"Robert Sturgeon" wrote in message
...
On Wed, 31 Dec 2003 17:12:11 -0800, "PrecisionMachinisT"
wrote:

Ed Huntress" wrote in message
et...
"PrecisionMachinisT" wrote in message
...



Heh.......

But what this all has to do with the price of tea in China is *all*
I
wanna to know.

Gunner posted a quote from some blog to the effect that the US has much
less
total crime than major European countries, based on Interpol reports.
It
isn't so, which is what the discussion has been about ever since.


Well, Duhh......I knew that all along, Ed.

Its all about attempting to correlate two or more sets of facts or
statistics that are at often best remotely connected so as to represent a
single "fact" in order to promote some agenda.

Kirk said it quite well IMO.........

Suppose, for instance, one set of data is gathered as to the price
fluctuations of tea in China over a period of time.

Another set of data is gathered concerning deadly automobile accidents in
the USA over the same time frame.......

Suppose further, when one chart moved positive, the other moved
negatively
more or less, and did so throughout several cycles over the time period.

Would this mean there is a definite correlation between the two data sets
???

Why, of course not !!!

I sure hope some genuine statistician jumps in here and
straightens out "Precision." I could, but not being a
professional, I might get some term slightly wrong and the
"Credentialed Society" nazis would have a fit.

But in fact, it is exactly this type of ( Faulty ) statistical comparison
that is *very* frequently used, and presented in order to sway people's
opinion on any number of topics and agendas..........it is simply one
area
where statistics and probability cross the line from being legitimate
tools,
useful for scientific analysis into the realm of being nothing other than
pure propaganda.

Yup.

I think that both of you guys had better go look up "correlation." Indeed,
the situation described is a definite correlation -- a negative one. It's
exactly the kind of correlation that linear-regression analysis looks for
(as in data-mining programs) and the software will flag you that a
correlation exists if it finds one like this.

What I think you mean is that there is no indication of *causation*, and
that's entirely true. But if you used the data to predict the next
upswing/downswing, and if it occurred as the data predicts, you would not
only have a correlation, you would have a predictive correlation. That's the
kind of data relationship that makes astute people rich, and the rest of us
scratch our heads.

Predictive correlations are the basis of much of the marketing statistics
that are in use today. Whether one event causes the other hardly matters --
likely there is some unknown, third factor which is the causative one for
both of the events you're tracking -- but the ability of the data to predict
is its value. It starts with a correlation like the one you've described.

Ed Huntress