This post is not really about programming, but visually displaying quantitative information. Here’s a grand example of what not to do when you create a graph. The Wall Street Journal created a graph that shows the income distribution of Americans.
http://online.wsj.com/article/SB10001424052748704621304576267113524583554.html
The Journal image looks like a histogram with a Gaussian shape; it shows income levels on the x-axis, and amount of taxable income on the y-axis. At first glance, it looks like those in the middle class have the most taxable income. The bars in the middle of the x-axis are highest and the bars are small at either end (the very poor or very rich).
However, a second glance reveals that although the bars are physically the same width, they represent vastly different income groupings. The first few bars are:
1-5 Thousand
5-10 Thousand
10-15 Thousand
So far, so good: same width, same income span. But then, the bars stay the same width physically, but the income widths accelerate:
75-100 Thousand
100-200 Thousand
200-500 Thousand
The last three bars kick it up a notch:
2-5 Million
5-10 Million
10+ Million
Brendan Nyhan shows how to vary the income groupings to tell whatever story you want to tell!
http://www.brendan-nyhan.com/blog/2011/05/the-use-and-abuse-of-bar-graphs.html
And here's another writer's take on the graph:
http://motherjones.com/kevin-drum/2011/05/fun-charts-making-rich-look-poor
I’m fascinated (and appalled!) that this sort of “junk chart” still gets published and widely distributed!
-- Mike Carniello