Misc. Information
Part 2: Visualization Redesign
One of the major problems with all of these visualizations is the color scheme, which consists of red and shades of grey. My first step was thus to choose a new color scheme. For all of these redesigns, I chose the yellow-to-green-to-blue color scheme from Color Brewer, which 1) has easily distinguishable colors, 2) is easy on the eyes, 3) looks good in print according to Color Brewer, and 4) is color blind friendly according to Color Brewer.Redesign of "Migration" visualizations
Redesign of "Going South, 2007" visualization
Here is my proposed redesign, created in Adobe Illustrator:
The color of the circles and the actual numbers in the circle encode the migration numbers. The location of the circles on the grid encode the "From" and "To" dimensions.
The main problem with this original visualization was that it tried to present two dimensional data in one dimension, and so it was difficult to see the overall trends. I first considered placing arrows on a U.S. map which would encode the migration data, but then I realized that this would just result in a very cluttered map. So I abstracted out the geographic data by making this simple grid. The left-side labels depict the "From" regions and the top-side labels depict the "To" regions. It is now very natural to locate data points and answer questions. For example, one might ask "of all the regions, which regions has the highest migration from the Northeast?" To answer this question in this new visualization, you need only scan across the row corresponding to "From Northeast" to learn that the South has the highest migration from the Northeast, with West in second and Midwest in third. Having the color encoding of the numerical data makes the high-level trends easily visible, while having the exact data helps confirm these trends. At first, the white text did not show up well against the yellow and light-green backgrounds, so I overcame this problem by putting thin black borders around the characters. Also, I added red X's for "From Northeast To Northeast" etc. since these points have no migration data. Finally, I felt that the data was not giving the whole picture since certain data points were left out (e.g. how many people moved from South to Northeast), so I created some dummy data points to imply a more complete picture.
Redesign of "Population change for select states" visualization
Here is my proposed redesign, created in Microsoft Excel:
This bar graph encodes the data that was previously in table format only. The lengths of the bars encode population increases. It is now very easy to see not only the overall trend, but also how changes in one of these states compares to another. For example, one can quickly see that the population increase of Texas is several times greater than the population increases for Washington, Colorado, Nevada, Tennessee, and New York. Such trends take longer to infer from looking at numbers alone. I also included a bar for "average" (whose figure I made up) for comparison, which gives an important frame of reference to the data and makes the statistics more meaningful.
Redesign of "Population distribution" visualization
Here is my proposed redesign, created in Adobe Illustrator:
The colors now encode the populations of states, not of regions. I felt that this more accurately depicts the population distribution than in the original visualization. For example, the "West" regions may hold a large share of the population, but most of that population could be concentrated in California. Ideally, I would have liked to create a heat map, but I recognize that geographically continuous migration data is probably not obtainable. Also, the original color scheme was the biggest flaw of the original visualization. This new color scheme is easy on the eyes and the colors are easily distinguishable, in particular on printed material (according to Color Brewer).
Redesign of "'Magnet' states" and "'Sticky' states" visualizations
Here is my proposed redesign, created in Microsoft Excel:
There were only two major flaws of the original visualization, which were the choice of color (red) and the lack of an "Average" data point for frame of reference. I alleviated these problems by choosing a cool color consistent with my overall choice of color scheme from Color Brewer and by adding (made up) average statistics.
Redesign of "Urban Sprawl" visualizations
Here is my proposed redesign, created in Adobe Illustrator and Microsoft Excel:
The first thing I considered when making this redesign was the city diagram. In the end, I decided that this diagram did not add to the explanation of the given terms, so I chose instead to just display the definitions of "urban" etc. in a list. Next, I liked the "small multiples" pie charts, but wanted to change the color scheme. Since I needed a nominal color scheme, I went back to Color Brewer and picked a new color scheme that would fit this task but is still similar to my first scheme and also still readable on print. Lastly, I included a time-series scatterplot to show the data trends over time, using the same color scheme. From this graph, we can now clearly see how the suburban share of population increases while the urban share of population decreases.