Changing my mind on group work

Embed from Getty Images

Group work is a default strategy for many teachers. Many forms of differentiation, such as the model used by the L3 program in New South Wales, require group work. Other teachers use it as a way of adding variety to lessons. If pushed to think about the research on group work, they may suggest it is good practice and link this to Vygotsky and social constructivism.

I tend to avoid group work in my own teaching. The only time I feel the need to create groups is when I want science students to complete practical work and resources won’t allow for them to work individually. However, my avoidance has not been due to scepticism about the effectiveness of group work. Until now, I have been persuaded by the arguments of Robert Slavin.

Briefly, Slavin surveyed the evidence on collaborative learning and found that it can be effective if two crucial conditions are in place; group goals and individual accountability. In other words, the groups need to be working towards some clearly defined objective and everyone in each group needs to be held accountable. For instance, a teacher may inform the class that one member of each group will be feeding back at the end of the session but that the teacher will decide who that is at the time. This way, everyone in the group needs to be fully briefed.

These conditions are necessary to overcome the greatest practical obstacle to group work; social loafing. Social loafing is a well-documented tendency for some group members to reduce effort and let others do the work. Most incarnations of group work don’t apply Slavin’s conditions and so this has been my main criticism.

However, I recently decided to review some of the evidence that Slavin cites in his reviews. I took a meta-analysis as my starting point.

It really hit me when I read this meta-analysis that Slavin draws a contrast between cooperative learning where there are group rewards, individualistic learning where there are rewards for reaching a certain level of performance, and competitive learning where there are rewards for performing better than other students.

This seems odd. Rewards are a feature of all conditions and the research seems to be addressing the best way of apportioning them.

One of the studies included in Slavin’s meta-analysis makes clear what this looks like. The subjects are Grade 9 science students:

“In the cooperative condition students were instructed to work together as a group, making decisions by consensus, completing the assignments together, making sure that all group members contributed their suggestions and ideas, seeking help and assistance primarily from each other rather than from the teacher, and with the teacher praising and rewarding the group as a whole. Students took the tests individually with the scores of each group’s members being totaled and averaged; each group member received the group’s average grade. A criterion- referenced evaluation system was used.”

So the students are graded on the performance of their group. For the competitive and individualistic conditions:

“In the competitive condition students were instructed to outperform all the other students in their class. A large wall-chart prominently displayed each student’s number and his or her daily and cumulative progress on all tests and assignments. The chart provided frequent and instant feedback as to the current status of each student. When the students changed teachers, the chart moved with the competitive group. The teacher verbally encouraged individual effort, made sure that students sat apart from each other, and instructed students to work on their own, asking only the teacher for help and assistance. Students were evaluated strictly on a norm-referenced basis of how their performance compared with that of their peers. The teacher praised and rewarded students who “won.”

In the individualistic condition students were instructed to work on their own, avoiding interaction with other students, seeking help and assistance from only the teacher, working at a self regulated pace, and completing as much of the assignments as possible. A criterion- referenced evaluation system was used where each student’s performance was compared to a preset criterion of excellence. The teacher praised and rewarded each student on the basis of how his or her performance compared to the preset criteria.”

Two issues stand out from this. Firstly, I cannot imagine applying any of these conditions in my routine class teaching. I tend to praise – or rather positively reinforce – effort over performance and I wouldn’t display students’ performance on a wall chart. Secondly, where is the teaching? It seems that students are working through individual assignments. Yes, they can obtain help from the teacher but there appears to be little explicit instruction. Under such conditions, I am not surprised that cooperative learning is more effective.

However, it is not that effective. The reason Slavin emphasises certain conditions for group work is that not all studies show a positive effect. In his meta-analysis, only 63% gave a significantly positive result. I suspect that there are a number of studies of this type that showed no effect and remain unpublished, so this is likely to be an optimistic figure.

I cannot locate many of the studies Slavin analyses because they are stated as ‘in press’ or are otherwise difficult to find. However, the ones that I am able to find all seem to have similar issues with the control conditions that cooperative learning is compared with. For instance, the authors of one of the studies admit that it is confounded (more than one factor was changed at a time) but suggest that the teachers involved thought the confound wouldn’t make any difference. It also seems like the cooperative learning group engaged in more retrieval practice – through the use of a game – than the control.

So I am starting to change my mind. I am not sure that the evidence for the effectiveness of group work, even if implemented under Slavin’s conditions, is sound.