University of Southern California Dissertations and Theses

Alternatives In Research Design: The Measurement Of Goal Conflict

(USC Thesis Other)

Alternatives In Research Design: The Measurement Of Goal Conflict

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS
This material was produced from a microfilm copy of the original document. While
the most advanced technological means to photograph and reproduce this document
have been used, the quality is heavily dependent upon the quality of the original
submitted.
The following explanation of techniques is provided to help you understand
markings or patterns which may appear on this reproduction.
1. The sign or "target" for pages apparently lacking from the document
photographed is "Missing Page(s)". If it was possible to obtain the missing
page(s) or section, they are spliced into the film along with adjacent pages.
This may have necessitated cutting thru an image and duplicating adjacent
pages to insure you complete continuity.
2. When an image on the film is obliterated with a large round black mark, it
is an indication that the photographer suspected that the copy may have
moved during exposure and thus cause a blurred image. You will find a
good image of the page in the adjacent frame.
3. When a map, drawing or chart, etc., was part of the material being
photographed the photographer followed a definite method in
"sectioning" the material. It is customary to begin photoing at the upper
left hand corner of a large sheet and to continue photoing from left to
right in equal sections with a small overlap. If necessary, sectioning is
continued again — beginning below the first row and continuing on until
complete.
4. The majority of users indicate that the textual content is of greatest value,
however, a somewhat higher quality reproduction could be made from
"photographs" if essential to the understanding of the dissertation. Silver
prints of "photographs" may be ordered at additional charge by writing
the Order Department, giving the catalog number, title, author and
specific pages you wish reproduced.
5. PLEASE NOTE: Some pages may have indistinct print. Filmed as
received.
Xerox University Microfilms
300 North Zeeb Road
Ann Arbor, Michigan 48106
74-21,511
STUBBS, Roy Charles, 1938-
ALTERNATIVES IN RESEARCH DESIGN: THE
MEASUREMENT OF GOAL CONFLICT.
University o£ Southern California, Ph.D., 1974
Political Science, public administration
i ]
|
[ University Microfilms, A X E R O X Company, Ann Arbor, Michigan
© Copyright by
Roy Charles Stubbs
ALTERNATIVES IN RESEARCH DESIGN: THE
MEASUREMENT OF GOAL CONFLICT
by
Roy Charles Stubbs
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requi rement s for the Degree
DOCTOR OF PHILOSOPHY
(Public Administration)
June 1974
UNIVERSITY O F S O U T H E R N C A LIFO R N IA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES. CALIFORNIA 9 0 0 0 7
This dissertation, written by
Roy Charles Stubbs
under the direction of hi.?.... Dissertation Com
mittee, and approved by all its members, has
been presented to and accepted by The Graduate
School, in partial fulfillment of requirements of
the degree of
D O C T O R OF P H IL O S O P H Y
MMITTEE DISSERTA'
Chairman
INTRODUCTION
Intellectual advance occurs in two ways.
At times increase of knowledge is organized
about old conceptions, while these are ex
panded, elaborated and refined, but not
seriously revised, much less abandoned. At
other times, the increase of knowledge de
mands qualitative rather than quantitative
change? alteration, not addition.
— John Dewey
"The Need for a Recovery of a Philosophy."
It is only two years since Edgar Schein said:
Our research models still derive from the
experiment or laboratory model, even though we
know that we cannot control key variables in
organizations. One of these days we will have
to face up to the fact that research on dynamic
open systems requires new kinds of research
models and put our creative skills to the task
of inventing such research models.1
It would be foolish to regard Schein as the origi
nator of this point of view, just as it would be foolish
to imagine that the greater proportion of social re
searchers are not aware of the problems, or are not seek
ing ways to circumvent them.
However, it is possible that, just as the medi
eval scholars declared miraculous that which they did not
know, we too may be expecting magical processes to solve
our future problems, when workable solutions may already
be at hand in some unrecognized, untried form. What may
be needed is not a new technique, but simply a new ap
proach: alteration, not addition.
At the present time, for instance, there is much
discussion of phenomenological research, which attempts
to perceive the truth through "bracketing," a procedure
which Husserl never satisfactorily elucidated, and which
still awaits its modern systematist. Of course, phenome
nological researchers have been working in sociology for
many years — MacLeod's 1947 article is typical of the
genre^ — but, with their existential underpinnings, such
theorists have seemed to deny the conventional forms of
analysis in favor of a perceptual approach which tends to
stop short at the receipt of sensation. By emphasizing
the "newness" of every examination, the phenomenologists
minimize analogy.
With such limitations, conventional research can
seem to offer little more than the interview as a suitable
tool. More complex methods seem to lie somewhere "out
there."
It is one of the points of this paper to indicate
that such limitations are unnecessarily restrictive. In
doing so, we shall lay no claim to great originality.
iii
Instead of striking out for new techniques, we shall sim
ply lay out the process by which problems appear to us,
and the way in which a research design tries to solve
them. After stressing what seems to us to be the desir
able characteristics of such a design, we will put the
model into practice by running it through a rudimentary
problem. The design will incorporate three things. It
will attempt to satisfy current philosophical demands;
it will represent analytic methods which are presently
accessible; and it will lay the foundation for a more in
clusive, dynamic model of some organizational processes.
The entire structure of the paper was inspired by
a single article of John Dewey on "The Need for a Recovery
of a Philosophy." One need only read our chapter quotes
to follow Dewey's method, and our search for knowledge.
TABLE OF CONTENTS
INTRODUCTION ................................... ii
LIST OF TABLES . . . ............. vi
LIST OF FIGURES................................. vii
Chapter
1 1
II.............................................. 8
III.............................................. 22
IV.............................................. 40
V.............................................. 52
VI.............................................. 63
VII.............................................. 78
VIII.............................................. 107
IX.............................................. 115
X.............................................. 127
CONCLUSION..................................... 144
REFERENCES..................................... 148
BIBLIOGRAPHY ................................... 156
V
LIST OF TABLES
Table Page
7-1 Ideal Decision Losses ................... 89
8-1 Probability Rule Decision
Frequency Comparison ..... .......... 108
10-1 Maximum Entropy T a b l e .................... 139
vi
70
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
133
LIST OF FIGURES
Decision Tree
Loss Function, Decision al
Loss Function, Decision a2
Loss Function, Decision a3
Loss Function, Decision a4
Loss Function, Decision bl
Loss Function, Decision b2
Loss Function, Decision b3
Loss Function, Decision dl
Loss Function, Decision d2
Loss Function, Decision d3
Loss Function, Decision d4
Loss Function, Decision d6
Minimum Loss Envelope, al/a2 ...........
Minimum Loss Envelope, a3/a4 ...........
Minimum Loss Envelope, bl/b2/b3 . . . .
Minimum Loss Envelope,
dl /Conflict Pattern for Differences
in Goal Time Perception .............
vii
CHAPTER I
The point seriously at issue in the notion of
experience common to both sides in the older con
troversy thus turns out to be the place of thought
or intelligence in experience. Does reason have
a distinctive office? Is there a characteristic
order of relations contributed by it?
— Dewey
In his historical development of empiricism,
Dewey^ points out that there have been at least three
conceptions of it. The classical Greeks considered em
piricism as experience; the Bacons, and Locke, looked for
what was empirical in experience; and the modern thinkers
tended to take a more active part in the process, by
setting up empiricisms which would later be experience.
To put it more simply, the Greeks considered empiricism
as what Dewey called a mere honest accounting of that
which had gone on. Locke saw value in this only so far
as we could produce associations from the mass of sensa
tions with which Nature coerced us. Today, we milk these
sensations, and empiricism, from Nature, by a process of
experimentation.
Within Western culture, then, the conscious pur
suit of empiricism is a relatively modern phenomenon.
2
Kaplan^ estimates that it is only within the last 100
years that we have seen the accumulation of empiricism
in both its epistemic and semantic forms.
From a historic point of view, then, we might ex
pect that experience would take temporal precedence over
association, or theory. But experimentation has made this
relationship ambiguous. We now have Wittgenstein^ assert
ing that the roles of epistemology/semantics and empir
icism are reversed; in other words, that all concepts are
theory-laden. To investigate the formation of new con
cepts is to investigate the changes in old theories.
Administrators, apparently practical on the outside, are
actually relying on their background as theory-makers to
form their decisions.
The two points are not necessarily antithetical,
however. Rather than examining them in terms of cause
and effect, we are better off viewing them as "functional
relationships," in Simon's words;^ in which case cause
and effect become mutual. We have, for instance, Weber's
definition of law:
. . . laws are legitimate if they have been
enacted; and the enactment is legitimate if
it has occurred in conformity with the laws
prescribing the procedure to be followed.
This circularity is intentional.7
3
But to accept this, we have to make our definition
of administrative behavior transcend to a more general
level: we must say, for instance, that the administrator's
view of the world now and in the future will depend on
what his view of the world has been in the past.
Now, this has seemingly generalized the definition
into triteness? but we shall see that, in this form, some
useful developments can be examined.
First, the definition admits experimentation as
a dynamic, rather than a static value, and allows for in
novation through experimentation, if such has been an
element of the past.
