Each Tuesday, Eurry Kim, a student in our class, will pick one example of data visualization to share with us. Eurry wrote:
Hi Rachel,
For this week’s viz, I decided on the following New York Times graphic:
Source: Jason Deparle and Matthew Ericson/The New York Times 05/09/09
http://www.nytimes.com/interactive/2009/05/09/us/0509-safety-net.html
New York Times again?? Yes, I have a good reason. Wait for it.
The graphic demonstrates a breakdown of government aid through the use of small multiples — a series of small charts sharing the same design and the same relative axis. These offer quick and easy comparisons between categories. Ed Tufte says that “an economy of perception results; once viewers decode and comprehend the design for one slice of data, they have familiar access to data in all the other slices. As our eye moves from one image to the next, this constancy of design allows viewers to focus on changes in information rather than changes in graphical composition (Envisioning Information, p.29).” Plus, notice the use of horizontal bars to display the percentages of government aid given to the various groups. Do you see the light gray bar under each percentage? This is a demonstration of Weber’s Law. The law basically says that a reference is needed to achieve perceptual judgement between small differences. Take a look at the first column and distinguish the differences between Vermont’s 49% and 46%. If that gray bar wasn’t there, it would be tough to tell the difference between the two values (the percentage labels notwithstanding — haha).
I am partial to uses of data science in journalism because of the respect paid towards the importance of visualization. Journalists have a duty to provide lucid and effective content to their readers — the same level of communication is demanded in their conveyance of data. Data scientists should feel the same responsibility. Our job is not over even after implementing the machine learning algorithm. Manually. Without a package. Communication of the findings is just as important as the statistical methods underlying the findings. Have you read Ed Tufte’s description of the botched presentation on O-rings before the fated Challenger explosion?
Finally, I just found out about a MOOC (massive open online course) offered through UT Austin and the Knight Center. It’s a course on data visualization starting on the 28th! And it will be using Adobe Illustrator (vector graphic design software) in its content. http://knightcenter.utexas.edu/00-11587-knight-center-launches-its-first-massive-online-course-introduction-infographics-data-visua
Thanks!
Eurry
Interesting graphic. I have three pieces of critique for it though.
1) I started looking at it and it took me a while to figure out what I was actually looking at. I thought at first it was the total percentage of people receiving benefits, but then when I saw MO food stamps at 98%, I was like no way and realized that it was actually welfare benefit UPTAKE. That’s an essential part of the takeaway and needs to more prominent, maybe in the graphic title rather than in the paragraph description that I only read when I was confused.
2) The color scheme of low and high was not necessarily logical, it kept me checking whether blue or black was high or low, this could have been avoided by using warm and cold colors for high and low, might not have gone with the NYT color scheme, but it would have been easier to take in. They could have also put this on a color scale- like a heat map.
3) The size of the maps make it hard to see the small Northeastern states (ie DE and RI).
One thing that as a statistically minded person I would also like to see is above the slider bars for ranges, a histogram so that we could tell how the rest of the states were distributed, but at the scale of the info graphic, that might have been difficult.
I agree with your first point. Admittedly, I chose it to demonstrate the aspect of small multiples and Weber’s Law. But as to your second critique, I’m not sure about the warm and cold color scheme. I think the NYT tried to be as apolitical with the colors as they could. Imagine if they had used a warm red hue to indicate a larger share of each sub-population… I think the blue was meant to communicate neutrality towards the issue at hand.
Also, did you notice the way in which the states were sorted? They’re sorted by an overall reliance on government aid — definitely an interesting way to view the national distribution.
Interesting, I did not actually notice the ordering. That does give an interesting perspective, but again, being the careless reader I am (which maybe when we are doing data visualization we should assume that all of our audience is) I did not read the text description…
Beautiful graphic!! Hope I could make one in future!!!