Hi,
I have been trying to get a good visualization of the cumulative causes of changes to a pension fund’s unfunded liability by cause:
-I have tried a waterfall graph – but it is very difficult to “read”.
-I tried a stacked column chart – but some changes are negative, so this doesn’t work well — the height is plain wrong.
-I adjusted the numbers so that the totals added up correctly, but then the negative contributions are invisible.
A full write-up on the topic here.
![Spotify](https://policyviz.com/wp-content/uploads/2023/09/main-og-image.png")
I don’t really have time to look deep enough into it to propose an alternative, but here are a few pointers from a quick look.
– Be mindful of your copy. Some of your labels are confusing. Try to imagine how the NY Times would write them, in clear copy. Lose the title capital letters for instance.
– Label your categories directly on the graph, not in a separate legend. A reader who works less to understand the meaning of each color will understand your graph better.
– Chose your colors intentionally. Now, they appear rather random. Two shades of the same color suggest that two categories are similar or linked. Neutral colors attract less attention (“Others”). Red and oranges draw attention to problem areas (Employer contributions? Changes in assumptions?).
– Also be intentional about the order of your categories. I’m not sure how it is chosen now, but you might consider grouping similar categories (in shades of the same color) and then putting then in order of size for instance (largest category at the bottom, perhaps on top of the initial unfunded liability).
– Remove the data labels on the chart. Your graph’s purpose appears to be communicating large trends. These numbers are a distraction. Also, they take up space that you could use to label your categories.
– Consider small multiples (several area charts) to show the evolution of each category. They could be much smaller, without much labelling and be fitted in one row at the bottom of your larger graph.
– Feel free to annotate, such as announcing when a certain decision or event changed the course of history.
I guess what you want is actually a new type of graph that would make your data clearer and I wish I had the time to dive into it enough to make suggestions, but in my experience, improving an existing graph can go a long way to make it clearer. Good luck!