Second, the concept relates to other schemes of
innovation. For instance, Schon® has characterized all
concept building as the transfer of metaphors from our
past into our present. The success of the technique
depends purely on the migratory ability of the metaphor:
whether, like a good madeira, it can survive radical
changes in the environment. Miller has noted an amusing
example of this:
It is amusing that so many psychologists
who abhor subjectivism and anthropomorphism
unhesitatingly put telephone switchboards
inside our heads. In 1943, for example,
Clark Hull . . . could take it as self-
evident that the brain "acts as a kind of
automatic switchboard." . . . However, the
important adjective, "automatic," is a re
cent accomplishment. The telephone engineers
4
who had to build and maintain those early
switchboards that reflex theorists loved so
well were dissatisfied with them because
they required a human operator to make the
connections. Eventually, of course, the
operator was replaced by more elaborate
machinery, thus rendering reflex theory
scientifically impeccable at last. But in
1892, when Karl Pearson wrote The Grammar
of Science, he unblushingly provided a
"clerk" who carried on the same valuable
services in the brain as he would in a
central telephone exchange.9
This dynamic view of truth as a traveler, rather
than as a static, essential thing is necessary because of
the inability of the static model to yield "exact" an
swers, or inherent truths. Though Hume had come to this
conclusion in a philosophical way, it was left to Heisen
berg to formulate the idea in more scientific language.
The dynamic view, on the other hand, allows us to accept
truth as a transient quantity, dependent on circumstances.
Inherent truth then becomes a mere special case of tran
sient truth.
Satisfying as this definition may be, however, it
means that in practice, we will never be able to pin down
our meanings, because of the potential infinity of envi
ronments. In place of this, we must rely on two things.
First, the conscious choice of environment, which is
classical experimentation. And second, we rely on the
observation of recurring associations to achieve a re
peatability which we label truth. The measure of our
5
success in following this truth is called predictability.
Further, Schon takes the view of Heraclitus that
even our observation of new associations is subject to
change; that, in making the necessary adjustment in our
thinking to accommodate the new, we implicitly change the
way we thought about the old association. The new meta
phor places the older one in a broader environmental
scheme than that in which it originally existed.
Therefore, an empirical search for truth boils
down to a search for associations in which;
1. We think we can recognize elements of our past
environments;
2. By applying the "truths" we have accumulated
in those former environments, we can predict with a sub
jective accuracy the behavior of the new association; and
3. By an internal, rational process, we will
transcend the experiential aspects of these associations
to produce an "expected image" of what future environments
will be.
Because of the way in which they bias the direc
tion of our thought — thinking of organization man as a
rational cypher or as a psychological variable can produce
quite different results — the forms in which these meta
phors occur become crucial. Much thinking has been
6
directed toward hedging types of definitions, in which
the metaphor scans several environments, in much the same
way that an author intends his book to be judged in its
entirety rather than by any one page in a chapter. It is
in a sense a Gestalt method of definition.
Thus, J. S. Mill defined analogical reasoning as
occurring when:
Two things resemble each other in one or more
ways: a certain proposition is true of one,
therefore it is true of the other. . . . Every
resemblance which can be shown to exist af
fords ground for expecting an indefinite num
ber of other resemblances.10
The first section of the quotation is deceiving,
in that Mill appears to be happy enough with the ele
mental similarities? but the second part leaves no doubt
that he considers such similarity as useful only insofar
as it forms a link in a much larger metaphoric chain.
This line of thought reappears in Wiener,H who
sees such metaphors as patterns or arrangements, charac
terized by the order of the elements of which they are
made, rather than the intrinsic nature of the elements
themselves. And Kaplan echoes this view, defining iso-
morphs as:
. . . Systems having the same structure, in
the sense that whenever a relation holds be
tween two elements of one system, a corre
sponding relation holds between the corre
sponding elements of the other system. It
7
is important that the systems need not stand
in any causal connection: it is enough that
we can put them into correspondence.12
Thus, Kaplan injects two important developments
into our definition. First, the process of metaphoric
transfer is an arbitrary one performed by us — an asso
ciation, in Locke's sense, rather than some inevitable
and intrinsic truth. Second, the way in which we put
these metaphors in correspondence is an heuristic tech
nique, based on our experimentation to that point.
CHAPTER II
While it is a sign of an illiberal mind to
throw away the fertile and ample ideas of a
Spinoza, a Kant, or a Hegel, because their set
ting is not logically adequate, it is surely a
sign of an undisciplined one to treat their
contributions to culture as confirmations of
premises with which they have no necessary
connection.
— Dewey
So far, we have examined the process as if it were
a scientific one capable of philosophical analysis. We
have not concerned ourselves with the possibility that it
might also be a philosophical problem capable of scien
tific analysis: in other words, that the process may not
be acceptable if it does not conform to certain philo
sophical modes of thinking.
Now, this is antithetical to Durkheim's method.13
For Durkheim:
It is entirely independent of philosophy.
Because sociology had its birth in the
great philosophical system and thus has
been continuously overburdened with it,
it has been successively positivistic,
evolutionary, idealistic, when it should
have been content to be simply sociology.14
But the logic of its inclusion is fairly simple.
Berstein has seen philosophy as an attempt to explain
8
9
human behavior in logical terms. Durkheim, with his be
lief in causality, would have to agree with this. For a
theory such as our to be philosophically correct, it must
also be consonant with human behavior.
Further, if truth is indeed a traveler, then it
is not necessary to state our theory in some universal
form. It is sufficient if we can show that it is viable
in some different philosophical environments. But which?
Dewey's classification of empiricism emphasized
three different stages: preoccupation with the environ
ment, preoccupation with the observer, and preoccupation
with the interaction of the two or experimentation. Now,
while it must be emphasized that the implications of the
classification are no longer the same, yet a similar
division has occurred within philosophy on the grounds of
positivism, existentialism, and pragmatism. The validity
of such a division will now be examined.
It has become almost a reflex action for students
of public administration to place positivism and existen
tialism in antipathy, and then examine the merits and
demerits of both (generally, it might be added, to the
detriment of positivism). Rather than concentrate on the
splintering which has occurred in modern philosophy, we
might follow Dewey's advice not to be illiberal, and
10
instead examine the areas where the different viewpoints
still possess common characteristics. Prom these, no
matter how hostile the match, we will attempt to strike
out in some practical direction.
To prove that our present theory is consonant with
logical positivism could put us in a perplexing position;
for such is the drubbing that social scientists have
handed out to the school over recent years, that proving
consonance could result in wholesale condemnation of the
practice of metaphoric transfer.
Why should we wish to show consonance? For sev
eral concrete reasons: because positivism was a philos
ophy held by many scientists of the day; because it can
be held up as a system explaining their behavior; and
because, under this philosophy, these men made significant
empirical and conceptual advances in many fields.
Most arguments ignore these important points.
Indeed, most arguments are based on a "straw man" concept
of positivism, in which the postulates are frozen, while
the environment — including our sphere of knowledge —
is allowed to change. No one today would seriously en
tertain a democratic system based on the Greek model in
which slavery was acceptable; het, we accept democracy as
a desirable process, when it is modified to fit our modern
11
environment. Similarly, it is unlikely that any
nineteenth-century positivists exist among present-day
natural scientists; for not only is the fight to extract
uniformities from Nature a humbling experience, but to
genuinely believe in Heisenberg's Principle, as a suc
cessful scientist of today must, is to guarantee that the
world is indeterminate.
However, the loophole — it is hardly a flaw —
in the Principle of Indeterminacy lies in the the fact
that these ambiguities occur at the very ends of the
spectrum — in mathematical terms, as the problem tends
to infinity. Hence, it is perfectly true that, given the
impossibility of measuring temperature to the last infin
itesimal decimal, we shall never really know at what
temperature water freezes; but within the practical limits
of thermometer manufacture, and our ability to read such
an instrument to an interval of a degree, we can say
that, to all intents and purposes, water on this planet
always freezes at zero degrees centigrade, and we shall
be able to observe this every time we see a thermometer
immersed in freezing water. Thus, in most cases, and in
all that the Positivists were successful, the choice of
environment — the choice of limits — makes indeter
minacy academic.
12
Now, choice of environment can be stated another
way: which is, that we be aware of the environment.
Given this, what elements of positivism seem worth pre
serving? Kaplan sees it as an attempt to distinguish
between:
Scientific statements, and unscientific state
ments, not on the basis that the latter were
false, but on the basis that they said nothing
at all . . . the criterion was a needle with
which to prick the metaphysical pretensions
in science.15
The mere process of separating the scientific
from the unscientific is a barren one, however; the sci
entist's real motivation becomes one of identifying and
clarifying meaning. This clarification within a specified
environment can, it seems, be regarded as another way of
stating the system of metaphoric transfer; and we may
therefore pursue our task along the path of generality
which was so important to the positivists. This is not
to say that we shall accept all positivist premises, how
ever. Again, in Dewey's terms, we will not be so undis
ciplined. Our chief criticism of the positivists will be
discussed at a later time.
For now, it may be noted that, in missing the im
portance of the environment on determinacy, and thus re
jecting positivism on superficial grounds, social scien
tists have tended to repeat some historical mistakes.
