The dataset for week 1 was nice and simple, containing average salary by gender for various occupations in Australia.
The original article represented the data in two separate ranked tables (one for each gender), making comparison between genders for the same occupation difficult.
My first thought was that a slopegraph could show this difference effectively – I also wanted to learn how to create one in tableau, having just endured the frustrations of creating one in excel (and recolouring multiple series manually).
My second thought was a connected dot plot might also do the job – I was sure I had seen some recent examples displaying differences in pay by gender – and I was keen to learn how to do this technically.
Then I stopped myself – was this an Excel mindset? One of the most important advantages of tableau is the ability to explore data visually. When a chart takes seconds to create, the capacity to experiment grows exponentially. It was time to play with the data.
Exploring the data
The bar chart below shows the highest paid professions by gender. Even when the whole chart is ordered by female pay, many, if not all, of the male bars appear longer.
So what if we order by rank?
With a couple of drag and drops we get the following chart – and our the central message is clear – the top ranked occupations for males (now in orange) pay much more than the top ranked occupations for females (in blue).
By creating a combined field for gender and occupation we can display the same message in a different way – the chart is a sea of orange.
The beauty of Tableau is that with just one or two clicks you can visualise the data in different ways – here are two which have their merits, but didn’t make the final cut.
I was still keen to show how pay differs for the same occupation so it was back to my first though – the slopegraph.
Although this isn’t a standard tableau chart, a quick google returned several online tutorials, including this one from Andy Kriebel.
A few minutes later, I had this:
We can see that almost all the lines slope upwards (indicating higher salaries for males), with a few interesting exceptions (magistrate, futures trader), but there are too many categories, and occupations with a salary under 100k are obscured.
I could filter the results, de-emphasise the lower lines, and colour code the results according to whether male or female pay was higher, but as time was ticking, I returned to plan B.
Connected dot plot
Again this is not a standard tableau graph so I was relying on the tableau’s extraordinarily helpful online community to show me the way.
After some google image searching I found this helpful post from Ryan Sleeper and learnt that the chart has two aliases: the dumbbell chart and the DNA chart!
Within a few minutes I had this: the bare bones of my final visualisation:
From there, it was just a case of tweaking my chart– by changing the default order to variance, adjusting the size of the marks, moving the dots behind the connecting the lines and adding a (not particularly great, and quite possibly dated) title.
My priority in week 1 was to start relearning tableau, so learning how to create two new charts was a big win. The final visualisation displays my key finding fairly effectively: there are some large gaps in pay between genders for the same occupation, with the men almost always being paid more than the women.
The visualisation feels incomplete (don’t they all?), and I feel there is work to do in terms of editorial focus, moving away from the default display to something more aesthetically pleasing, and using multiple charts to show different aspects of the data.
But it is a start!