With all respect and amity to the likeable and communicative David Colander,
I do not believe there can be data to "start with". There must be defined
units about which data is collected and presented. Economists have not given
careful enough thought to defining their terms. They have, of course,
tortured dozens of issues in the definition of national income, but the
present mess is the result of too much thought without enough careful
thought. Plus a dash of politics.

 

Often the data must be ranked, and if it is a correlation study one must
choose the ranking variable, which makes a big difference. For example the
U.S. Census of Agriculture ranks farms by "acreage", rather than value. Then
some analysts calculate Gini Ratios therefrom. Worse yet, some (friends I
will not name) then note that the "largest" farms have lower values per
acre, and conclude that the Gini Ratio for the value of farms must be lower
than that for the area of farms. Ugh.

 

Mason Gaffney