13
Consider the following quote from 1955:
We envisage a far-off scientific Utopia in which
we can reduce to comparable dimensions the Oedipus
complex, repression, submissiveness, physiological
traces, acculturation, the pH of the blood, and
every other factor related to behavior.16
In the limit, Comte lives.
Of the two systems, existentialism appears to be
further from positivism. The positivists are "allowed" to
discover universal truth; the existentialist allows no
truth but personal authenticity. Where the positivists
saw structure everywhere, the existentialists considered
the universe a random, contingent place, in which any
absurd thing may happen with equal probability. Engaging
as this may be intellectually, it does not appear to be
very scientific. While Kaplan and Polanyi may feel the
scientific world is arbitrary, yet they do not appear to
doubt that it also contains a high degree of order.
Miller has put this even more forcefully: "Recognition
of analogy or formal identity underlies all generaliza
tion and all science."17
And yet contingency seems to rule out the possi
bility of such formal identity — of structures, of
patterns.
Much of this must be laid at the feet of the
French existentialists with their almost generic love of
paradox. Dewey has unwittingly characterized the
14
existential movement when he dwells upon: "... the
precious contributions made by philosophic systems which
as a whole are impossible."I®
However, as with the positivists, a loophole has
been left. In the positivist case, we could progress by
ignoring the highly unlikely, the faint exception. In
existentialism, we are better off considering only the
exceptions; and since this is a contingent world, every
case is an exception.
How then are we to decide which are important to
us? We appear to do so on the basis of either relevance
or familiarity. Those cases least relevant or familiar
will be relegated to the extreme ends of the spectrum,
just as the least likely cases were in positivism. The
major difference between these two spectra of possibility
is that, for the positivist, the spectrum was chosen by
Nature; for the existentialist, it was partly chosen by
Nature — the world out there — and partly by himself.
Whether they do in fact differ is an interesting, but un-
provable point.
To take this a step further, we can regard the
existential world as a Wilsonian one, in which our im
pression of chaos derives from the fact that we are con
tinually being bombarded by a plethora of choices. Since
15
we are free to choose any path we feel authentically cor
rect for us, it is quite possible that we might choose a
path of logic and analogy. The only thing that matters
in such a case is that we be aware of the existence of
alternative paths, and that we be prepared to abandon the
logical intentionality at any time it ceases to be
authentic.
We must also consider Monod's suggestion^ that
once the initial choice is randomly made (in the sense
that the concatenation of events prompting our decision
is due to chance), then the rules for that particular
system are set? progress along that path has to follow
certain rules, until some other variable intervenes. His
examples are drawn from the formation of the DNA molecule,
and represent the grapplings of a scientist who recog
nizes a world in which order and chaos appear to exist
side by side in mutual contradiction — the classic ex
istential dilemma. It is up to us to decide whether the
leap from "inert" DNA to "active" man is a legitimate
one. In any case, both Wilson and Monod offer encour
aging evidence to an existentialist who wishes to include
metaphoric transfer in his world view; the minimization
of analogy in a phenomenological method can be seen as
due to a too inflexible view of the philosophical pos
sibilities.
16
While it is fairly easy to establish some basic
picture of positivism and existentialism, it is much
harder to do the same for pragmatism. Part of this is,
no doubt, due to the tradition that the pragmatists were
"fuzzy-minded,"^0 that "it is easier to see what Dewey
opposes rather than what he precisely means."21 But much
of it is also due to the fact that, while the first two
provide us with an end, pragmatism presents us with a
means. This cryptic statement will become clearer after
we examine Bernstein's remarks, from which we shall quote
extensively.
It is tempting to draw connections between posi
tivism and pragmatism, and indeed some of the early
pragmatists did lean towards an eventual joining. But
Pierce was clear enough on the point that pragmatism in
cluded not only the real world, but action and thought as
well. Similarly, connections can be made between exis
tentialism and pragmatism, particularly in terms of the
perspectives which both schools have of the real world:
by which we mean experience for Pierce, and human exist
ence for the existentialists. Indeed, all these disci
plines saw Nature as coercive; but only the latter two
saw the other life nature of experience. If it is co
ercive, then so are we; it reacts as we do. Pierce's
17
Secondness, the name he gave to this inanimate beast,
joins our knowledge and our experience like Siamese twins.
But this is the point at which differences begin
to appear. And here we must defer to Bernstein:
Pierce directly attacks what has been called
the "foundation metaphor" of knowledge and the
"spectator" view of the knower. The conception
of knowledge that Pierce criticizes as mistaken
is one that claims that knowledge does — indeed
must have a basic fixed foundation. . . . From
the perspective of this paradigm the task of the
philosopher is to discover just what this foun
dation is or ought to be, and then show pre
cisely how more complex knowledge rests on this
foundation. If we can know what this foundation
is . . . then we will be in a position to "legit
imize" our knowledge claims. . . .
Pierce . . . argues that the quest for such
epistemological foundation is misguided. Knowl
edge and inquiry neither have nor need such a
foundation. It is certainly true that in any
inquiry, there are starting points, procedures,
methods, rules, etc., that are taken -as fixed
and unquestioned. . . . But it does not follow
that there are absolute starting points. . . .
The alternative paradigm of inquiry or knowl
edge that Pierce begins to develop . . . is
. . . a self-corrective process which has no
absolute beginning or end points and in which
any claim is subject to further rational crit
icism. . . . Our claims to knowledge are
legitimized not by their origins . . . but
rather by the norms and rules of inquiry itself.
These very norms, rules and standards are them
selves open to rational criticism. The falli
bility of knowledge is not a sign of its defi
ciency but rather an essential characteristic
of knowledge, for every knowledge claim is a
part of a system of signs that is open to
further interpretation and has consequences that
are to be publicly tested and confirmed. In the
18
continuous process of inquiry we may be called
upon to revise our knowledge claims, no matter
how certain and indubitable they may appear.
. . . The very meaning of our concepts depends
on the role that they play in a social context
of rules and norms.
The real, then, is that which, sooner or
later, information and reasoning would finally '
result in. . . ,22
As Bernstein says, with remarkable understatement:
The shift from the foundation paradigm to that
of inquiry as a continuous self-corrective
process requires us to rethink almost every
fundamental issue in philosophy.23
The essential opposition of the pragmatist phi
losophy thus lies in its refusal to accept a preconceived
universe. Classical philosophies — and both positivism
and existentialism fall into this category — assume two
things. First, the individual will follow some single
and consistent form of logic. Second, the environment
must also be considered consistent. Thus, existentially,
an enlightened Joe Bftsplk is continually rained on by
the absurdities of fate; while positivistically, rational
man strides out into a lawful universe.
Pragmatism makes no such grandiose claims. It is
only interested in the difference it makes to us if the
incoming information is true. The modification of the
isomorph is more important than nature of the isomorph,
19
since that nature is an arbitrary image picked out by us
at an arbitrary time.
We must, however, beware of falling into the trap
of seeking one "best" philosophy out of these three. In
stead, we must look for some reasonable way of reconciling
them. This becomes clear if we recognize the peculiarly
static way in which classical philosophy views the world.
If, instead, we view the individual as moving through
time to meet his crises, or his environment, a different
schema appears.
Let us consider, as a model, Harris' popular work
on transactional analysis.24 jt is, perhaps, unsatis
factory in that the contrast between Freudian preconcep
tions and Adlerian methods is never fully resolved; but
the assumption of Freudian Parents and Children may not
be necessary.
If we borrow Erickson's view of recurring life
crises, then the child is no longer a child, but merely
the first observer of a new situation. Because the situa
tion is so new, no logic can be extracted from it, and it
therefore appears as contingent in an existential sense.
The Parent, on the other hand, as the archive of
past experience, tries to apply past analogies to the
new situation. Since this is often unsuccessful, the
20
Parent is represented by rigid, "positivist" responses.
With both Child and Parent rendered inoperative,
the individual goes into a search mode in which correct
— coping — solutions are found heuristically. Deci
sions, and behavior, are carried out on a trial and error
basis combining feedback from the Parent in the-of
analogies, and from the Child in the form of perceptions.
This indecisive, pragmatic mode is the Adult. Because of
its nature, we can see why pragmatism seems so indefin
able, when compared to the more fixed existential and
positivist sectors.
This analysis provides a complex picture in which
the three philosophies interplay over a period of time;
the falsity of trying to contrast the various modes
should be evident. Kaplan,2^ for instance, has contrasted
positivism and existentialism on the grounds that the
former seeks universal truths, while existentialism seeks
only individual truths. Our model indicates the fruit
lessness of the technique, unless time, and the nature of
the individual/environment balance are taken into account.
The two philosophies, far from being alternatives,
are merely opposite views of the one conceptual picture,
both presupposing a universe and devising means to deal
with it: one from the environmental side, and one from
the individual.
21
Further, our analysis has attempted to show that
the two are merely information sources, one from the pre
viously perceived universe, and one from the newly per
ceived, which we join by a third method, the pragmatic
to achieve a picture of the presently perceived. In a
peculiar fashion, the phenomenological approach inverts
time by allowing us to transfer metaphors of the future,
newly perceived but not fully understood world, back into
our still forming present world.
