From the Maddowblog:
‘This has gotta be some kind of record!’
By Will Femia
Wed Jul 11, 2012 9:13 PM EDT
There has been a lot of talk about weather records lately. If you’ve been doing any fact-checking and finding yourself lost in a system of inscrutable pull-down menus and “animation” that doesn’t load, here are some shortcuts.
via ‘This has gotta be some kind of record!’ – The Maddow Blog.
A glimpse at the data, uneditorialized, shows a map like this:
At the National Climatic Data Center, you can look up all record highs, lows, precip and so on over the past 160 years (their year widget goes back to 1850). Draw whatever conclusions you want.
Recently, MoveOn.org published an infographic purportedly showing the comparison of various annual salaries of people ranging from welfare recipients to Fortune 500 CEOs:
This is an extremely inaccurate depiction of the differences between these bits of information.
Here is a spreadsheet showing the various numbers indicated in the graph above, and some ratios showing numerical comparisons:
If we take the same representation, i.e. a set of disjoint circles, to show the comparison, using the data above, this is what we end up with:
Mine isn’t nearly as pretty, but you can see where the infographic designer erred in making the comparison. The incredibly huge CEO compensation circle is now normalized to the rest of the circles, assuming the area of the circle represents the compensation amounts. (In the full picture, the diameter of each circle is represented by the normalized compensation ratio.) But this is a truly horrible way to represent data for comparison if one is at all interested in showing things in a way that people can actually make reasonable judgements.
Traditionally, comparisons of single group data are shown via a pie chart:
This might be more useful to see the relationship between the compensation values. However, pie charts are also notoriously hard to accurately reflect data comparisons. People have a difficult time judging the ratios between slices by looking at an area. In this particular set of data, if all you really wanted to show was the disparity between CEO compensations and the other groups, it may serve. However, an even better way to show it to achieve proper visual perception would be as a bar chart:
The data represented above in no way diminish the intended message. CEO compensations are way out of the norm. I hate to see the message be distorted by an egregious use of the graphical comparison of statistics. Such only serves to let those who aren’t inclined to agree to be able to push it off as dire histrionics and point to it as an example of how misleading we are.