Re: To Jim on Weaving

Rue L. Cromwell (cromwell@KUHUB.CC.UKANS.EDU)
Sat, 09 Mar 1996 00:08:36 +0000

>Content-ID: <>
>Content-type: text/plain
>Attach file.
>Content-ID: <>
>Content-type: text/plain;
> name="WEAVE"
>Content-Transfer-Encoding: quoted-printable
>Thanks for expressing your interest in the problem
>of "weaving constructs." I am sure this has been =
>a concern of yours for many years and that you are =
>aware of the extensive ramifications of the problem =
>for not only construct theory but for science in =
>general. Some of our colleagues may not be familiar =
>with the issues, however, so I would like to hold =
>back from presenting the mathematical details in =
>this posting, and turn instead to clarifying the problem.
>Joe Rychlak's philosophy of science has been most =
>instructive to me on these issues. In particular, has =
>been his treament of Aristotle's notion of formal, final, =
>efficient, and material causes. Final cause concerns =
>the classification or shape of an item; final cause =
>concerns the purpose behind the generation of an =
>item; efficient cause concerns the actions that generate =
>an item; and material concerns the stuff of which =
>something is made. A chair is a chair because it
>contains a back, four legs, seat and arms (formal
>cause). It is a chair because someone wanted to
>build something for people to sit in (final cause). It is
>a chair because a craftsman sawed, drilled, glued, and
>sanded wood in its construction (efficient cause). =
>And it is a chair because it is made of wood and =
>not air (material cause).
> =
>At the beginnings of what we call modern science,
>the formal and final were pretty much outlawed
>and the efficient and material were accepted as =
>the only legitimate paradigms for causal inference.
>This probably came about because of the rejection
>of the church dogmas. An order of being had been
>hypothesized by the church fathers but there was
>no way of demonstrating the validity of the order of
>being, except by a kind of blind faith. The new
>scientists wanted to get away from things spiritual
>and used the manipulation (efficient cause) of
>material objects as their vehicle. Materialism
>eventually arose, with behaviorism and biological
>approaches being, probably the clearest expressions
>of this materialism in psychology.
>As "science" stands now, the "weaving of nature"
>(causation) can only be revealed by experimentation.
>This is because there are no purely formal cause
>methods for demonstrating the inclusion of objects
>in larger objects.Thus, if we say good will+ability+
>community=3Dscientific advance, the only way modern
>science knows to test this equation is by the
>manipulation of the presumed independent variables.
>This makes the study of ideas, constructs, and
>so forth difficult because they are not literally physical
>objects and because of ethical constraints that arise =
>should we manage to operationalize these terms in
>some physical expression. The best we can do with =
>the less physical types of being is to use correlations
>(factor analysis, etc.) But correlational methods do
>not disclose causal relations. They can not tell use
>if constructs are being woven together causally, but
>only if they occur in one another's temporal or spatial
>presence. Such co-occurence may be incidental and
>not causal. This is a restatement of the problem of
>experimentation versus correlation, that Cronbach =
>discussed years ago. It has been central to the division
>of appied (usually clinical) and experimental
>psychologists into disparate groups. =
>SCIENCE is not just a static concept developed
>in the wake of Bacon, Locke and Hume. Its been
>around a long time and will prosper for a lot longer.
>There is no need for us to restrict ourselves to an =
>interpretaion of science that was largely a reaction
>against the struggles of the old church. The church
>could not demonstrate the chain of being =
>mathematically from observations. In order to do
>this they would have had to have shown that by
>mere observation and analysis , we can say A is =
>a subset of B. That is, they did not have a
>mathematical/logical way of measuring formal
>cause. And we have all been paying the price,
>ever since.
>In a series of papers that were edited by Joe
>Rychlak, I proposed a mathematical way of =
>measuring formal causes. The method was =
>developed from my attempts to synthesize Rychlak's =
>philosophy with Kelly's notion of superordinacy. =
>It is not the sort of thinking that most scientists =
>and mathematicians engage in today. The received
>view is that no mathematical procedure for =
>measuring formal cause is possible, even in =
>principle. The papers on corresponding regressions, =
>suggest otherwise. Many simulations (over 30,000)
>and analyses of real data support the validity of the
>method of corresponding regressions. At first I sent
>the paper to Psychological Bulletin. They kept it for
>6 months and returned a review by only one reviewer.
>The logic of the method was completely ignored, as
>was the data. The reviewer said that I was simply
>"trying to turn iron into gold." This is the likely =
>response anyone will receive should she attempt
>to measure the "weaving together of constructs
>" in a nonexperimental but none the less empirical
>I mention my frustations not simply out of self pity.
>I sent copies of the paper to over 100 psychology
>methodologists, asking for their criticisms. About
>6 old and eminent psychologists responded. These =
>guys were some of the most highly esteemed psycholgists
>alive. I admit, they gave my ego quite a boost. They told
>me that they could find no error in the method's logic
>and that it may well prove to be a "revolution " in
>statistics. But none cared to say this in public, nor
>apparently to use or test the method. They were =
>apparently just too tired or wary to test the method
>publically because of the revolutionary nature of the
>idea. This is what we who would trace the weaving
>of constructs are up against. Without Joe Rychlak,
>the method would never have been published. I think, =
>whether or not the method ultimately survives the
>tests of time, that Joe and Ray Russ deserve our
>respect and appreciation. Joe's affinity for Kelly's work
>is strong. I think, if nothing else, our addressing these
>issues is testimony to just how important construct =
>psychology could be for all of science. If we solve this
>problem of weaving- and we apparantly are the only =
>folks who will dare to try- then we will make a very =
>significant impact on the advance of science. Please =
>do not take this as grandiose. I may have proposed a
>mathematical solution, but I know very well that it
>arose from my participation in the PCP endeavor. I =
>also know that unless some sharp but open intellects
>put the method and ideas to the test, it will probably
>die with me, whether it is correct or not. Without =
>help from the PCP community, I doubt that the ideas
>will ever be tested and elaborated. I do believe, =
>however, that if it works, WE will show the materialists
>that we are capable of solving a lot more than our
>family fights. I am not suggesting the method might
>do it all. It really makes no sense with out Kelly and
>Rychlak. That makes me and my ego a footnote.
>I have been called worse.
>If ya'll are interested, I will walk you through the
>method of corresponding regressions. I'll leave
>you with just a thought. Would it be useful to be
>able to say from a grid or collection of grids that
>a person's construct D is the synthesis of his
>constructs A and B but not C? If you think so too,
>I could sure use your help in sorting things out.
> =