Later, we shall show that this interplay was not
always necessary: as long as closed systems successfully
described what happened in our experiments, there was no
need to look further. With our striking out into the
open, uncontrolled experiment, however, recognition of
the place of the three techniques becomes essential.
Further, the recognition of the characteristics of the
pragmatic approach must also lead to some fundamental
changes in our expectations of desirable research design.
CHAPTER III
Historic empiricism has been empirical in a
technical and controversial sense. It has said,
Lord, Lord, Experience, Experience; but in prac
tice it has served ideas forced into experience,
not gathered from it.
The very point of experience, so to say, is
that it doesn't occur in a vacuum.
— Dewey
In considering the nature and historical evolution
of our present research analogies, we must first establish
their level of conceptualization.
C o l e m a n 2 6 has indicated that mathematical abstrac
tions occur on four levels. First, they may be a simple
description (the percentage of people favoring a candidate
in a poll), a proportion of rate at which someone does
something (Durkheim's suicide rates), or as an indication
of structure (graph theory).
On the second level, measurements begin to order
quantities, as in Coombs' unfolding technique.^7 it dif
fers from the first in indirectly measuring some proposed
state; the first, clearly, measured observable phenomena.
The majority of organization research in public adminis
tration can be placed in these two levels.
22
23
The third level considers the possibility that
these quantities can be connected in some systematic way,
which may be called an empirical generalization. Thus,
Stephan and Mischler^ have produced an exponential equa
tion which, they feel, approximately relates the partici
pative rank of a group member to the frequency of his
participation. The method thus tries to produce a func
tion in the mathematical sense. In Simon's words: "the
concepts to be defined all refer to a model — a system
of equations."30
Finally, the fourth level consists of the expli
cation of the above generalizations, and prediction from
them. These predictions take the form: "Given A, then
if B is true, C will occur." (Note that A thus limits
the environment in which B and C can be causally related.)
Now, these theories can occur in two ways. Either
we observe phenomenon B, and derive theory A to explain
it — what Coleman calls "explanatory theory"? or we can
devise a theory containing certain postulates, and attempt
to test the postulates — this he calls "synthetic
theory." Within the social sciences, psychology has been
concerned with explanatory theory, while economics has
concentrated on synthetic theory. Coleman points out that
reliance on one or the other type depends to a great ex-
24
tent on the subject matter, and its amenability to study.
Two points should be evident from this. First,
the synthetic type is prone to normative generalization;
the theory may be as much a product of the theory-maker's
background as it is of the subject matter. Second, Cole
man has simply stated the problems inherent in our philo
sophical discussion of the previous chapter, in which
explanatory theory represents information going in the
environment-individual direction, while synthetic theory
represents the same interaction "outward bound." To say
that theories should be one or the other, then, is erro
neous; in pragmatic terms, both must occur for the theory
to have validity. The implication — that, if this is
really the method by which psychology or economics theo
rists work, then their theorizing is incomplete — is
recognized and grasped.
As a final note, we should state that isomorphisms
will only occur in the third and fourth levels. Coleman
believes that we cannot arrive unannounced at these
stages; that there must be prior definitions of con
stants, variables, etc., which go to make up the func
tions. Isomorphisms thus imply levels one and two; the
reverse is not the case. If we have read Dewey cor
rectly, however, we must regard this tendency to reduc-
25
tionism with extreme caution.
Further, in his lucid analysis of Durkheim's
writings, Bierstedt has pointed out that one of the few
things which joined Durkheim and Weber was the concept
that it is impossible to explain a variable by a constant.
In other words:
Psychological factors are too constant in
their effects to be able to explain the
particular variations that characterize
social institutions.30
Or, in Durkheim's own words:
The determining cause of a social fact should
be sought among the social facts preceding it
and not among the states of the individual
consciousness.31
Durkheim was aware, of course, that social facts
were produced by action on psychological factors. He
felt that psychological training was more valuable to a
sociologist than the biological training pursued by the
social Darwinists; but he was convinced that such training
would only be useful if, in Kantian fashion, the sociol
ogist absorbed psychology, and then abandoned it — went
beyond it.
He must abandon psychology as the center of
his operations, as the point of departure
for his excursions into the sociological
world to which they must always r e t u r n .32
Both Durkheim and Dewey would therefore lead us
to expect that psychological reductionism would be of
26
limited use in the complexities of public administration.
If our isomorphs are to include psychological elements,
then they must be at a relatively constant level. In a
theory of a model of man, for instance, whether he be
rational, risk-taking, or irrational, a factor such as
parental influence, or foetal ingestion, would have to
have an effect so minor — we are reminded of Weber's
ordered causations — or so far removed in time and cir
cumstance, as to be neglected in a general theory.
Now, reductionism implies two related things:
that it is possible to "freeze" past actions in such a
way that their origin becomes definable; and that this
freezing can make the definition determinate. By this
we mean that reductionism in its ideal state requires a
static model of limited probabilities. We must therefore
ask ourselves if our isomorphisms tend to have these
characteristics? A short digression into history may
prove illuminating.
It seems useful to say that the modern charac
teristics of these two properties begin to arise around
1600.
There is, for a start, the appearance of Bacon
(1561-1626) as the great scientific thinker of his day.
Opposed as he was to scientific analysis as it had been
27
developed, first by Aristotle, and later by Aquinas,
Bacon's path could almost be described as predestined.
The intellectual clash is standard fare in any course on
philosophy, and it is not worth our while to go deeply
into it. Suffice it to say that Aristotle's error, as
far as scientific investigation was concerned, was to
rely too heavily on the validity of the syllogism, without
going into detail as to how the original premise might be
established as a legitimate generalization.
In opposition, Bacon codified the inductive
process, which, in Broad's words, was:
A method by which we could slowly and cau
tiously rise from observed facts to wider
and deeper generalizations, testing every
such generalization at each stage by delib
erately looking out for possible exceptions
to it if we actually found such exceptions.33
The similarity between this passage and the model
we described in the last chapter is obvious. But system
atic as this was, it was only a matter of time before the
inferences behind such a statement were subtly altered to
satisfy intellects less gifted than Bacon's. Broad's
following paragraphs represent a model of this destruc
tive distortion:
Perhaps his greatest service here was to show
the importance of testing every generalization
by devising and performing experiments which
would refute it if the result turned out in a
certain way, and would confirm it if the re
sult turned out in a certain other way.34
28
While superficially this seems the same, in fact,
there are important differences. The first statement
deals with a pragmatist universe in which we gradually
push out into the unknown. The second constructs a theo
retical universe first, and then relates the inferences
of this back to the data obtained in the experiment — a
positivist/existentialist technique.
The reason why lesser intellects found the latter
easier to deal with lies in the fact that apparently un
ambiguous answers are obtained. In the former technique,
we are never totally sure that we are on the right track;
it simply seems a better bet than other answers. We are
then genuine experimenters, betting on the probability of
being right.
To put this another way, the followers of Bacon
attempted to separate the scientific method from the
nature of the experiment; so that, no matter how chancy,
how fragmentary the latter might be, the former would
stand, logical, inviolate. The method will tell us how
close the answer is to being right or wrong. Bacon's
implication, on the contrary, was that we also needed
something to tell us if the method was close to being
right or wrong; and that indicator was the return to the
experiment itself for further verification. In spite of
29
this, the standard research design became:
1. Construct a null hypothesis.
2. Construct an ideal environment which contains the
basic elements of the hypothesis.
3. Inject a variable.
4. Record the results.
5. Accept or reject the null hypothesis.
But we are jumping ahead of ourselves; the nature
of this design has to be explored more fully. The split
ting of method from experiment contains within it horren
dous problems, if the experiment concerns probabilities
of occurrence; for we are then faced with a determinate
method telling us whether a probabilistic method is cor
rect or not. The elaborately artificial statistical
techniques devised in modern times by Fisher, et al., are
responses to this problem.
But there are historical, as well as intellec
tual, roots here. There is, for instance, the rise of
the Age of Reason, which was so complementary to the dis
torted scientific method that it all but wiped out
previous thought on the subject. It is, therefore, some
what surprising to realize that, just as present day
existentialists are intrigued by the pervasiveness of
chaos within our supposedly organized society, there were
30
in the seventeenth and eighteenth centuries scientists
and philosophers (perhaps they were one and the same) who
were equally intrigued by the regularities they found in
chance. Specifically, they were gamesters — experi
menters. We have, for one, Condorcet, who, in speaking
of judgment, said:
These applications have taught us to ascertain
the various degrees of certainty that we can
hope to attain: the probability according to
which we can adopt an opinion while making it
the basis of our thinking without violating
the rules of r e a s o n .35
A viewpoint which hardly supports our ideas about clas
sical thinkers. The fact that Condorcet had, through
necessity, been forced to teach his rules of chance to
young nobles as an aid to their .gambling, and had only
written them down to prevent plagiarism by others even
more "scurrilous" than himself, hardly helps his standing.
