[Marxism] International Socialist Review on Pharmaceutical

Jeff meisner at xs4all.nl
Sun Apr 26 10:18:04 MDT 2009


At 11:39 26/04/09 -0400, you wrote:
>
>Jeff seems to have an oddly constricted definition of "science," since  
>he excludes from it all the discoveries from many thousands of years  
>of human experience (except, perhaps, for that small portion which has  
>been torn out of its living context and subjected to a statistical  
>rigamarole
By casually referring to "statistical rigamarole", it appears you have no
idea how science is done. Almost every scientific result RELIES on
statistics, an essential part of scientific methodology. That is because
almost every measurement includes the effect of what is technically called
"noise," that is, a level of randomness in the result which reduces its
precision. It is ONLY with statistics that noise can be taken into account
in order to draw conclusions (and those conclusions are only valid to
within a "statistical significance" or quantitative results to within a
"confidence interval," which are themselves quantified by statistical
analysis).

Scientific conclusions which are so stark that they do not require
statistics were usually discovered a long time ago and are not on the
cutting edge of scientific research. For instance the direction of gravity:
down rather than up. That can be figured out without using statistics. But
it does not defy statistics! I could conduct an experiment: let go of 8
apples and record whether they fall down or up. If the apples fall down in
8 out of 8 cases, I could conclude STATISTICALLY that gravity is downwards,
with a statistical significance of 99.75%. Granted, that experiment is
rather trivial. But a measurement of the actual gravitational acceleration,
for instance by measuring the time that it takes an object to drop a
calibrated distance, will not give an exact number, but a statistical
result whose precision is dependent on the measurement apparatus etc.

In other words, use of statistics is at the heart of physical sciences.
When you abandon statistics, then any result can be conjectured without a
valid test. For instance, if I give 100 subjects sugar pills and 100 salt
pills, I might find that more people who took the sugar pill got relief
from their headaches. Another researcher could conduct the same experiment
and find more success with the salt pill. We could both proclaim our
"conclusions," that one is better for headaches, and we'd both surely be
wrong. That was not a scientific test precisely because it didn't appeal to
statistics (which would have shown lack of significance in the result). Nor
would it be reproducible in a consistent sense (especially by independent
researchers, where the bias of the experimenter has been eliminated as a
possible factor in the data).

Yes, there is a great deal of knowledge which was obtained before (or
simply without) use of rigorous statistical analysis. For instance many
traditional medicines (yes, including ones that aren't embraced by the
western medical establishment). But the important thing is that they COULD
(and indeed, should) be confirmed using scientific testing, if they are
valid. And that is how you can differentiate between the ones with REAL
value, and the many others which have survived the passage of time by
virtue of the placebo effect. 

- Jeff





More information about the Marxism mailing list