning: A Meta-Analysis Summary.
• EEF. Teaching and Learning Toolkit — collaborative learning.

Try one cooperative structure this week

Pick one lesson where the task actually benefits from joint reasoning — not one you could run as a worksheet. Use Think-Pair-Share. Add an individual accountability check as the exit ticket. Listen to one group's talk for three minutes and note the quality of the explanations. If you want full cooperative lesson plans generated, create a free TAyumira account.