So low was the regard for the field, that John Stuart
Mill called it a "scandal of mathematics,"36 and Laplace
rejected Condorcet's work in favor of more deterministic
models. (It was of little consolation that Poisson, in
his turn, rejected Laplace's w o r k .37)
Moreover, through its industrial (ergo, mechan
ical) aspects, the Age of Reason linked this determinism
with another feature of classic design, the isolated
environment. Later, we will attempt to show that this
31
environment also came to imply a static model; but for
now, we need merely indicate that, by a process peculiarly
economic, the development of these two concepts produced
an industrial success which legitimized their raison
d'etre.
For the seventeenth century also marked the devel
opment of the first efficient scientific machines.
Wheels, mills, and cutting tools have, of course, been
with us since the time of early many but the precise manu
facture of instruments for the abstract purposes of
scientific research only come into its own after the
Middle Ages. The compass is the exception which proves
the rule; for though, as far as Western Europe was con
cerned, Gilbert had been using navigational instruments
in the late sixteenth century, they failed to provide the
stimulus necessary for prolific invention, because the
principles of magnetism were not fully understood.
The air pump, to the contrary, played a crucial
role in history. Spurred on by such sixteenth century
industrial development as mining, men found the economic
incentives necessary to visualize the full possibilities
of the instrument. Von Guericke was one of the first to
consider using the pump to evacuate air instead of water
(1635-1654?), but his apparatus was crude and its appli
32
cation limited. Boyle's air pump (1658-1679) was con
ceived with many more uses in mind. Certainly it evacu
ated air from a container, but more than that, objects
could be placed within the container, and the effects of
various substances introduced in the presence or absence
of air could be determined. The rationale for scientific
experiment became routinized, and two important criteria
became indispensable.
First, the container, by being isolated from the
accidents of the outside environment, came to represent an
idealized state, or universe. Second, the injection of
variables in a sequential manner laid down the visual
analogues necessary to the object/cause/effect chain
which became the mainstay of the scientific method.
Further, the ability of the container to store over time
led to consideration of the possibility of "freezing"
time and its events, so that they could be analyzed. The
static model became a reality; and because it was easier
to attain and control, it came to represent reality more
than reality did itself. The successes of the scientific
revolution — Lavoisier's isolation of oxygen is the
prime example — all bear this stamp of isolation, ideal
ization, experiment and analysis. And their spectacular
achievements left no doubt as to their efficacy in the
33
modern world.
Curiously, Newton's laws of motion retain the
same fiction: the usual Newtonian problem begins with
something like: "Given a frictionless universe. ..."
In terms of the environment, then, the method implies
that the initial state is always known (because it has
been consciously manufactured), and the final state can
always be found through simple observation, within the
limits of the framework imposed by the instruments. The
circle is thus always completed: there is always an end
to the experiment.
It was not until the first experiments on radio
activity that the possibility of an alternative way of
investigating was examined. The late nineteenth century
experiments of Becquerel, Curie, etc., were influenced by
the fact that the initial results had been obtained by
accident — photographic plates in a drawer were affected
by radioactivity — where there was no question of pro
viding an isolated environment. Indeed, if the environ
ment were insulated, no effects were visible. Since
there appeared to be no metaphors to explain these cir
cumstances, the investigative procedure followed was pure
hit-and-miss. Expose a plate to the air, and see what
happens. Make a guess about events, and expose another
34
plate. Both method and analysis become probabilistic,
heavily dependent for verification on reality.
This emphasis on probability as an explanation of
process, rather than as a technique of judgment, has been,
until recent years, the major difference between experi
mentation in the social and natural sciences. Why did not
the same transition occur in the social sciences? Why,
for so many years, were they forced to carry on with the
old air-pump model?
The answer can only be guessed at but it must in
clude the relative novelty of social sciences. The first
great crest of thought in sociology, the time of Durkheim,
Toennies, etc., coincided with the routinization of what
Bacon and Boyle had been up to; it did not coincide with
the inception of the techniques but with the rationaliza
tion of why they had been so successful.
The work of the early physicists did not become
well known until the early twentieth century at a time
when sociology was attempting to prove itself independent
of the natural sciences. By the time interest in possible
relationships was reborn — probably with Lewin's force
field theories of the Thirties — the mathematics and
philosophy of nuclear and chemical physics had reached a
complexity beyond the grasp of the average sociologist.
35
The lessons remained unlearned.
What were the views of sociologists in this form
ative period? Russett-^S has traced developments under
Comte, Ward, and Spencer, and while her approach is pre
dominantly oriented toward equilibrium, we can still
readily perceive that our final point, the contrast of
static and dynamic models, lay near the heart of their
differences. She notes, for instance, that even amid the
ranks of the positivists there was dissension:
One society could never find equal room for
both Comte and Spencer, since their ultimate
values were so hopelessly at odds. Comte
lauded order and collectivity; Spencer, lib
erty and the individual.39
With such a slogan, it is little wonder that
Spencer was the one who most influenced American socio
logical thought.
However, all three espoused the dynamic point of
view. The interesting aspect is how differently each saw
it. Thus, for Ward:
Dynamic action is progressive, and, instead
of leaving the world in the same condition
as before, leaves it in a changed, i.e., in
an improved condition.40
For Spencer:
Social statics may be aptly divided . . .
into statics and dynamics; the first treat-
ings of the equilibrium of the perfect so
ciety, the second of the forces by which it
is advanced toward perfection.41
36
And for Comte, social statics investigates:
The laws of action and reaction of the dif
ferent parts of the social system — apart,
for the occasion, from the fundamental move
ment which is always gradually modifying them.4^
A statement which both Russett and this author see as
interposing short-term human interaction on the larger
social structure. Dynamics, for Comte, was the examina
tion of the same interrelated phenomena when seen over
time: "a kind of anatomical cross-section of the pro
gressive development of mankind."43
The value orientation of the first two writers
stands out. While Comte was hardly the ideal neutral
theorist, of the three he seems the one most motivated
toward producing a method separable from the ends to
which it is employed. (It is important not to confuse
this concept with the blending of theory and experimen
tation that we have considered desirable in previous
chapters.) It is true that Ward, in later years, tended
more toward a Comteian view, but Comte was on the verge
of decline by that time, and Ward's conversion was inef
fectual. In any case, the seeds of normativeness remain:
Progress consists in setting up dynamic activ
ities in the social structures themselves. A
structure represents a state of equilibrium but
it is never a perfect equilibrium, and the con
version of the partial equilibrium into a moving
equilibrium, provided it moves in the right
direction, is social progress.44
37
Spencer, on the other hand, was in no doubt.
Progress resulted in: "the greatest perfection and the
most complete happiness."45
Believer in laissez-faire that he was, he saw
happiness achieved largely by doing nothing. By some
sleight of hand, dynamism became a static affair. Comte,
on the other hand, considered progress as a drive toward
harmonious cooperation; and this required organic inter
vention in response to a change in the environment.
Now, it is true that, if we view these three in
the rough, so to speak, there is little to choose between
them. All view dynamism as progress toward a static con
dition of perfection or equilibrium. All sense the
existence of the evacuated container at the end of the
chemical reaction. All are closed systems thinkers.
But given the intervention of some later re
searcher with less subjective views, Comte could have
been adapted to fit modern theories of research; after
all, much of systems theory is an offshoot of his orga-
nismic ideas. In the meantime, however, Darwinism with
its emphasis on conflict theories of society, had risen
to shake the sociological world, and there was no longer
room for Comteian cooperation. For Spencer, fortune
smiled, in Russett's words: "Spencer's social thought
38
. . . was itself a form of Social Darwinism."46
And if there is any doubt of this, consider
Russett's own opinion of Comte;
His attraction to organic social theories,
which stress cooperation rather than con
flict, interdependency rather than inde
pendence, the collectivity rather than the
individual.47
All of these things she considers "illiberal" and "author
itarian." One cannot help wondering how Ms. Russett views
the presently evolving world of industrial participation.
In any case, Comte's dynamic method was relegated
to the back row, while Spencer's ideal universe became —
ideal. Fashion, and good timing, prevailed.
The Positivists represent for us the last linger-
ings of social thought within the world of the natural
sciences. Darwinism was possessed of sufficient identity
for sociologists to declare their conceptual independence,
and to attack the problems of social science as separate
entities.
Praiseworthy as this impetus was, however, it
also left the infant science with a crippling legacy —
the archetypal presumption of a science itself already
undergoing rapid change from within. Social Darwinism,
and its multiplying siblings, took for granted a dynamic
model of the world which was, paradoxically, static.
39
All interaction was for nothing else but eventual contain
ment within a closed, ideal, container. As the sociolo
gists of the time might have said, in the Beginning, there;
is the End. Functionalism, hypothesis testing, the models
of Man — are but the inheritors of an ancient tradition.
CHAPTER IV
Empirically, then, active bonds of conti
nuities of all kinds, together with static
discontinuities, characterize existence. To
deny this qualitative heterogeneity is to re
duce the struggles and difficulties of life
. . . to illusion: to . . . the "subjective."
— Dewey
It would be possible, though pedestrian, to trace
the implications of the last chapter through the develop
ments of the last seventy years of public administrative
thought. Much more vivid would be to encapsulate the
struggle of past and present within the work of one many
and this we can do. For no man has loomed larger in his
contributions to thought and orientation than Max Weber.
