- WA Sep'13
- WA Apr'14
How to interpret the diagram
Nodes on the left represent primary votes. The 15 groups that either won seats or were eliminated latest are tracked individually, while the remaining parties are aggregated into Other groups node. Intermediate nodes are positioned in the order in which they are excluded from the count (or fill a quota with insufficient excess to continue). Up to this point they collect primary votes and any lower preferences that have thus far been distributed. After this point, their votes will be distributed as preferences to others. Nodes on the right are individual elected senators, plus a Balance node which simply acts as a placeholder for votes that were still in play when the last seat is filled. First preference votes that contribute to a filled quota flow directly to the elected candidate. Intermediate nodes simply collect and distribute preferences in these cases.
Votes are counted, and preferences distributed, to individual candidates who are (usually) members of a group. I have aggregated votes and preference transfers to group level.
Some preference transfers may result in a flow from members of the final 15 groups ‘backwards’ to members in Other groups. These have been ignored. They won’t amount to much in the scheme of things.
Drag nodes to move them up or down. The initial placement of nodes results from an algorithm that tries to find the flow with the least entanglement. Rearranging the nodes may help to find a more visually pleasing configuration.
Data for these diagrams was obtained from the AEC
A detailed breakdown of the preference flows can be found at the ABC election results site
An alternative visualisation of these results, with an animated step through the process, can be found here
Thanks to Mike Bostock’s example of the Sankey plugin for D3.js.