0 comment

Decrease in Internet Explorer usage leads to lowering of murder rate

… well, certainly not. But a brillant article (because it’s as short as possible, consisting simply of an illustration) of Chris Blattman shows, how data and information can be used in a most misleading way. Correlating two sets of data leading to a seemingly interdependence. His example is the decrease of the rate of murder compared to the decrease of Internet Explorer usage. Speak it out loud and you’ll think: thats the dumbest shit I’ve ever heard. But look at his correlation graph. Doesn’t it look very persuading?
That is what I mostly dislike in the current trend to info graphics: ever graphically capable designer, blogger, journalist or whatever is able to create an info graphic not of “reality” (because there is no such thing as absolte truth, but thats a different post about epistemology…)  but of his preconception of “how things are”.
I’m not criticizing the use of info graphics at all, but we should be aware that they not only contain pure data but also the authors opinion of what that means. In the case of Chris Blattmans correlation there are just two sigmoidal curves. Layering them this way expresses an explicit wish to find a common cause for both. But sigmoidal curves just happen to be found nearly everywhere where things in- or decrease.
Having said that, I only can agree that the constant usage of Microsoft software can lead to a very special mind setting :)

[Update on 6. Jan. 2014] Here is another brilliant example of correlation vs. causation: “On correlation, causation, and the “real” cause of autism“.

Leave a Reply

Required fields are marked *.