Such is his standing that he might well be considered the
bureaucrat's Marx, the working man's Nietzshe; he is
quite possibly acquiring, with each passing year, the
quality that preoccupied him so much: charisma. The
reader need only go over the traditional "great" works of
organization theory to realize how the ideas of Weber
have touched them ally how many have touched the forelock
to the man.
40
41
Yet, when we contemplate the dubious testaments
handed down by former prophets, such as Comte and Spencer,
some uneasiness must appear. Did Max do the right thing
by us? Let us try to decide.
It is fitting that, in defining the historical
disciplines, Weber felt compelled to stress the plurality
of causal factors. For Weber's work is itself the result
of several disparate events of the nineteenth century.
As a friend of Weber's, Honigsheim,4® traces his method
ology in respectable linear fashion through Dilthey,
Rickert, etc. But for us Weber fits into a larger field
of influence.
There is, for a start, Darwin's voyages after
1831; voyages which certainly led to a dynamic view of
evolution, but which, through the doctrine of survival of
the fittest, seemed to emphasize the circular arguments
of Functionalism. In any struggle, the fittest will sur
vive; therefore those we see around us are the fittest.
It is a dynamism which relies on the closed system to
defend the status quo. Elsewhere, Marx was to adopt the
schema after 1844 for his struggle of the masses, who
were to ultimately triumph by virtue of their superior
fitness.
42
While this view of society as movement was being
pursued, the contrary view of sociology, as the science
of the classification of the forms of social relation
ships, was being developed by Simmel, and later by
Toennies. The monumental Gemeinschaft und Gesellschaft
was published in 1887, the same year in which Weber began
his doctoral dissertation. Thus, while Toennies later
acknowledged his debt to Weber for the classification of
Wesenwille and Kurwille, this was more a case of youth
outstripping age (Weber was nine years younger than
Toennies), than it was of one man's work influencing an
other s in sequential fashion.
Thus, the thrust of English science was toward a
peculariarly static form of dynamism; that of German sci
ence was toward purely static classification; and Weber,
for better or worse, was immersed in both.
Yet he did attempt to escape. His dissertation
was an attempt to examine the legal principles behind
risk-sharing in enterprises — hardly a classical trea
tise for a sociologist. But then Weber at this time was
considering himself an economist first, a sociologist
last. And indeed, his later theories on bureaucracy
developed, not from a desire to further Simmel's work,
but to counteract a third major influence in his life —
the doctrines of Marx.49 To describe this solely as
43
political or economic motivation would be to insult Weber
with simplification; but his opposition to Marx was ele
mental. Yet, paradoxically, Weber did not wish to throw
Marx out the window and begin afresh. His plan, instead,
was to improve on him in such a way as to make him
obsolescent.
For underlying this is Weber's ambiguous attitude
to the fourth influence, Hegel. Bernstein50 has pointed
out that Marx was a determined, even a deterministic,
Hegelian, and the process of change, through the struggle
of thesis and antithesis, is prominent in his work.
Weber, on the other hand, once called Honigsheim "worse
than Hegel":
He agreed completely with someone who said,
"Whoever uses the word synthesis today,
without making the meaning explicit, de
serves to be boxed on the e a r s . "51
There was in Weber a deep appreciation and caution
toward the dynamic of Hegelianism. Perhaps — it can
hardly be proved — Weber felt the multiplicity of actions
that take place within any one Hegelian struggle, and con
sidered Marx's borrowings dangerously simplistic. Cer
tainly his Protestant E t h i c 52 Was an attempt to point out
that economics might not be the only motivating force?
that religion might have its say too. So we are left
with the curious situation of Weber drawing us away from
44
Marx by the use of Marxist (i.e., Hegelian) techniques.
But the times did not permit Weber to be simply
an economist; he was, by academic definition, also a
sociologist, and it was part of his immense ability that
he was able to incorporate the two quite different modes
of attack, of Hegel and Simmel, into one method. He was
inspired to produce a dynamic theory of authority — of
the traditional/legal/charismatic process — which at the
same time could be classified into three "ideal" types.
Now this theory is what Parsons has called a
mosaic theory; what we might today call a matrix theory,
similar to Etzioni1s classification of types of compli
ance. It is comparative and static, in that the final
states are emphasized at the expense of the transitional
states: the moments of change between the periods of
ideal charisma, tradition, etc. Thus, while Weber's
theory gives the appearance of dynamism (through the
process of routinization), in fact it is only an examina
tion of discreet points along a spectrum, which is itself
not analyzed. Moreover, the ideal nature of the types
implies a deterministic view, in which the end product
is known, and whatever occurs is seen merely as an incom
plete representation of the postulated universe.
45
This statement must inevitably draw cries of pro
test. Freund states in no uncertain terms that:
[Weber] nonetheless starts from the basic assump
tion that empirical reality is extensively and
intensively infinite. . . . We can never come to
the end of our exploration of events and of their
variations in space and time or act on them all
. . . that it is impossible to describe even the
smallest segment of reality completely or to take
into account all the data. . . . Knowledge and
action are never completed, all knowledge leads
to further knowledge and every action to other
actions.53
And more importantly:
Weber was opposed to all systems — whether
classificatory, dialectical or other —
which, after constructing as close a network
of concepts as they are able, claim they can
deduce reality from it.54
The first passage is as close a description of
pragmatism as we are likely to get outside of Dewey. The
second is equally desirable (for us) — except that, if
we are to judge from history, Weber was not above carrying
out just such a classification and letting others draw
the conclusions. As Mannheim puts it:
Max Weber's actual historical analyses do not
always correspond to his theoretical precepts.
In his theoretical writings, he insists on a
causal explanation; in his historical work,
he very often proceeds according to the
"documentary" method.55
What is wrong? We have noted the stress Weber
was under in trying to reconcile dynamic and static proc
esses. He even dabbled with the concept of entropy, but
46
did so only from concern for it as an end product: the
pull of the ideal type was too strong.
Of course, his methodology left him little choice.
In classifying — again — the sciences by a nomothetic/
idiographic dichotomy, he segregated all regularities to
the natural sciences, while "the ideographic sciences are
concerned with knowledge of unique events."56
By definition, then, the natural sciences alone
can have periodicity and continuity, while the social
sciences can only be taken in discreet — static — units.
Now, we have shown that Weber also did not mean this.
Within the ideographic sciences, there is the pluralism
of causal factors. It is only at a nexus of these factors
that an event occurs. Unable to classify all the proba
bilities, Weber throws up his hands and describes the
complex predictability as unique. But the two words are
not interchangeable; a multiplicity of dynamics does not
produce a static result, at least, not in the social sci
ences. Indeed, the very fact that he should attempt to
classify authority indicates that multiple factors may
coalesce often enough for the authority phenomenon to be
noted.
So Weber's apparent clarity dissolves into an
unresolved tension. Its effects on us have been unfortu
nate; for we have come to believe that classificatory
47
techniques not only describe dynamic processes adequately,
but also minimize the necessity of analyzing the dynamic
process (as opposed to the end product). Further, by con
centrating on the end products, we bias our work in favor
of particularist, reductionist concepts which, in their
turn, infer a predominantly static, closed universe.
But this may begin to sound polemical. Did Weber
really have any choice; by inference, do we? Were there
alternatives?
At the time of Wilson's statement of the issues in
public administration in 1887, political science was di
viding itself into two areas — that of economics/poli-
tics, and that of public administration. It seems fair to
say that the former held sway in Europe, where administra
tion became a mere application of higher legal forms. In
the United States, the applicability of Wilson's thesis
was appreciated, so that by 1904, when Weber first visited
the country, public administration was a burgeoning dis
cipline.
The significance of this lies in the fact that one
of the most useful alternatives to Weber's method turned
out to be an economic one. Schumpeter's theory of re
newal was one which posited the rise of innovators during
periods of so-called overlapping geists as an explanation
48
of the ebb and flow of society. MacDonald has convinc
ingly demonstrated that what is at stake in any comparison
between the two men is the distinction between statics
and dynamics.
While both start from a similar viewpoint
(Weber's traditional capitalism versus Schumpeter's cir
cular flow), Schumpeter's development shows several
significant deviations. First, the Calvinist innovator,
whom Weber saw as an ideal type, is replaced by a super
normal entrepreneur. A unique person, in other words,
becomes a merely special person, who potentially exists
in any society.
Next, Weber saw the rise of capitalism as a state
which overcomes a previous state, feudalism. Schumpeter,
on the other hand, saw capitalism as a process which pro
ceeded gradually, admittedly by discontinuous steps, with
innovative cycles alternated with moments of equilibrium.
Finally, Weber saw the animus for this change in
the Protestant ethic — a particularist concept. For
Schumpeter, there was not one geist, but many, which over
lapped, and whose differing values provided the impetus
for the innovator's advance.
The interesting thing about these differing views
is that they did not start from differing paradigms?
49
Schumpeter's work derives directly from pages 27-29 of
Weber1s Die Protestantische Ethik und der Geist des Kap-
italismus of 1904. The static nature of Weber's work is
the product of his own conceptualization.
