The Southern Women Dataset

This page demonstrates JUNG's ability to create bipartite graphs, to fold them into single-mode graphs, and to interactively visualize user changes to the data. It uses Davis, Gardner, and Garder's dataset from their 1941 Deep South (as reprinted in Freeman's 2002 paper, figure 1).

This applet presents three panels. On the left side, the full bipartite graph is laid out, matching women to the dates of parties. Vertices are rendered with blue squares for women's names, red squares for the dates. The graph is laid out with the Kamada-Kawai algorithm.

The middle panel shows the woman-to-woman graph. Two women are linked if they co-attended at least one party. The nodes are laid out with a force-directed spring embedder. The thickness of the line between a pair of women is proportionate to the number of parties they co-attended.

The right-most panel shows the list of dates of parties.

Removing Parties

It is the different parties that makes this dataset interesting. Were some of the parties removed, the graph of connections looks very different. It is the parties on 23-Feb, 15-Mar, 8-Apr, 19-May, and 16-Sep that bring all the women together. Clicking on a party in the left graph causes the middle graph--the folded version--to be recalculated without that party.