Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: [Phys-l] explanatory and response variables (was calibration )



On 08/07/2007 06:22 PM, Ludwik Kowalski wrote:

I see no harm in using well established mathematical terms -- dependent
and independent variable -- in particular contexts.

What evidence is there that "independent variable" is a well-established
/mathematical/ term? If I search a large mathematical encyclopedia,
I get zero hits for that term:
http://mathworld.wolfram.com/search/?query=independent+variable
The top 4 near-misses are:
-- Exogenous Variable
-- Independent Statistics
-- Independent Events
-- Separation of Variables
which use the relevant words but don't have anywhere near the
relevant meaning.

I think the distinction between "independent" and "dependent" variables
is one of those thinks, like "significant figures" and like "chemical"
change versus "physical" change that exist only in intro-level textbooks
and don't exist in the real world.

Yes, objects do not
explain objects and variables do not explain variables.

Agreed.

Explanations
are logical chains of cause-and-effect created by humans. A statement
like "too much fat in what you eat explains your overweight" is a
shortcut. A doctor is saying that an acceptable cause-and-effect chain
can be established between the diet and the overweight. The same is
true for a statement "more rapid loss of potential difference, between
two initially-charged plates, is caused by humidity."

Agreed.

In this case
humidity is the cause (in an explanation) while the time after which
one half of the charge is lost is the effect. What kind of harm might
result from saying that, in this context, humidity is the independent,
or explanatory, variable while the half-discharge time is dependent, or
response, variable? One may argue about validity of some explanations.
But it would be difficult to argue that attempts to explain reality are
not useful. That is the essence of what we do physics.

We agree that finding cause-and-effect relationships is very valuable.

One of the most basic skills involved in finding such a relationship
is _recognizing when one has been found_ ... and conversely another
vital skill is _recognizing when one has NOT been found_.

For more on this, http://www.av8n.com/physics/causation.htm

It is well known in anthropology that people have a tremendous tendency
to find pseudo-cause-and-effect relationships when no real, objective
cause-and-effect relationship is present. The classic example is the
story of a tribesman who tripped and injured his leg. The witch doctor
"explained" that it was "because" the guy had worn mismatched socks
that day. It stands to "reason" that if you go against the natural
order of things, bad consequences are likely to follow.
(This is *not* a made-up story. Alas I can't presently lay
hands on the reference.)

So the question is, are we witch doctors or physicists?

IMHO it is manifestly harmful when somebody abuses cause-and-effect
by applying it to a task such as "calibration" or "curve fitting",
because the task is so general that it may or may not have anything
to do with real cause-and-effect relationships.

Let's be clear: I am 100% in favor of cause-and-effect reasoning.
I think it is really, really important. It is so important that
I think it should be done right.

A theorem that is correct /part of the time/ is not a theorem. By
the same token, terminology that is correct /part of the time/ is
not acceptable terminology. Sure, you can find examples where
curve fitting is used in connection with real cause-and-effect
relationships ... but the problem remains that there are innumerable
examples of where it is used otherwise. Using cause-and-effect
terminology when no such relationship exists sets science back
several hundred years.