[Peace-discuss] Mean Streets (2) (David Sirota on Populism, too)
E. Wayne Johnson
ewj at pigs.ag
Mon Apr 27 10:46:15 CDT 2009
The Italian mathematical economists made a tremendous contribution to
the field of the study of distributions in statistics, and that
contribution is seldom given its rightful position. Perhaps it's
because they were Fascists. But whether or not one wants to be subject
to their ideology, their statistical mathematics is quite robust and
interesting.
Much of the inferential statistics that is done in the sciences assumes
as a first principle that the data is normally distributed, and there is
quite often a substantial artifice employed to invoke the central limit
theorem and force the data into a Procrustean bed and torture the data
until it confesses the original premise that prompted the study...
Laplace and Quetelet contributed much to the development of the familiar
and symmetrical bell-shaped "Normal" curve with its skewness of Zero.
JC Kapetyn, an astronomer and mathematician, noted in 1903 that in
Biology the distribution of populations is quite often skewed rather
than "Normal", and proposed that the Underlying distribution was indeed
"Gaussian Normal" but that a second distribution of events, or a series
of unfortunate transforming events is at work that modifies the native
normal distribution to a skewed distribution. Kapetyn demonstrated how
the log-normal distribution could result from such a "Kapetyn
Transformation." It is quite common in the design and analysis of
experiments to get rid of skewness. But in the real world, it's the
Skewness itself that is sometimes more interesting than inference from
means because the skewness tells us about something that is happening to
individuals. All animals are created equal (normally distributed) but
some are more equal than others (Kapetyn transformed and Orwellian
Skewed).
The Italian economist with the unlikely name of Luigi Amoroso recognized
about 1910 or so that the series of unfortunate Poisson distributed
events are described by Euler's gamma distribution, and if one plugs
that back into the normal distribution as a Kapetyn Transformant, the
significantly useful "Generalized Gamma Distribution" is the result.
But Amoroso's work didnt get much traction in the Sir Ronald Fisher
dominated field of inferences and normal distributed data, but the
Generalized Gamma got rediscovered about 50 years later and is deeply
related to the Negative Binomial distribution although the pedants will
inform us that the one is discrete while the other is continuous as if
there is some information transferred in such a declaration. The
neg-binomial is the distribution of the number of failures in a sequence
of tries needed to get a specified number of successes, while the Gen
Gamma tells us basically the same thing about errors and unfortunate
events that occur along some cumulation of X. And success can be
fractional and messy.
In a parallel development in 1905, Max O. Lorenz published a paper
called /Methods of measuring the concentration of wealth,/
he details a method for graphical representation of the distribution of
wealth by plotting the cumulative distribution function of wealth.
Perfect equality (uniform distribution) gives a straight line at 45
degrees. The Lorenz curve lies below the line of perfect equality.
In 1912, the Italian Corrado Gini offered a mathematical representation
of the difference between the Lorenz curve and the
line of perfect equality. This "Gini coefficient" has a wide range of
applications in descriptive and inferential statistics.
Population biologist Christian Damgaard has recently (2000) developed a
Lorenz Asymmetry Coefficient which is based on the shape of the Lorenz
curve that tends to point out where the wealth or stuff is concentrated.
Anthony Pomonis wrote:
>
> As "soft" as economic or political sciences /may/ /be,/ any move to
> dismiss their insights/ /might be too casual.
>
> http://en.wikipedia.org/wiki/Pareto_principle
>
> Thank you to all for a lively discourse...
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Peace-discuss mailing list
> Peace-discuss at lists.chambana.net
> http://lists.chambana.net/cgi-bin/listinfo/peace-discuss
>
-------------- next part --------------
Skipped content of type multipart/related
More information about the Peace-discuss
mailing list