|Also available as an Acrobat File|
Visualisation in the Social Sciences Workshop
Towards a Research Agenda for Social Science VisualisationIn an attempt to derive a research agenda for future visualisation work in the social sciences, workshop participants conducted a pyramid exercise. In this each first wrote down four items they saw to be the most important for future work. These were then merged with the ideas of others in successively larger groups of two, four, and then eight. Eventually, after much discussion, twelve issues emerged from the group (see Table 1). Tables 2, 3 and 4 summarize the identified issues at each stage in the process with each labeled A, B or C according to the 'family tree' from which it emerged. Note that not all participants completed arrived at four issues and that with twenty-two participants not all groups were exactly of the labeled size. Inevitably, numerous issues emerged from the exercise, but it is possible to recognize three general areas of research need:
1. Aspects of the use of existing visualisation methodsIf visualisation is to be as successful in the social as it has been in the natural and physical sciences, there is a need for work on the experiences to date using visualisation in the social sciences. A major task is to widen the audience to include a much wider range of disciplines than was present at the symposium. Such work needs to examine the effectiveness of specific visual techniques as well as that of the general strategy. Research into the effectiveness and usability of visualisation systems is an area in which the social sciences can contribute to the physical and natural, a good example being work on the perception of colour and its use in displays.
2. Methodological and technical developments associated with the specific needs of the social sciencesOutside of GIS and mapping, existing visualisation systems supported by the higher education computing community (such as AVS, and IBM Explorer), as well as the graphics available in popular packages such as Excel, MINITAB and SPSS, do not readily enable visualisations of several types of social science information. Examples include non-hierarchical text such as CD-ROM and interactive maps, highly multivariate data (hypervariate is the term that was used), the outputs from statistical and mathematical models, and temporal and spatial dynamics in populations of individuals or aggregates of individuals. Since many problems in the social sciences, including many of the so-called 'qualitative' type, can be addressed by data organized as a network or interaction matrix, there is a clear need for technology to be developed to enable the visualisation of large and complex networks. In a computing context, such visualisations may well be the only way that we can use to address the complexity of the Internet and its use. A practical but important issue is the evident need to provide existing visualisation systems with appropriate interfaces to enable them to access data coded into standard relational or GIS databases.
3. Issues related to the environment of visualisationFinally, a cluster of 'human' issues were identified, particularly a need for work on how a visualisation strategy is incorporated into research, to assemble a library of suitable visualisation algorithms and case studies of their use, and to enable visualisation to be a collaborative activity.
Ways by which these agendas might be addressed formed part of a subsequent Workshop discussion.David Unwin
Graphics Multimedia Virtual Environments Visualisation Contents