It is official. “Power Law” and “Pareto Distribution” have become two of the most misused, abused, and pundit-mangled statistical terms around. Not ringing any bells? Power law distributions are often referred to as “Long Tails,” of Chris Anderson/Wired magazine article fame. Apparently, many data distributions referred to as power law curves may be something else entirely. Fortunately, vigilant statisticians have sprung into action, issuing a paper on how to determine whether a particular data set really contains a power-law distribution. (Warning: not for the mathematically faint of heart!)
- The article: Power-law distributions in empirical data.
- The blog entry about the article: So You Think You Have a Power Law—Well Isn’t That Special?
- Another blog entry about the article: Power laws and all that jazz
A flavour of the “So You Think You Have a Power Law—Well Isn’t That Special?” posting:
Lots of distributions give you straight-ish lines on a log-log plot. True, a Gaussian or a Poisson won’t, but lots of other things will. Don’t even begin to talk to me about log-log plots which you claim are “piecewise linear”.