----- Original Message ----- 
> On Mon, 10 Sep 2007. Ken Steele wrote:
>
> http://tinyurl.com/3yvxlm

It is possible that I am just behind the times or out of 
touch with what is interesting in the research coming
from various areas of neuroscience but I do get
annoyed reading popular media accounts (or even
some of the journal articles) such as that presented
in the news story linked to above.  Just a couple of
points to keep in mind when reading the story:

(1)  The researchers make the assumption that there is
a unidimensional scale which runs from "conservative"
at one end to "liberal" at the other end.  This type of
"model" of political "stance" may be tempting because,
as students of S.S. Stevens would know, we rely upon
an internal scale of magnitude which allows us to arbitrarily
assign different degrees of magnitude to different discrete
entities which may embody different qualitative properties
(e.g., in giving magnitude estimates of severity of crime,
some might rate/generate larger estimates for, say, rape,
relative to physical assualt but there are likely to be
differences amongs different groups, such as whether
males and females have ever been assaulted or raped).
A unidimensional scale for liberalism-conservatism ignores
important differences among groups initially may be thought
of similar (e.g., both economic conservatives and social
conservatives may be thought of as being highly similar but
libertarians probably don't see that they have much in 
common with Christian conservatives, especially when it
comes to issues of having laws restricting gambling, drug
use, abortion, etc. -- their commonality may actually be
based on their dislike/distrust of liberal democrats use of
the federal government [e.g., restricting gun ownership]
which is different from basing differences in magnitude on a 
unidimensional scale of liberalism and conservatism).
I would suggest looking at John Dean's "Conservatives
Without Conscience" to appreciate how many different
types of conservatives there are and some of the reasons
why this is the case (e.g., there are no "canonical" documents
that all conservatives can point to as the basis for their
political beliefs while liberals can reference to the U.S.
Declaration of Independence and other political documents
that assert the equality of all peoples and the rights that
all people are entitled to).  One has to ask, what does a
correlation between a unidimensional scale like conservatism-
liberalism to any other variable mean when the actual
variable (i.e., conservatism-liberalism) may be multidimensiona?.

Sidenote:  Bob Altemeyer's Authoritarianism Scale is presented
as being unidimensional but consisting of three "subareas"
which Altemeyer has "mixed together" in the items of his
scale.  Others, such as Funke (2005; Political Psychology,
26(2), 195-218) have revised Altemeyer's scale to disentangle
the three subareas have has found that a correlated three 
factor model (one factor for each subarea) is a better
description than a single factor model, as predicted by
Altermeyer's theory and research.  Although a single factor
model is simpler to understand, it may be an oversimplication
that obscures potentialy interesting  differences among groups.
One might make the same argument regarding a unidimensional
conservative-liberal scale.

(2)  I wonder if people, especially Ph.D. and M.D. types
are more likely to confuse CORRELATIONAL relationships
with CAUSAL relationships if the correlational relationship
involves neurologically based variables?   For example,
if there is a correlation, say, between scores on a unidimensional
conservatism-liberal scale and activity in the anterior cingulate cortex, 
(ACC) which of the following statements is correct:
(a)  holding a conservative or liberal belief determines the level
of activity in the ACC?
(b)  level of activity in the ACC determines whether one holds
a conservative or liberal belief
(c)  the observed correlations between conservatism-liberalism
score and ACC activity to due to some third variable, possibly
cognitive, neural, social, experiential, etc., that was not controlled
in the experiment.
(d)  there is insufficient information provided to allow one to
develop a coherent explanation without making ad hoc, post hoc
speculations which were not tested by the presented research.

I'm not sure sure which answer above is really correct though I
do lean towards (d).

I would also like to suggest a research project for anyone who
is interested in such issues or may have students who are interested
in such interests:

Construct vignettes where two variables X and Y have either
correlational or causal relationships and then "frame" these in
neuroscience and non-neuroscience situations.  So, the deep
structure of the problem is determining whether the presented
relationship is correlational or causal but the surface form of
the problem is a neuroscience context vs. a non-neuroscience
context (it should fairly easy to construct the 2x2 table for this
design).  It could be just me but I have the feeling that people,
ranging from the proverbial "man on the street" to undergraduates
to professional researchers may call correlational relationships
causal when presented in a neuroscience frame relative to the
correlational relationship being presented in a non-neuroscience
framework.

I would be interested in seeing the results of such research
because it is my impression that correlations in neuroscience,
as described above, are more readily interpreted as causal
in nature though I'm not really sure why (is this due to the
way neuroscientists present their results, does having a 
neuroscience/biological frame bias one towards asserting
causal relationship, etc.).

Anyway, I guess the most useful bottom line statement is that
everyone should maintain some skepticism about popular
media presentations of psychology.  Remember Freud and the
iceberg metaphor.

-Mike Palij
New York Univeristy
[EMAIL PROTECTED]





---

Reply via email to