[Martin Taylor 2012.06.01.07.01]
[From Rick Marken (2012.05.31.1745)]
Jim Wuwert (2012.05.31.1410)–
You mentioned in a previous post when you presented
the data written by Robert Reich and published in the
New York Times that conservatives never present data.
They just point out flaws in your data. The data
should stand on their own merits. It is valid to point
out flaws in methodology and data. The data that you
present is what is at issue not the fact that we may
or may not be able to prove it to be untrue. Or, the
fact that we may or may not present a new model.
RM: The problem here is that pointing out "flaws" implies that
you know what the “unflawed” data are. Unless you can point to
the unflawed data that supports your point of view saying that
the data are “flawed” is just a dodge.
What could it mean to say data are flawed? Let's ask this question
from a PCT perspective.
What are 'data' within the PCT framework? I can see no answer other
than that data are inputs to one or more perceptual functions, the
outputs of which is the variables to be controlled or monitored (not
all perceptions are being controlled at any moment; those that are
not are said to be being monitored, either consciously or not).
If data are inputs to a perceptual function (let's limit the
discussion to one, for the moment), what could it mean to say they
are “flawed”?
"Flawed" means imperfect. Data can be imperfect in several ways: (1)
They misrepresent the value of an environmental property (e.g. you
tell me it is sunny and dry outside, but when I go out I get rained
on); (2) they are insufficient to allow the perceptual function to
generate a value (e.g. you are on a boat in the middle of a lake in
the fog, and you cannot determine the direction of your home dock);
(3) they correctly represent the wrong environmental property (e.g.
a mirage that correctly represents the sky but is seen as
representing water on the ground); (4) they are incomplete, and
represent the environmental property only approximately (almost all
data have this flaw).
Which sense of "flawed" is intended by those who say that the
various sets of data presented by Rick over the years are flawed?
As for proving something to be true or untrue, since all we have
access to is our own perception, “proof” in the logical sense is not
something that can be publicly communicated. Quite apart from that,
the real world appears to be rather complicated, and whereas
mathematical proof depends on a defined set of axioms, real world
“proof” depends on observations that may well omit some crucial
thing that could have been observed and was not. All we can ever do
in respect of the real world is talk about conditional
probabilities: if there are no unobserved variables that would
influence the result, and if one or other of these two (or more)
theories is true, then the data are more likely to have been
observed if the world conforms to this theory rather than to that.
In cases of dispute, one of the unobserved variables is quite often
the mode of selection of the data. That is why scientists are
usually told very early in their training that you can’t test a
theory by looking at data already gathered. Actually, you can, but
it is all too easy unconsciously to select data that support a
theory that gives results conforming to your reference values for
some of your controlled variables. The same applies to
randomization. Randomization is intended to prevent you from
unconsciously choosing the conditions so that the results you want
will be more likely to be the results you get.
"Flawed" data involving unobserved variables, particularly the
variable of data selection, can readily result in case (1) above,
the data misrepresenting the value of an environmental property.
However, in the absence of demonstrable selection of the data to
conform to one theory rather than another, it is probable that the
theory to which the data conform well is closer to the truth than
the theory to which the data conform poorly.
So I can refine my question above: Is the "flaw" in Rick's data
claimed to result from improper selection of data either by Rick or
by the sources that Rick has quoted? If so, could the problem be
resolved by providing complementary data to reduce the
incompleteness of the data available for consideration by
uncommitted readers?
Martin