ABC News Australia recently published a scrolling visual data analysis of Australian voting history that made use of a ternary-axis chart. Here is one slice of the story.
The story nicely explains how the three axes work together since it’s not a common chart type and walks the reader through some key moments in the Australian voting results. I think the main take-away is that if you group the political parties into three main groups: Australian Labor Party (ALP), Liberal/National Coalition (L/NP), and all others combined, you can witness the rise of other party candidates at the expense of the two main parties. In the chart, it’s a literal rise, as dots higher in the graph indicate a higher number of votes for the other parties. The dots start at the bottom and rise higher over time.
However
While organizing the “other” parties together to create three main groups is the key step that allows for the article’s insights to happen, I wonder if the ternary data requires a ternary chart. In this case, using three axes for the three groups means there is no axis left for time. Instead, you have to scroll through the article to get a sense of time patterns, and there’s never any all-in-one time view.
In case it’s not already obvious, I know nothing about Australian elections. So I’ll just share a few visual alternatives with minimal commentary. I’m mostly using the same level of detail but showing each individual result (three dots per district per year instead of just the winner).
First, a paneled view with a separate chart for each party/group. Triangles indicate won seats and dashes are losses. The rise of the Other group over time is apparent. You can even see a few yearly shifts, including how in 1990 when Others had their first surge, it was mostly at the expense of ALP. All the years are shown, and you can even see the occasional irregular spacing of the years. However, you can’t really see each result because of the overstriking.
The next view combines everything in one chart (except dropping a few early years) and adds dodging jitter to avoid overlapping points and show the distribution. The rise of the Others group is not only apparent, but it’s easier to see it relative to the other parties. What was once a separate blob of dots became fully merged with the main blob. though still mostly in the bottom half.
A bonus insight made by a Bluesky user from this chart is the dark gray dot at about 12% in 1996. That dot is present in the original ternary chart and my paneled chart but is obscured. A win with 12% result? I had to look it up. It seems that dot is for Pauline Hanson, who was “disendorsed” by ALP after ballots were printed and took office as an Independent.
Finally, a side-by-side panel view is interesting, though I could only include a sample of the years because of space. The time trends are more pronounced.