Your ideas, stated above, are of central interest to me from a number of
vantage points. (I often wish I were free of other commitments so that I
could give central attention and effort to working through these interests.)

>From one vantage point, it seems to me that de Boeck's (and Rosenberg's)
HICLAS (hierarchical cluster analysis of matrices) approach the matter of
whether "construct D is the synthesis of his constructs A and B but not C."
True, the HICLAS method is based upon set theory (allowing asymmetry) rather
than correlation (forcing symmetry). Moreover, it can deal with only binary
(dichotomous) constructs (or else a repetition of binary operations). Yet,
some HICLAS solutions appear to be more pwoerful in repesenting (predicting)
events than does a correlation approah. In what way is your approach the
same or different?

As a parallel question, in what way is your approach the same or different
from structural equation modeling (as by Bentler and others)? Is it that
yours does not require an a priori model? Are there more distinctions to be
made between your approach and that approach?

>From an entirely different vantage point, I have long been struck by the
fact that PCP (within the narrow range that I know it) attends to one kind
of constructive relationship but not to another. Stated simply, the
structure of languages all over humankind characterize transitive and
intranitive relationships. Stated more concretely, they involve sentences
with either (a) forms of the verb "to be," (b) forms of the verb "to have"
(which in fact can be a variation of the verb "to be"), and (c) action verbs
(whether in the active or passive voice grammar does not seem to be
important). To me, the Kelly rp grid approach, as so far applied, deals
essentially with the first two of these but not with the third. Yet, to me,
it is the third which reflects the heart of how people conceptualize cause
and effect. To me, the latter is distinctly different from dealing with the
associations among attributes (wherein PCP and rep grid have made great
strides. But, when I have made pleas for analytic procedures and rep grids
which deal with cause effect (action verb propositions), I tend to get a non
response. I am led to conclude that I am naive and overlooking some
development in PCP so that this particular idea is outmoded, or that I have
not made myself clear. I find it difficult to believe that the idea is so
new or ahead of its time that it is hard for others to fathom.

Care to comment on the questions raised by these three vantage points?




Rue L. Cromwell