Re: Chaos and topographic diagrams

["Alla Linetsky" <alinetsky@home.com>]

Bob Adsett has brought up a point that has plagued statisticians for
eons. Chaotic data cannot be modelled with a neat statistical
distribution, while excessively complex data consist of many
superimposed distributions. The only way I know of to distinguish
chaos from excessive complexity is to try to segment the data by
whatever criteria makes sense for those data. For socio-economic data,
the segmentation criteria are usually income, education, family
status, age, country of origin...  Of course, the most important
criteria are extremely difficult to determine and are impossible to
quantify:  the closeness of the family unit, people's nutritional
habits, how secure they feel , their upbringing...

I've often suspected that some studies are deliberately designed in
such a way that the test population is excessively complex, and as a
result, the study shows no significant effect. A case in point is a
study of the effectiveness of echinacea, reluctantly carried out by
some pharmaceutical companies (surprise, surprise, they found no
evidence that it made any difference for viral infections). The same
fudging is probably going on with trying to determine a link between
mad cow disease and Creutzfeld-Jacob disease, or between mad cow
disease and animal-source feed for cattle, or between CJD and
brain-wasting diseases in other cattle.

As for Bob's comments regarding pouring money into community projects,
the problem here is probably the opposite of excessive complexity. It
is very easy for a politician to find at least one person who,
possibly as a result of the program, has improved his lot in life.
Such people become poster boys and girls for the initiatives, although
the public sees the far greater proportion of people for whom the
program has had no effect. Did the program actually help a certain
type of "client"? It would be extremely difficult for an outsider to
determine that; one would have to gather the people who benefitted and
those who didn't and try to determine what distinguishes the two
groups. It's pretty thin ice.

Alla Linetsky