outcomes, but the students may have benefited in terms of collaborative activity.
Developments over time may be an important factor in the efficacy of collaborative interactions. This is an aspect of computer-supported collaborative learning which has not previously been investigated. There may be an optimal way of working together and this may develop over time - or conversely, students may begin by collaborating effectively but after some period of collaboration, this way of working may break down and the collaboration become inefficient. It is therefore important to look at the nature of the interactions over time.
Time-based analyses were used in both studies. In the Secondary school study, an advanced computer- based tool was used to investigate a selection of the interactions. This facilitated graphical representations of the students interactions over time. Consequently, patterns of talk and behaviour were discerned and these were interpreted relative to the products and outcomes of the collaborations.
In the Primary school study, Timelines was not used because it is not appropriate for analyses which involve relatively large units of time. Instead, a hand- made representation of the interaction was created and this provided a description of the interaction as well evidence of patterns within the interaction. For example, from the representation it is apparent that the collaboration broke down over time and the children stopped working with one another and split the task. It shows that the majority of the collaboration occurred while the children were recording the sound. It is also obvious exactly who was working with who, and the teacher interventions can be seen. It also shows the distribution of the four different types of activities associated with this type of interaction.
The time-based analyses clearly show the benefits of this type of analysis. These representations provide a way of describing the interactions and enable us to see patterns in the interactions. This type of analysis facilitates relating products, interactions and outcomes to one another.
This paper has provided an overview of two empirical studies on computer-supported collaborative learning. These have been discussed from various perspectives. In particular, they point to the necessity to take a wider view of assessing effective learning situations, which incorporates affective as well as cognitive factors and distinguishes between products, outcomes and interactions. The studies illustrate the value of time-based analyses, particularly when investigating collaborations which occur over more than one session.