Although I like the theory behind it and the derivation of component statistics, most of the analysis hits into correlation-implies-causation.<p>The linear regression especially: a large number of points <i>causes</i> a large number of comments, not the other way around.