Im guessing author used data like this (http://blogs.wsj.com/wealth/2007/02/01/rich-o-meter-20/) which is household, not individual