On Wed, 12 Aug 2009 23:08:12 -0700, Allen Esterson wrote:
Michael Sylvester wrote:
If scientific findings represent flawless objectivity,
A curious statement if there ever was one. Is Dr. Sylvester actually
posing a counterfactual in contrast to referring to actual beliefs or
conditions? His statement seems to have more in common with
something like:
If pigs had wings
why do need replications?
One would, of course, have to be clearer by what one means by
flawless objectivity, but it appears to embody a highly deterministic
viewpoint about the nature of reality, that is, if everything that needs
to be known about a phenomenon is known AND THERE IS NO
SOURCE OF RANDOM OR SYSTEMATIC ERROR, then a
single demonstration of an outcome in the context of an experimental
design would be sufficient and the need for replication would be negligible
because under these conditions it is expected that the same result would
always be expected (recall the recent thread on whether coin-tossing
representts a real random process; when all relevant variables are controlled
and environmental conditions are kept constant, only one outcome will be
produced).
If you presuppose an erroneous premise as here, the question is redundant.
Well, it is not clear what the premise is and given Dr. Sylvester's
non-eurocentric mode of thinking about things it is not clear that the
concepts of error, truth, or meaningfulness (outside of a limited
community) are relevant. He could, however, be stating a counterfactual
in which case the truth of the premise is not the issue: start by assuming
that the premise is true and then extrapolate/deduce what follows.
If, however, Dr. Sylvester's conditional statement is actually meant to
refer to actual scientific practice, then it would be worthwhile for him
to provide cases where replications are not necessary.
Working scientists, on the other hand, might view Dr. Sylvester's statement
as either (a) bizzare because few scientific findings are based on flawless
objectivity (flawed objectivity, that is, observations are made with error and
this
is common practice [psychometrics is concerned with estimating and controlling
different types of error] because there are few situations where errors can
be forced to be zero or close to zero) or (b) the purpose of experimental design
is to identifiy all of the variables that may affect a causal relationship that
is the
focus of the experimental design and attempt to control for them or minimize
the effect of these nuisance variable on variables involved in the causal
relationship under study.
A researcher with good experimental design background will be familiar with
the threats to internal validity which would undermine one's claim that a causal
relationship exists, threats to external validity which would limit the extent
to
which the causal relationship applies, threats to statistical conclusion
validity
which would undermine whether one's analaysis of the causal relationship is
in fact statistically valid, and so on with other types of validities such as
construct validity, etc.
Scientific findings are unlike mere anecdotes where the focus is on how good
a story is and whether our cognitive heuristics readily process the story
points
(e.g., does the story match up with concepts activated by the representative
heuristics, does the availability heuristic work to activate relevant recent
event
to serve as partial support for the story, does the simulation heuristic allow
one to
construct the causal thread that connects one elements/episodes in the anecdote
into a coherent thread that holds all of the statements made in the anecdote
together?).
People may believe that what they say is true, honest, accurate, and sincere.
Of course, as research on eyewitness testimony has shown, belief in what one
thinks and the factual foundation for what one thinks are two different things.
As an exercise, I found suggest watching Akira Kurosawa's film Rashomon.
In one sense it is a police procedure like a CSI or an Land Order episode:
a crime was committed and the authorities must decide who is responsible.
The trick that Kurosawa uses is that of the unreliable narrator, that is, the
narrator in a story, play, film cannot be trusted to provide an accurate,
factual,
and honest narration (a reliable narrator obeys Grices conversational maxims
while an unreliable narrator violates them and more); see:
http://en.wikipedia.org/wiki/Unreliable_narrator
Kurosawa embeds narrations within narrations and the naive film view will easily
lose track of who is relating what (it is the woodcutter who provides the master
narration and relates what he remembers of the narration of other actors -- but
it becomes clear that he is an unreliable narrator because he does not want to
admit to the crime *HE* committed, namely, stealing a pearl handle knife that
may have been used in killing one of the main characters).
The film Rashomon is a great story
