I’m working on a way to display trends in cancer survival over time broken out by stage and race. The purpose is to compare trends in survival between races so that if, for instance, whites have seen a large improvement but Asian/Pacific Islanders have not, that will come across in the graph. So far, this is the best, but it’s pretty hard to read and interpret. I was hoping to get some advice on how to design the graphs to make them clearer and easier to read. I’m struggling with the layout more than the actual implementation, so sketches or ideas on what to do is more important than any kind of code or how to do it (although the latter would be appreciated!).
The data are slightly different than what’s used in the graphs since it uses a different staging method (Localized, Regional and Distant instead of I,II,III and IV). In addition, it contains values for all sites combined while the graph is only for colorectal cancer. The results won’t be exactly the same, but should look roughly similar and since I’m more concerned with how to display the data than the actual results I hope this won’t be too big a deal. If it is, let me know and I’ll figure something out.
Thanks!
I think what you have is good: straightforward and something anyone can read. The message I get is that there isn’t much difference in race or year but there is a big difference in stage. If that’s not the intended message, then some adjustments may be needed.
Other comments:
It would be helpful to know the population sizes to get an idea about the accuracy of the proportions, especially if each dot represents different sample sizes. A text annotation may suffice or you might represent the sample sizes graphically with confidence intervals or bands.
If you want to emphasize the long term trend rather than year-to-year variation then consider smoothers instead of connected lines, as shown in my sample.
It’s good that all the panels share the same scale, but it’s not strictly necessary to start the Y axis at 0 for line graphs unless there is some convention for your field or institution.
testing this post, please ignore
I think the lines are too scrunched together, making it hard for people to differentiate between the races. I’d look for a way to separate the information to make them easier to understand such as this https://public.tableau.com/profile/antonio.peredo#!/vizhome/CancerSurvivalbyRace2/Dashboard2 Another idea would be to show people the difference in survival rates compared to the first year. It would provide a new way to look at the data. https://public.tableau.com/profile/antonio.peredo#!/vizhome/CancerSurvivalbyRace/Dashboard4
Sorry I meant to upload another version of the picture. Click on it and it will expand