From CS 294-10 Visualization Sp10
Design Description – Haiti Aid Data
I focused on the “TOTALS” tab of the spreadsheet. “ISO COUNTRIES” and “UN POPS 2009” are reference tables that don't have any Haiti specific data. “FUNDED AND UNFUNDED” was too brief, and “FULL DETAILS” was, well, too detailed. I thought that if someone were trying to get there head around the main story in this data, they would start with “TOTALS” first, then drill down.
The story told in the “TOTALS” tab is about three quantitative values: population, total funding, and dollars per person. The tab also provides denominators for these three values to be scaled by: world population, world funding, and a maximum dollars per person value (Canada). Each record therefore contains 6 pieces of quantitative data and at least one nominal identifier (two for countries). In fact there are only really two pieces of quantitative information per record, “Funding, committed and uncommitted, $” and “Population”. The other two fields are derived from these.
Working with this data, the story is mainly a comparison of countries with regards to these to properties and the additional derived data. I saw three ways one might choose to tell this story focusing on population, total funding, or the ration from dollars to person. By ordering the data in one of these three manners, the viewer can choose the frame for their analysis and explore other relationships.
Story Not Told
One of the stories not told by this data is the how the ~40% of the world population that has not provided any funding is divided. This could be done by incorporating the data from the “UN POPS 2009” tab.
In general I tried to follow principles from the reading with a focus on data density, and a high data ink ratio. I also specifically liked the “Tokaido Line at Yokohoma Station” graphic from Tufte's Envisioning Information (p47), which showed how the same ink can serve more than one informational purpose. That example focuses on two sets of data per row. I tried to use a similar style to focus on three.
My graphic relies heavily on the bar chart design, which I believe lends itself to quick comparison of countries along different variables. Each row actually has three bars in it, each aligned to a separate axis. The key is read across the top for clarification. I felt the horizontal order of the data read naturally. China for example: 20% of the population, or 1,345,751,000 people gave $0.009 per person for a total of $13,005,286% or roughly 1% of total world funding.
The main visual encoding is length, which the eye has a relatively strong capability to process automatically. Not wanting to sacrifice accuracy, I also included actual data points next to the lines for those who wanted to drill down into more detailed information. The only piece of data not explicitly stated is “$ per person” / max( “$ per person”), although this value is coded in the orange bar chart length. Color was used sparingly, only to separate boundaries between the different bars.
This design allows the data to be sorted in three ways from top to bottom, emphasizing different orderings of the country/organizations, but allowing for all the data to be considered regardless of the sorting.
Each sorted view lends itself to different comparisons, and facilitates different explorations of the data. Sorting by any one of the three dimensions will show where exceptions are seen in terms of the ordering of both of the other two.
Finally, I tried to make the graphics small enough so that a small multiple effect could be used. The images would be indexed by sorting criteria and provide a quick way to see how the different criterion ordered the data. The design and horizontal position of the data remain constant so that the user only has to learn one design and can apply the pattern to the others quickly.
Further work could involve better management of text color, logarithmic scales, and incorporation of other data from the spreadsheet, as well as user interaction components.
Small Multiple Version