Ice Cream Crimes

16 Sep 2014

Ice Cream Crimes

Colonel Mustard in the Observatory with the Ice Cream Scoop

Many textbooks on statistics and criminology alike use the correlation of ice cream sales with the rate of violent crime to highlight that correlations does not imply causation. Yet, time and time again this mistake, along with other common pitfalls, leads to data being used to back up a false claim.

In the name of all that is true and verifiable, I’m starting a series of posts marked with the Ice Cream Crimes tag to find examples of data analysis techniques being used incorrectly. In terms of categories of dispute (as laid out in my last post), these will be mostly disputes of fact. That is, does the data suffieciently support the claim? Example infractions will include insufficient sample size, ignoring other variables/explanations, or even using poor visualization techniques to mislead.

So look for these posts in the future: I’m sure I’ll have a lot of material.

comments powered by Disqus