Cooperative learning: what it is, how to set it up, and what the research says

What it is

Cooperative learning is small-group work designed so that students need each other. The group has a shared goal, and the group only reaches it when every member has learned the material, not when the strongest student has done the task. That design is what separates it from ordinary group work, where four students share a table and one does the worksheet.

The theory behind it, set out in David Johnson and Roger Johnson's review, is social interdependence: how students' goals are linked shapes how they treat each other and how much they learn. When goals are cooperative (I succeed when we all succeed), students help and explain. When goals are competitive (I succeed when you fail) or individual (my result has nothing to do with yours), they don't.

In practice, cooperative learning comes in three sizes:

• Quick structures inside any lesson: think-pair-share, turn and talk, numbered heads together. One to five minutes.
• Formal cooperative tasks: a group project, a jigsaw where each member becomes the expert on one part and teaches it, a team review before a test. One lesson to a few weeks.
• Base groups: the same small group meets over a term or a year to check in, help with work and keep each other on track.

The common thread is two design features: a group goal, and a way to make sure each member is responsible for learning (individual accountability), such as random calling, an individual quiz, or each member presenting one part.

Why it works

Explaining something to a classmate makes the explainer retrieve it, put it in order and notice the gaps. The classmate gets an explanation in words close to their own, from someone who just learned it, and can ask questions they'd never ask in front of the class.

The largest review of the goal-structure idea is Cary Roseth, David Johnson and Roger Johnson's 2008 meta-analysis: 148 studies of over 17,000 early adolescents in 11 countries. Classes where students worked toward shared goals had higher achievement and more positive peer relationships than classes where students competed or worked alone, and under cooperative goals achievement and good relationships went together.

The evidence on programs in real schools points the same way. Robert Slavin, Cynthia Lake and Cynthia Groff reviewed 100 studies of middle and high school math programs that had control groups and lasted at least twelve weeks. New textbooks and computer-assisted instruction had very small effects; two cooperative learning programs had positive effects. The authors concluded that programs that change daily teaching and how students work together hold more promise than new textbooks or technology alone.

Learning in groups also seems to help students use what they've learned on new problems. Hui-Hua Pai, David Sears and Yukiko Maeda's meta-analysis of small-group learning found a positive average effect on transfer, though results varied a lot between studies.

How to start, by grade

Examples

• Think-pair-share with a random reporterGrades 1–12

  Ask a question. Thirty seconds of silent thinking (or writing). Two minutes telling a partner. Then call a name at random and ask: "What did your partner say?" Every student thinks, every student talks, and everyone listens, because anyone might report their partner's answer.

• Numbered heads togetherGrades 2–8

  Students in groups of four number off one to four. You ask a question; heads go together until everyone in the group can answer it. Then you roll a die or spin for a number, and that member from each group answers on a whiteboard. Groups make sure the weakest member understands, because they don't know who'll be called.

• Jigsaw on a readingGrades 4–12

  A chapter on ecosystems has four sections. Each home-group member takes one section, meets with the members from other groups who have the same section, and becomes the expert. Experts return home and teach their section. A short individual quiz on all four sections follows, so each expert's teaching matters to everyone's score.

• Team review with improvement pointsGrades 3–8

  Before a unit test, teams of four review together with a set of practice questions, making sure every member can answer each one. After the test, teams earn recognition for how much each member improved on their own previous scores, not for the raw total, so the team's strongest student gains nothing by doing the work alone.

• Partner math with jobsKindergarten–2nd grade

  Partner A builds the number with counters while partner B counts and checks. Partner B writes the number. Then they switch jobs for the next card. You walk around listening for counting and agreement, and use the team points chart to notice good partnership.

Common mistakes

Group work without a group goal.

If the task can be done by one student, it will be. Design tasks where the group succeeds only when every member can explain or do the work, and check that by calling on members at random.

A shared group grade.

One grade for the group invites the strongest student to do everything and the others to coast, and it hides who learned what. Keep individual accountability (individual quizzes, random reporters) and use team recognition for improvement rather than a shared grade.

Assuming students know how to work together.

Listening, explaining, asking for help and disagreeing politely are skills. Teach them explicitly with models and sentence starters, and practice them like any other skill.

Groups that are too big.

In groups of five or six someone always drifts. Use pairs for quick structures and threes or fours for longer tasks.

Using it for everything.

Cooperative learning works well for practice, discussion, review and applying knowledge. It's a poor way to meet brand-new content that nobody in the group understands yet; teach that explicitly first, then use groups to practice.

Letting groups run themselves.

Circulate, listen and step in on how the group is working, not to give answers. Close with a quick reflection on teamwork so the next session goes better.

Time and cost

Time: quick structures take one to five minutes inside a lesson. Formal cooperative tasks need more class time than a worksheet, and teaching the social skills takes time at the start of the year that pays back later.

Money: nothing beyond what you have. Our free group, timer and picker tools below cover the logistics.

Training: moderate. The structures are easy to learn; designing tasks with real interdependence and individual accountability takes practice.

Debates and open questions

Does structure matter? Pai and colleagues noted that earlier reviews of cooperation and achievement favored structured groups over unstructured ones, while their own review of transfer to new problems found no extra benefit from structure. One reading: structure matters most for making sure everyone learns the content; for applying ideas to new problems, the discussion itself may do much of the work.

What about strong students? Parents of high achievers sometimes worry their children spend group time teaching others. The research we cite reports average effects, and we haven't yet added a study on high achievers specifically. Rotating roles and using ability-mixed and like-ability groups at different times are common answers.

Who did the research? Two of the reviews cited here (Roseth and colleagues, and the Johnsons) test or describe social interdependence theory, which the Johnsons have written about at length. That's normal in research, but it's fair to weigh it, and to give extra weight to program evaluations with control groups, such as those in the Slavin, Lake and Groff synthesis.

Tools that do it

Free, on this site, nothing to sign up for. Class lists stay in your browser.

• Random group maker: make pairs, threes or fours in a second, or keep the same groups for weeks.
• Random name picker: pick a random reporter after group talk, so every member must be ready to explain.
• Dice roller: roll for the member number in numbered heads together.
• Countdown timer: show groups exactly how long they have for each step.
• Discussion sentence starters poster: the words students need to ask, explain, agree and disagree in groups.
• Team points chart: recognize teams for improvement and teamwork rather than raw scores.
• Seating chart, pods of four: plan a room for groups of four.

Questions teachers ask

What's the difference between cooperative learning and group work?

Design. Group work puts students together on a task; cooperative learning sets up the task so the group succeeds only when every member learns, with a shared goal and individual accountability. Without those two features, one student often does the work while the others watch.

How big should groups be?

Pairs for quick structures like think-pair-share; three or four for longer tasks. Larger groups make it easier for someone to drift and harder for everyone to talk.

Should I grade group work as a group?

We'd avoid a single shared grade for learning. It invites one student to carry the group and hides who learned what. Grade individual work individually, and use group recognition for improvement or teamwork.

Does cooperative learning work for young children?

Partner structures with clear jobs work well from kindergarten. Start with pairs, model every step, keep tasks short, and build up to groups of three or four later in the year or in later grades.

Is think-pair-share cooperative learning?

Yes, one of the simplest forms. Everyone thinks, everyone talks to a partner, and a random reporter keeps everyone accountable. It fits into any lesson in two or three minutes.

What does the research say about peer relationships?

Roseth and colleagues' meta-analysis of studies with early adolescents found that cooperative goals were linked with more positive peer relationships as well as higher achievement, and that achievement and good relationships went together.