A paper by Retnowati, Ayers and Sweller has recently been published that casts new light on the value of group work. If you have access then it’s really worth reading the whole thing. The introduction contains cogent explanations of a number of areas such as the worked example effect, the borrowing and reorganising principle and even the research evidence for collaborative learning where the authors note the strong positive evidence for effectiveness when certain conditions – such as group goals and individual accountability – are met.
I am going to focus here on the experiments described in the paper. All involved maths and participants were schoolchildren in Indonesia where, apparently, traditional teaching from the front of the classroom is effectively banned by the national curriculum.
In the first set of experiments, students were randomly allocated to one of two conditions. In the first, they studied worked examples, individually and in the second condition they worked in groups. The groups had been established previously and the students were used to working in them. The tasks involved example-problem pairs: on the left hand side of a piece of paper was a worked example and then on the right-hand side was a similar problem to solve. The students were given the final numerical answer to each problem to check their answers against.
The researchers thought that the level of complexity – or ‘element interactivity‘ – might affect the experiment so they created a further division within the experiment. Some students completed simple then more complex examples and others completed complex then simple ones. The complex examples were essentially the same as the simple ones but were expressed as word problems, which meant they required extra steps to first convert them into mathematical form.
After this, students completed a test to see how much they had learnt. This included similar problems to the ones in the task as well as ‘transfer’ problems that required students to apply the ideas in different situations.
The results showed basically no difference between collaborative learning and working individually for the simple-to-complex condition. Oddly, in the complex-to-simple condition, working individually was superior to working as a group. This seems at odds with the research evidence that suggests group work is an effective strategy.
In the second experiment, the researchers decided to only use the complex-to-simple condition due to the fact that this was the one that showed an effect in the first experiment. However, this time they included a comparison group that were not given worked examples and were instead given two problems to solve in place of each example-problem pair. Again, they were given the numerical answers.
This time, group work was more effective than individual work, but only when the students were solving problems without worked examples. This result also indicates that the groups were not dysfunctional – a possible alternative explanation for the results of the first experiment. However, the worked example group still outperformed the problem solving group overall.
It might be the case that worked examples provide all of the guidance necessary to successfully learn and so adding group work to this is unnecessary and potentially distracting. Collaborative learning might be more effective when guidance is lower because the higher cognitive demands of these low guidance tasks can be shared across group members.
This has interesting implications for the classroom – perhaps providing full guidance is optimal but, if we can’t provide this for some reason, group work may be beneficial.
Which, as an aside, gives me the opportunity to raise another point about group work that has been bothering me. We are often advised to use it to do things that a single student could not complete on their own. For instance, different group members might possess different information that they bring together for a task. The point I wish to make is that completing a task is not the same thing as learning from it (see the ‘multiply by three and add 29‘ experiment’). If your brain is totally occupied with doing something then there is little capacity left for learning – this is why a task can make you think hard yet lead to little learning.
Filling the pail 
Reblogged this on The Echo Chamber.
Thanks for drawing this to our attention Greg. I am pleased to see the subtlety in this account and in the design of experiments. I am increasingly unconvinced by absolutist pronouncements on the worth or worthlessness of a technique. If nothing else we have surely learnt that learning is complex and people are different (both teachers and students). I think you final comments about times when “thinking hard” can be inimical to learning are particularly interesting. I still suspect that “not thinking” is unlikely to lead to much learning except where items are to be recalled without thinking (e.g. The alphabet, times tables or riding a bicycle. I suspect the crucial addition to the experience of having to think hard to solve a problem is then being supported to reflect on and make explicit how you achieved the solution (and where else you could use the approach)
Recall does involve thinking, and it’s necessary in order to learn new things that we have a foundation to build on.
Hi Greg
Great article and summary of the report. It doesn’t surprise me that the complex to simple wasn’t as strong in group compared to an individual as the chance of distraction is much higher.
It is worth considering that not all objectives of the classroom are topic related though. Working as a group is a skill that needs to be developed for many jobs (at least as many that require high levels of.mathematics). How we ‘test’ this level of skill is the tricky part.