Information visualizations can improve collaborative problem solving, but this improvement may depend on whether visualizations promote communication. In an experiment on the effect of network visualizations, remote pairs worked synchronously to identify a serial killer. They discussed disparate evidence distributed across the pair using IM. Four conditions, respectively, offered (a) spreadsheet only (controls), (b) individual unshared visualizations, (c) view-only shared visualizations, and (d) a full-access shared visualization of all evidence. We examined collaborative performance, use of the visualization tool, and communication as a function of condition. All visualization conditions improved remote collaborators' performance over the control condition. Full access to a shared visualization best facilitated remote collaboration by encouraging tool use and fostering discussion between the partners. Shared visualization without full access impaired performance somewhat and mad...
Aruna D. Balakrishnan, Susan R. Fussell, Sara B. K