Although the results of the evaluation demonstrated bene�ts from using tablets, such as high engagement and similarities with the touch table, there is potential for further work to improve upon the design presented here. A sensible next step would be to make some small changes based on learnings from the evaluation. For instance, it seems necessary to create more aware- ness of what others in the group are typing, to encourage negotiation and discussion. This could potentially be achieved by requiring all members of the group to con�rm a text input once it is typed, and before inserting it, through a voting mechanism. Further re�nement could also be considered to reduce the risk of individuals claiming "ownership" over particular tablets. This could possibly be achieved by building in an induction tutorial that encourages all users to touch all tablets.
In order to con�rm the �ndings of this study, further studies using larger data sets would be needed. In particular, studies that involve a wider age range, socio-economic status and schooling style may provide important insights as to whether the�ndings presented here are generalisable to a wide population.
It would also be worthwhile conducting studies over longer time periods, enabling the use of more typical individual learning outcome measures based on formal testing. However, learning outcomes may be di�cult to demonstrate clearly. The link between collaboration and higher individual learning is clear, as well as the bene�t of improving this collaboration through more structured cooperative activities. It would therefore make sense that improving collabora- tive or cooperative learning through the use of technology would improve learning outcomes, however this may be a small e�ect and di�cult to clearly demonstrate.
Based on the promise of the prototype system tested here it seems worthwhile exploring the incorporation of more of the attributes of the original envisaged system (as described in Fig- ure 3.6). For example, using the tablets system to support not only collaborative work but also cooperative learning. Cooperative learning exercises like "Jigsaw", where students learn about a topic in an "expert" group then return to their original group to aggregate learning, could be e�ectively supported with this system of networked linked tablets. The use of spatial interactions between tablets, as proposed in the envisaged system, could also be explored to determine their e�ectiveness in enabling users to navigate larger volumes of information than was required in this study.
This envisaged system also explored the incorporation of tools for teachers. Teacher support would be necessary for a commercial application of this technology, therefore future work that develops, re�nes and tests the e�ectiveness of a teacher control panel would be bene�cial. Inter-group interaction could also be explored. For example, di�erent groups could collabora- tively learn about the di�erent parts of a topic - for instance, the water cycle - with the various parts�owing to the next group. Ideas from mobile learning could also be explored, such as creating location-aware learning situations based on GPS data.
Design Guidelines & Conclusion
8.1 Design Guidelines
These design guidelines provide a short list of things to be aware of when designing di�erent applications for similar tablets systems to maximise collaborative behaviour in students be- tween ages eight to twelve. They are based on qualitative observations taken from this study. • Ensure that group focus is drawn to events, especially if these events are not always vocalised. For example this could be done either through an addition being highly visible (in this case adding images), or a more explicit method such as displaying the addition on all screens.
• Multiple input sources should be carefully considered to ensure that they do not reduce group awareness, as per the above guideline.
• If tablets are moved or rotated the application should automatically adjust to maintain the concept that the tablets are windows onto the canvas. In this case, the system did not do this, and by rotating the tablets users could cause links to not align correctly, causing confusion.
• Shapes crossing the boundaries of the tablets should be straight (rather than curved) as these are easier to follow when crossing the bezels. Figure8.1shows this e�ect. Area1 shows links not aligning and area2shows the issue with curved elements crossing the edge.
• Consistent with our initial speci�cations, do not locate any permanent menus or buttons along edges that are next to other tablets.
F�����8.1: Shows tablets that have been moved and rotated causing links to not connect correctly.
• Do not digitally space the tablets to match bezel size. Objects should appear on the next tablet as soon as they are dragged o�one, rather than being "hidden" by the bezel. This reduces the chance for objects to be lost between tablets.
• Objects on screen must be normalised to real-world display size, rather than pixels, to ensure they stay at the same size when dragged between tablets.
• Allow�icking actions to pass objects between tablets.