Yet, if Schumpeter's theory was better adapted to
a dynamic view of world progress, why was it not accepted
as such? Weber's view of the world after the waning of
charisma was quite deterministic, compared to Schumpeter's
world made rich in potential by virtue of the creative
human impulse. Weber offered no satisfactory escape from
routinization, other than the mystical uprising of another
charismatic figure. Are we to found a new religion every
few hundred years? Why?
But Weber's theories were accepted, and Schum
peter's remained solely for economists. Several reasons
are apparent. First, MacDonald points out that there is
no inherent superiority in either a static or dynamic
model.
From Schumpeter's standpoint, several of Weber's
concepts are seen to be special cases of a single
conceptual process, and at least one of Weber's
problems is seen to be a spurious one, arising
from the inadequacies of "ideal-type" tools. On
the other hand, Weber's distinction between class
and status, which vanishes in Schumpeter's proc
ess model, is of vital importance in analyzing
social systems which may properly be regarded as
stable, and a process model tends to gloss over
such distinctions. . . . The comparison illus
trates both the convenience of static models, in
50
making it possible for certain variables to be
dropped entirely as well as the dangers in
volved if the system is actually in a process
of change.57
But the times were against Schumpeter. War was
still regarded as a regrettable deviation, empire was a
stable, honest vocation for monarchs. Process was simply
not in the image the world had of itself.
Wbrse, Schumpeter himself did not see his method
as social theory. He did not consider it sociological,
because, unlike Weber, he was an economist first and last:
"his reputation as a jealous guardian of the integrity of
economics, remain[ed] unchallenged."58
Indeed, he never apparently regarded it as a
theory, for he never expounded it systematically. It was
simply a technique for solving economic problems.
Weber's theory was in the mainstream of "modern"
German social theory, required no great jump from the
ideas of Simmel and Toennies, and, even more signifi
cantly, in his descriptions of the victory of capitalism,
he was translating for the New World the progressive opin
ions of Comte and Spencer. The American industrial ethic,
the J. P. Morgans, found their justification: it was in
stars all along. For sociologists, the combination was
51
formidable. For students of public administration, given
their first systematic exposition of bureaucracy since
the Cameralists, it was well-nigh irresistible.
CHAPTER V
The human being has upon his hands the
problem of responding to what is going on
around him so that these changes will take
one turn rather than another, namely, that
required by its own further functioning.
— Dewey
Bloated with background material, we are now in a
position to contrast our research design to the tradi
tional one. The characteristics of the latter have been
taken as:
1. Presumption of an ideal, prior universe.
2. This universe can be considered, for experimental
purposes, closed, so that there is a definable end
to the experiment, and variables can be controlled.
3. These variables can thus be isolated, classified,
and made unidimensional (Reductionism).
4. Examination of the variables is done on the basis of
their descriptive characteristics, rather than on the
processes by which they interact (Static Analysis).
5. The research method is determinate, relying on the
statistical acceptance or rejection of a null
hypothesis for its legitimacy.
52
53
By contrast, the characteristics of our alterna
tive design will be:
1. A non-committal attitude toward the universe.
2. The universe is considered in open interaction with
the experimenter, so that there is no end to the ex
periment, and variables are not controlled.
3. Variables are defined phenomenologically on the basis
of what the universe "considers" relevant. They are
functional, rather than classifiable, and are probably
complex (multi-dimensional).
4. Examination of the variables is done on the basis of
how they interact (Dynamic Analysis).
5. The research method is heuristic, relying for its
legitimacy on application of predictions concerning
variable interaction back to the universe.
Item 4 needs some explanation. We have already
quoted MacDonald's statement to the effect that there is
no inherent superiority in either the static or the dy
namic method. However, since the static case is a special
type of dynamic case, we are in a better "insurance"
position if our results do not turn out the way we expect,
then we would be if we took the traditional, classifica
tory path.
54
It must also be noted that this design is pecul
iarly suitable for the social science case, in that the
universe is considered to act like a human being in many
ways? for the objects examined are not inanimate, but, in
a sense, experimenters in their own right. This, as we
shall see, raises serious doubts about the use of clas
sical statistical techniques for samples of less than
"acceptable" size, where the motivations of the sample do
not obey some constant, economics-type laws.
Finally, the classically-oriented reader had best
be prepared for the fact that the design may leave him
holding onto a bunch of loose ends in a very frustrating
manner. This is intentional.
Because experimentation is so much a part and
parcel of our method, we have decided to illustrate the
design by an analysis of decisionmaking within a public
organization. The theory of statistical decisionmaking
has been exhaustively, if normatively, treated in the
literature, and there will be no attempt to work this
material one more time. The basic assumptions underlying
our ideas of the field can be found in W a l d ,59 Raiffa,60
and particularly Chernoff and M o s e s .61
Given that we are in some public organization
and possess a subject, there are three possible ways we
55
could begin:
1. We could simply take the symbols used in decision
theory, and ask the subject to give numerical values
to them as best he could; or we could, by searching
the files, develop our own numerical equivalents
for these quantities.
2. We could abandon any attempt to fit the theory to the
situation, and simply try to build a model which sat
isfactorily describes the subject's behavior.
3. We can attempt to find concepts which the subject
uses, and construct analogues within the theory in
such a way that the facts are explained satisfac
torily.
The first achieves nothing, being self-fulfilling
as far as prophecies go. The second is far more useful,
in that we are at least simulating actual behavior.
However, unless some theory can be constructed to explain
this model, no further implications can be drawn. The
third looks the least satisfactory, because of its incom
pleteness; if we do not succeed in finding appropriate
analogues, we have no alternative hypothesis to sustain
us.
Yet it is the path we shall follow. We are in
terested in finding symbols in human thinking which have
56
their equivalents in decision theory, but the existence
of these symbols does not thereby legitimize decision
theory in toto. So let us first set up a group of terms
from decision theory which might be useful to us. (Note
that the sequence is not important here; we could equally
investigate the subject first and bring those variables
back to decision theory.)
Simon62 has noted that most decision models in
volve at least the following elements:
1. A set of behavior alternatives, which we shall label
as the set X.
2. The set of behavior alternatives that the subject
actually considers, which is a subset of X, and con
sists of xl, x2, ..., xn.
3. The possible future states of affairs (outcomes of
the X), which we shall call the set W, and which
contains wl, w2, ..., wn.
4. A "payoff" function, which represents the value of
subject places on each decision or its outcome,
depending on the causal relation between the two.
This is V(w), or V(x).
5. Information as to the probability of occurrence of
each outcome w.
57
Normatively, three possible strategies are con
sidered:
a. Minimax Rule. Always assume the worst possible out
come no matter what x is chosen. Then select the
highest payoff V, within that worst outcome, and
consequently choose the x which provides it.
b. Probability Rule. Since item 5 tells us the proba
bility of occurrence of outcomes derived from any
alternative x, we can calculate the expected values
of the payoffs E(V), and select the value which is
greatest.
c. Certainty Rule. Select the alternative x which pro
vides an outcome w with the largest payoff V.
Under this scheme, strategies which minimize the
subset (x) relative to the set X on the basis of time,
convenience, and innovation, are labelled as "satis
ficing." Its opposite — the need to expand alternatives
to seek new solutions — is labelled "search."
The classical objections to the rational choice
model are, first, that a total ordering of all payoffs
is impossible simply because the subject may not know
them all, and second, that the utility function which
calculates payoffs may involve combining concepts which
are insoluble in each other.
58
A final extension of the rational model lies in
Bayesian theory. Here, the decisionmaker is considered
to be an experimenter, who modifies the probabilities of
occurrence of outcomes (W) on the basis of data received
from his experimentation. He is thus a more sensitive
pragmatist than the simplistic decisionmaker outlined
above. He must also use a further set of concepts:
6. External observations Z (zl, z2, ..., zn), which in
dicate to the decisionmaker the probable states of W
before the experiment.
7. The frequency f(z:w) with which the observation z
represents to him a certain state w.
8. The prior probability u of occurrence of a state w.
9. The term u now becomes the a posteriori probability
of occurrence of state w. Bayes Theorem then states:
where
f(z) = uif(z:wi) + U2f(z:w2) + ... + unf(z:wn),
for n states of nature.
59
Our last preliminary step is to decide on the
nature of our sample. Since we have no universe in mind,
a representative sample is irrelevant. The purpose of
the sample is far more important. We are interested in
investigating whether human problem solving might contain
metaphors similar to those in decision theory; in a more
general sense, we are curious to know whether normative
Bayesian theory can legitimately claim that there are
rational people to carry out the necessary techniques.
In such a case, a sample of one human could conceivably
be a good start; for it could tell us at least two things
1. If the theory adequately describes that human's way
of making organizational decisions, then the theory
is at least applicable to some humans.
2. If the theory does not describe what our sample does,
then we may either (a) continue our search for a rele
vant human, or (b) begin a search for an alternate
theory.
It can be seen that sample size has no relevance.
We would need to investigate the entire population to
disprove (1), and (2a). (Remember our model as yet does
not accept the conditions of statistical sampling.)
Alternative (2b), if carried out before the entire popu
lation was exhausted, would merely require the same
60
pattern of investigation to be carried out at a later
date.
In fact, our procedure will be recognized as a
simple sequential analysis in which the probability of our
choosing a relevant human is unknown (the population is
both too large and too volatile). Our stopping rule is
that of Dewey — that the response should be that required
by its own further functioning — which is to say, some
consistent rule of survival.
