Figure 1: In the dataset visualizer interface (left, panels DV1-3), voting records from a constitutional convention dataset can be explored, and voting patterns from people can be visualized as projections. Data can be filtered and sampled, and information about the quality and composition of the projection itself can be visualized. Records from individual people can be investigated using the person visualizer interface (right, panels PV1-5), which contains committee involvement, keywords, relationships, event participation, and voting breakdown. The blue highlights show an example of workflow: parts of the timeline are chosen to be projected (A), a point of interest is identified in the projection as two people became closer (B) and the reason for that behavior -a large section of votes in agreement -is seen on the vote breakdown (C).