From CS 294-10 Visualization Sp10

Jump to: navigation, search


Image:Redesign2_6_-_final - 2.png

tools: MS Excel, MS Paint

I chose to emphasize two stories: 1) Haiti pledges of aid by country/organisation and 2) Amount pledged per person by country. I chose bar graphs to show data variation using position along a common scale and length, two simple "high accuracy" tasks (Cleveland & McGill). I used green to encode the funding/country data because of money's association with the color. But I wanted to use a different color for the smaller graph to separate both data, even though both involve money, so I chose a neutral blue. I borrowed Tufte's idea of increasing data-ink ratio. To emphasize data variation and remove distractions, I took out the y-axis line, erased unnecessary gridlines, which I made 50% transparent, removed unnecessary interval markings, and thinned the bars as far as Excel would let me. Furthermore, gridlines were extended only as far as they could serve as a useful reference marking.

My original redesign attempt portrayed the two stories using one double-column (funding + population) bar graph: original_redesign. I was unhappy with this visualization because it graphed population, and China's enormous population greatly affected the scale of the other data and was a visual distraction. When I decided to use two graphs, I realized the isolated total funding graph left a huge blank space on the right side of the graph. To maximize use of the real estate, I placed the smaller $/person graph in this space. The rectangle boundary of the smaller graph ensures that viewers know that the $/person graph is separated from the total funding graph.

The $/person by country graph was not sorted from highest ratio to lowest because I wanted the countries to be in the same order as they are in the total funding graph. Although it takes longer to order the countries by $/person, not ordering the graph allows quick country lookup between graphs. Although the ordering takes longer, using length and position to encode the data makes the ordering not difficult.

My redesign obscures the "% of total data" (although it is possible to calculate this from the graph). In my bar graph of total funding by country / organisation, I left out the world total bar because including this data would have skewed the graph, making the graph harder to read as well as visually distracting (the world total pledged was approximately 3 times as much as the amount pledged by the next highest category). In retrospect, including the total funding would have helped visualize the "% of total" data much easier. Instead, I just listed the worldwide funding. The redesign also obscures the population of each country listed, even if it could be calculated with the data in the graph, but I decided that I did not want to focus on showing population. I did include the world total data in my smaller graph because it did not vary the scaling of the graph in a distracting way.

Stephen Chu

[add comment]
Personal tools