file.newsgroup.med.58853 Maven / Gradle / Ivy
From: [email protected] (Russell Turpin)
Subject: Re: Science and methodology (was: Homeopathy ... tradition?)
-*----
I agree with everything that Lee Lady wrote in her previous post in
this thread. In case this puzzles people, I would like to expand
on two of her comments.
In article [email protected] (Lee Lady) writes:
> Avoiding mistakes is certainly highly desirable. However it is also
> widely acknowledged that perfectionism is inimicable to creativity.
> ... In the extreme case, a perfectionist becomes so paralyzed by all
> the possible mistakes he might make that he is unable to even leave
> the house.
One of the most important (and difficult) aspects of reasoning
about empirical investigation lies in understanding the context,
scope, and importance of the various arguments and pieces of
evidence that are marshalled for a claim. Some errors break the
back of a piece of research, some leave a hole that needs to be
filled in, and some are trivial in their importance. It is a
grave mistake to confuse these.
Past snippets from this thread:
>>> I doubt if Einstein used any formal methodology. ....
>> He also proposed numerous experiments which if performed would
>> distinguish a universe in which special relativity holds from
>> one in which it does not. ...
Back to Lee Lady:
> These are not the rules according to many who post to sci.med and
> sci.psychology. According to these posters "If it's not supported by
> carefully designed controlled studies then it's not science."
These posters are making the mistake that I have previously
criticized of adhering to a methodological recipe. A "carefully
designed and controlled study" is neither always possible nor
always important. (On the other hand, if someone is proposing a
remedy that supposedly alleviates a chronic medical problem, we
have enough knowledge of the errors that have plagued *this* kind
of claim to ask for a "carefully designed and controlled study"
to alleviate our skepticism.)
Rules such as "support the hypothesis by a carefully designed and
controlled study" are too narrow to apply to *all* investigation.
I think that the requirements for particular reasoning to be
convincing depends greatly on the kinds of mistakes that have
occurred in past reasoning about the same kinds of things. (To
reuse the previous example, we know that conclusions from
uncontrolled observations of the treatment of chronic medical
problems are notoriously problematic.)
Russell