Have you ever found a graph of some interesting information, but the graph was difficult to understand (or even misleading). I strive to fix those graphs - this time it's a graph of US immigration data...
It was an interesting topic, but I found the graph a bit difficult to understand, and even a bit misleading. Here are a few of the problems I see with the graph (in no particular order):
- It is difficult to read the text on the axes, since it is graph text on a black background.
- The colors in the bars do not match the colors in the legend (the bars seem darker). For example, I first thought the reddish color predominant in the bars before 1920 matched 'Oceania' in the legend (and I thought that very strange). I later found that matched Europe.
- The bar heights only go to about 10 million, but the vertical axis goes to 24 million (I assume this is to make room for the map in the background?).
- There is a world map in the background, but it doesn't add to the data analysis - it's just a decoration and a distraction.
- When I first looked at the graph, I thought it odd that immigration had dropped off after 2010, but upon closer examination I found that the 2010/11/12/13 bars represented single-year values, whereas all the other bars represented decades.
- And I'm not a big fan of creating the bars out of small 'building blocks' instead of using the traditional stacked bar.A different view of US immigration #dataviz Click To Tweet
I think this data is interesting and important, and it deserves a better graph - therefore I set about creating one. First I located the data (Table 2) on the US Homeland Security website. I wrote some SAS code to import the Excel Spreadsheet, transpose it, and create a more standard bar chart that is easy to read, and avoids things that could cause the user to misinterpret the data.
I'm not a big fan of showing the world map in the background, but I decided to add that in order to show how it could be done in a way that might actually help visualize the data. Notice that the regions in my map are color-coded to match the bars and legend. I created the map separately with Proc Gmap, and then annotated it into the Proc Gchart bar chart. Click the image below to see the full size graph, with html hover-text.
What other ways might you visualize this data? Feel free to share your ideas in the comments section!