[crpatino]

Sorry, but splitting "classes" in thirds is just ridiculous. The way the economy works is by concentrating wealth, so any class definition based on clustering according to income/net-worth cannot be linear. A logarithmic scale would make much more sense, by example: Poor:= 0-50%, Middle class:= 51%-75%, Upper class:=76-100%???

This model is extremely rough and inaccurate, but is a step in the right direction. The main problem it has is to assume that there has to be only 3 classes, which is arbitrary and does not describe the actual lives of real people. A second, related defect is that it does not recognize the existence of a small group of very vulnerable people below the "working poor".

For these reasons, I think it is best to model this clustering classification using a log-normal distribution, with at least five classes (Underclass < Working poor < Middle class < Upper-middle class < Upper class). Still, the "upper class" category lumps together the merely wealthy (smallish business owners, the most successful professionals) with the extremely rich, and all possibilities in-between.

[ahomescu1]

> Still, the "upper class" category lumps together the merely wealthy (smallish business owners, the most successful professionals) with the extremely rich, and all possibilities in-between.

It also lumps in the high-income earners with the high-net worth people (the ones everyone thinks of when they say "rich").

[crpatino]

agreed, you can only do so much when modeling a complex system with a single variable distribution.