Common Ground: Abstraction vs Construction
Tue, 5 Mar 1996 11:03:02 -0500
Content-ID: <>
Content-type: text/plain

Attach file abcon.
Content-ID: <>
Content-type: text/plain;
Content-Transfer-Encoding: quoted-printable

Common Ground: Abstraction versus Construction
The creation of common ground is not the same
as abstracting common ground. Pretending will =

do no good in the long run. This issue has plaqued =

the analysis of repertory grids. Do we measure =

abstractions or constructs?
Kelly linked the development of constructs to =

abstraction of common themes in life's real events. =

He may have been short sighted in this in that =

abstraction in its purest sense, entails the sloughing =

off of features that are not shared by all of a group =

of constructs. Abstraction essentially is preemptive =

(exclusive). For example, in classical psychometric =

theory (CPT), total scores are derived by adding a =

group of questionnaire items. The assumption of =

CPT is that each item is a parallel test of the =

single trait they all have in common. Their =

differences are assumed to be due to either random
measurement errors or to the fact that a deviant =

item does not really measure only the factor =

under consideration. Adding factorially pure items- =

those that only differ in terms of random measurement =

errors, creates an abstraction, not a construction. =

The random "errors" cancel out and the common =

variance that remains was already there, not created.
CPT works when there is only one factor or ground. =

With more than one factor, total scores may be still =

calculated but they no longer deliver an abstracted =

essence.This is most evident when items are =

uncorrelated (orthogonal). The abstraction of =

essences requires correlated items. Uncorrelated =

items have no common ground from which an =

essence can be abstracted. =

Of course sometimes people fake it.They ignore =

Cronbach's alpha. Others conduct a factor analysis, =

which is more honest, in that it allows us to create =

total scores (common ground) for each separate factor. =

This a common practice in grid research. The grid is =

factored and then a map of the construct system is =

plotted across the factors or components. This can =

be useful but there is a danger. It is assumed that =

the factors can legitimately be crossed and items =

plotted in their common space. The practice derives =

from the implicit assumption that the abstract factors =

actually interact in their expressions through the =

elements (items). They may not. It may be improper =

to plot elements on two factors at the same time. =

The factors may not actually share common ground.
Well, don't they share their elements? Aren't they both =

existent in the elements? Maybe but perhaps only =

incidentally. Piaget described this incidental clumping as =

syncretism. Coincidence in time or place are taken by =

the young child as sufficient grounds for equating two =

features. Poor scientists do this by seeking only to affirm =

the consequent in their research. The constructs that =

characterize people may have no interaction with one =

another semantically, but only coexist syncretistically. =

Factor analysis will not tell us one way or the other. =

Thus simply abstacting factors and clumping them
together will not necessaily deliver the dialectical =

synthesis of true common ground,i.e. construction. =

Construction ultimately consists of the synthesis of =

orthogonal essences. Construction is what you do =

after abstraction. The conjugation of essences has =

a long and distinguished history in science. Mendeleeff's =

Table of Elements is an example of how the interaction =

between abstraction and construction is useful. He used =

a logical framework to direct the discovery of yet =

undiscovered elements. The building blocks of chemical =

elements could be arranged in many ways logically. At =

first, a number of these logical possibilities had not been =

observed in nature. Using a latin square type approach, =

in which all combinations of orthgonal building blocks =

were combined, Mendeleef predicted certain undiscovered =

elements. But the prediction was only confirmed when =

nature's dialectic was discovered. Only logically =

conjugating a balanced sample of building blocks =

into an integratively complex soup would not have =

created the common ground of the yet undiscovered =

elements. Nature had to do the synthesis, =

fair and square, not syncretistically. Nature constructed =

the undiscovered elements. This creative surge of =

nature is what is missing in the stalemate grid.
We may clump our factors together in grid reseach by =

assuming their coincidence in elements. But this is not =

finding common ground, at least not with regard to =

construct systems. Until there is evidence that the =

constructs are being woven together by the person, it =

is not fair to say our factor plots are valid. They may be =

simply syncretistically clumped. How do we know that =

the person adds construct A and construct B in his =

construction of a person? A possible solution is to =

use the method of corresponding regressions to =

determine if some constructs are additive composites =

of others. Where such composition occurs, we may =

properly speak of construction. Otherwise, we may =

only be measuring piecemeal applications of =

abstractions. It would be like saying the coinicidental =

appearance of many points of view on Mr. =

Nightingale's net is evidence of construction and =

common ground. Such syncretistic posting may seem =

sexy, but it would really be something unnatural.
What do ya'll think?