What works well in this visualisation?
- Michigan clearly stands out as the state with the highest auto insurance rates due to it’s exceptionally long dotted line.
- The states are ordered from highest to lowest, starting with Michigan and rotating clockwise around the circle.
- The average cost of full and minimum coverage is displayed for each state.
- Annotation lines have been used to ensure the labels appear neatly for each state, without any overlap.
- The data source is quoted at the bottom of the chart.
What could be improved?
- The radial chart type makes it very difficult to compare the individual states.
- State names are shown with their abbreviations only, making it more difficult to identify individual states.
- The stacked dots make it difficult to compare the differences between the cost of each insurance level, by state.
- The length of the dotted line represents the total cost of insurance, aggregated at a state level. It does not make sense to aggregate these numbers. Customers would typically only select one insurance type, not both.
- The image in the middle is distracting.
When I saw the original viz and dataset, I immediately knew I wanted to use a dumbell / barbell plot. This is one of my favourite chart types for displaying comparisons between two values. A dumbell plot works particularly well in this case since it emphasises the difference between the cost of insurance types at a state level, while comparisons between states can easily be drawn.
After building the initial chart, I realised it would be beneficial if the list of states could be refined somehow to show comparisons between states in the same region. To achieve this, I sourced a list of states by region and brought this into Tableau. I then build a simple map to display the four regions with the intention of using this as a filter.
To keep the chart clean, I labelled each barbell with the state name and the cost of full coverage only. Reason being, I thought readers would get a sense for the cost of the minimum level of insurance from knowing the full coverage cost and from seeing the length of the line between the two points. The colour coding for the dots is included in the introductory text, meaning the need for a separate colour legend is redundant.
I actually used a gradient colour background in this viz. It’s a very subtle effect but I like it. I used this site to generate the gradient and exported it as a PNG file.
As always, I like to research the topic behind my viz in an effort to understand the data a little better. I was intrigued why the insurance rates in Michigan were so high and why the basic level of insurance in South Dakota was so cheap. Once I had found answers to these questions, I included them in annotations on the viz itself. I used Tableau’s in-built annotation functionality to design these. Since the annotations are linked to the data points, they remain in place when a filter is applied.
To complete the design, I included a thin white frame around the viz. This was a technique I have seen used by Judit Bekker and I really like the way it completes the look. While Judit tends to includes breaks in her frames, I decided to keep mine complete.
Here is my completed viz. Click on the image to view the interactive version on Tableau Public.
Thanks for reading!