For any subsequent use of collaborative design in a middle school context, we would make several recommendations over our original implementation. Some of these are adapted from other studies with different goals and age-groups; some are derived from the students' suggestions; and some arise out of our own observations:
Druin et al [Druin et al.1997] suggest that students and adults should work together in teams and that there should be more than one of each per team. The objectives are on the one hand to prevent the students from feeling overshadowed by adults and, on the other, to diminish the dominance of the ``school-like'' dynamic. A one-to-one ratio would probably be ideal but at least one adult to two students should be sufficient. Due to the small number of adults available during this project, it was difficult to achieve this goal. Druin also suggests that adult-to-adult interaction is important and that adults should be informal and playful in their interactions both with other adults and with students.
Klawe and Phillips [Klawe and Phillips1995] emphasize the importance of carefully explaining the role of a researcher and the kinds of activities that a researcher might undertake. This includes encouraging students to formulate research questions and to pay attention to rigor and detail while conducting research. Other strategies might include inviting researchers to speak about their work and having students conduct a preliminary research project of their own which might or might not be related to the final project. As well, students' own opinions on what they need to feel included might be solicited.
The students themselves made the suggestions that their roles as researchers would have felt more genuine had they had the opportunity to visit the PDG lab and to interact more often with members of the PDG team. Evidence to support the potential positive impact of this strategy emerged when the students were shown a screenshot of the applet building environment used by the researchers, Java Studio. They were immmediately captivated by the original and visual interface.
They began asking questions about the different components of the interface and how the teacher worked with it. They were able to compare their own activities with simCHET with the environment simCHET was emulating. Had the students been able to see the actual physical environment in which the PDG members worked, had they been able to sit at the computers and explore the interface, the expectation is that they would have developed a stronger personal connection to the project.
Clearly emphasis needs to be placed on creating an authentic collaborative experience. In our case, having the teacher lead the project at the school detracted from their sense of participation. Although the presence of a familiar person helps create a comfortable environment for the students, we propose that the teacher take on a more secondary role in any research collaboration.
As there are many different personalities within a student body, it is difficult to accomodate everyone's participation. For example, while oral debriefing was found to be most effective for eliciting useful feedback, there was a concern that weaker, less verbose students' potential contributions would be left out. A strategy which would address this concern would be to hold smaller group de-briefings in addition to full class ones. Another hybrid format might be to have smaller discussion groups in which one student was responsible for recording the comments and suggestions of the group members.
We were able to observe that the students were more motivated to collaborate upon seeing their suggestions implemented. A continuous flow of information between the students and the developers is conducive to creating an authentic collaborative environment. This might include trying several suggestions regardless of any judgements held by the developers. This would demonstrate the value of their input early and also show them how even unsuccessful suggestions contribute to the design process. In any case, it is not unusual for students to propose ideas which never occur to developers.
Druin et al [Druin et al.1997] argued that children ages 7 to 10 years old make the most effective design partners. This is mainly because they are mature enough to discuss what they are thinking yet young enough not to be too heavily burdened with preconceived notions of the way things ``are supposed to be''. In our case, the students were 12 to 14 years of age; we found them to be easy to work with, partly due to the existing sense of community among them.
In summary, our recommendations for future work with middle school students in guided collaboration projects are: