Yesterday morning I was speculating about the surprising abundance of instances where there appears to be a normal distribution of an attribute. My thought process was something along the lines of:
There seem to be an awful lot of instances of normal distribution.Today I come across:
Well, probably a lot of those distributions are near-normal but not perfectly normal distributions, so perhaps it is not as common as I assume.
But the utility for quotidian decision-making of switching from an average number to the concept of a distribution is so high, that it exceeds any error arising from deviance from normal distribution.
Even if they are only near-normal distributions, why does it occur with such frequency?
Lyon argues that, in many cases where people say that the central limit theorem can explain why something is normally distributed, it really isn't clear that it actually can.
— Philippe Lemoine (@phl43) February 1, 2018
Lyon is getting to a much deeper aspect of the "why" than I have the maths chops to follow.
Posts
No comments:
Post a Comment