How to make a statistical inference in fat tailed domain. Typically statisticians don't deal with power laws, which is left to physicists who, owing to the stability of their domain, do not do statistical inference. We show errors of using the sample mean as an estimate. The presentation discusses some papers including debunking Pinker's bogus statistics in his book on violence.
[munger]
Nassim Taleb: The Law Of Large Numbers And Fat Tailed Distributions