Moreover, while the procedure may be unusual, it
is by no means rare. Dukes^3 notes at least 246 studies
in the period 1939-1963 (in terms of journal articles) .
A large proportion of these fall, as our study does,
within the region of personality: motivation (7 cases),
perception (25), learning (27), thinking and language
(15), and intelligence (14), are typical areas.
Dukes also outlines the criteria for such studies:
a. The existence of a unique specimen.
b. The existence of complete population generality.
c. The desire to preserve functional unity, or to
dramatize a point.
d. The existence of dissonant character in the findings:
In contrast to its limited usefulness in
establishing generalizations from "posi
tive" evidence, an N of 1 when the evi-
61
dence is "negative," is as useful as an N of
1,000 in rejecting an asserted or assumed
universal relationship.64
e. The existence of a limited opportunity to observe,
either in terms of uniqueness of case, as in (a) ,
geographical situation, or time.
f. The possibility, if the study is carried out, that
questions, defining variables, and indicated ap
proaches may make substantial contributions to the
study of behavior.
g. Similarly, the possibility that such a study might
provide important groundwork for theorists.
In our present cases, criteria (d), (f), and (g),
are all good motivators. For doubters, we can only refer
to Duke once more:
The statistician who fails to see that important
generalizations from research on a single case
can ever be acceptable is on a par with the ex
perimentalist who fails to appreciate the fact
that some problems can never be solved without
resort to numbers.65
First, then, we are hoping to find out if the in
dividual follows decision theory procedures in deciding
a course of action. If not, the theory's universality
will be called into question.
Second, if the theory is satisfactory in some
ways, we can investigate the applicability of concepts
such as "state of nature," to a person who has had no
grounding in decision theory.
Third, if the individual does prove to be
"rational," we hope to be able to describe the strategies
as minimaxing, probability, etc.
Fourth, we must lay some groundwork for incor
porating our results (assuming they are positive) into a
dynamic theory of a higher order.
CHAPTER VI
The organism has to endure, to undergo,
the consequences of its own actions.
— Dewey
The following study complements, for purposes of
comparison, a paper entitled "The Juvenile Probation Sys
tem: Simulation for Research and Decision-Making," by
McEachern, e t _ a l . 6 6 Several points should be made at the
outset: —
1. There will be no attempt to prove that one
approach is right and the other wrong; on the contrary,
it will be apparent that both studies fulfill a useful
need in the search for truth.
2. The study has been chosen because it follows
the Weberian model in two ways. First, it takes the
nomothetic stance of generalizing from an impressive
group of statistics to the particulars of any one proba
tion officer's decision process. In doing so, it assumes
a utility function sufficiently general to encompass all
the sample, and by implying that the model is rational,
suggests that it be used to guide the actions of the in
dividual: "We must move away from simple cause-effect,
63
64
deterministic roles towards concepts of complex systems,
strategies and rational decision processes. "6’ 7
The emphasis is ours. This blanket assumption of
rationality has bo be treated with some caution. In
trying to assess his self-actualizers, even Maslow88 was
forced to conclude that he knew very little about their
distribution in society. His is a case of what Dewey
describes as "traditional accounts," that "have not been
empirical, but have been deductions, from unnamed prem
ises, of what experience must be."69
Of course, this does not put the SIMBAD study (as
the McEachern group's work will be called) automatically
in error. For it to dig back so far in premises would be
to put an intolerable strain on an otherwise tightly dis
ciplined work. It is precisely why research such as ours
is required: to fill in the gaps that a classic work
should not be called upon to do. The question as to
whether man is irrational, as the existentialists are
tempted to believe, or rational, as the economists would
prefer, has implications outside normal statistical
analysis.
Second, the SIMBAD study is a classificatory one,
in which the examination of variables in the utility func
tion drops off in efficacy after about twenty. By appli
65
cation of Bayesian statistics, the authors construct a
probabilistic model which can be used to simulate proba
tion decisionmaking. An end point — the SIMBAD model —
is reached, which is within the conceptual framework
established by the initial variables isolated, and the
reason for their isolation, the Bayesian model. The study
requires no further experimentation to carry it on past
the end point; it is complete in itself.
The Los Angeles Probation Department is a County
organization, under the supervision of the Chief Adminis
trative Officer. Among the list of its duties specified
by the CAO are the investigation and recommendation to
the courts on delinquent and pre-delinquent children, and
the supervision, control, and rehabilitation of such
juveniles.
The Camps' Intensive Aftercare Program is a devel
opment from the Camps and Schools Division of the Proba
tion Department, and is used to reinforce attitudes which
may have been developed within the camp program. The
Deputy Probation Officer (PO) is seen as providing the
necessary treatment patterns, and formulating the neces
sary objectives, that will enable the juvenile to make a
smooth transition back into community life. The PO will
use such techniques as individual counseling, group
66
activities, tutoring programs, and workshops, to achieve
this transition.
Because being sent to camp represents one of the
most severe things the County can do to a delinquent, the
population an Aftercare PO deals with is above average in
lawlessness. While the SIMBAD group dealt with a popula
tion containing roughly 50 percent prior offenders,70 the
Aftercare population has 100 percent. In the words of
the Probation Department:
The boys are "losers" by almost any definition
of the term. . . . Nearly half of these boys
came from families in which at least one other
member had a record of criminality or delin
quency. Seventy-one percent had divorced or
separated parents. Sixty-two percent had no
father, stepfather, or other adult male figure
in the home.
Twenty-eight percent of the boys came from
families receiving public assistance. The
median annual income of the families of all
boys was 23 percent below the countrywide
median. Sixty-six percent of all parents had
less than a high school education. . . .
Twenty-eight percent of the boys were no longer
attending full-time school, even though the
median age of all the boys was only 14 years.
The measured I.Q. of the boys placed 68 percent
of them in the dull normal or lower intelli
gence range . . . by the time these boys were
placed in camp, 86 percent had had prior police
contacts and 84 percent had previously appeared
before the Juvenile Court.71
Thus, while in the SIMBAD case, "improvement is a
. . . realistic measure of the effectiveness of proba
tioners in the sense that a juvenile who has been com-
67
mitting serious offenses before probation will commit less
crimes of a serious nature after treatment, there appears
to be little such justification of optimism in the After
care unit. This has significant effects on the values of
such units.
Instead, a major criterion in the dispensing of
decisions becomes, not improvement, but simply the buying
of time. By keeping the juvenile from more serious pen
alties, including those normally reserved for adults, the
PO buys time for the child. By maintaining a loose super
vision over his activities, the PO buys time for the
community. In the meantime, it is hoped that the child
will pass out of the destructive phases of adolescence
and his peer group pressures, and, by the age of eighteen,
be mature enough to place himself in a perspective against
the norms of society.
Aftercare Units thus spend much of their time
fighting rearguard actions. As a result, the formal
treatment process often takes on the unfortunate appear
ance of a judicial cycle, through which the child will
pass, apparently on an endless belt, several times before
he or she is eventually "retired" because of age. Though
we have no evidence for it, it seems quite likely that
the child must become aware of this, and, further, that
68
it should lead to an even more pessimistic view of his
ultimate destiny. Of course, the proviso must be added
that there are juveniles who escape the system. By defi
nition, however, successful "escapees" are not subjected
to the grinding repetition of public wardship.
The limitations of treatment are directly related
to the limitations of the environment. Pressures from
the police, who have a community to protect, pressures
from relatives, for intensely personal reasons, and pres
sures from the community, all act to constrain the PO*s
alternatives. Above all, there is the ever-present,
crushing weight of juvenile society, which few offenders
are fortunate enough to escape.
The goals of the unit, then, betray a certain am
biguity. Officially, "the reduction of recidivism is the
major objective of the penal system."73 But this seem
ingly straightforward statement can be broken down into
two quite different, and sometimes contrary goals: (1)
the rehabilitation of the individual, and (2) the protec
tion of the community.
Obviously, the achievement of the first should
automatically lead to the second. But attempts to deal
"humanistically" with the juvenile can place the community
in jeopardy if he should take it into his head to sniff
69
glue and walk the streets playing with a pistol. Equally,
"hard-nosed" approaches, such as automatic detention for
any infraction, may keep the juvenile away from the com
munity, but they may also prevent him from ever learning
what socialization is all about.
A PO can thus find this job full of unrewarding
dilemmas. Control from his superior, the Supervising
Deputy Probation Officer, is generally loose. While pro
cedure may vary from unit to unit, an incoming case will
generally be discussed by the PO and the Supervisor, and
a preliminary treatment plan devised. After this point,
the PO functions largely as an autonomous unit, referring
very few actions back to the Supervisor. Of course, ul
timate responsibility rests with the superior, and he may
ask for an update at any time? but delegation is exten
sive. The decision tree (Fig. 6-1), available to a PO
at the time of the survey is on the following page.
A few notes concerning the various decision points
should be valuable. First, the juvenile comes to the at
tention of the PO in one of three ways:
a) Referral.— An outside agency, generally the
police department, may apprehend a juvenile for a minor
infraction, and be in a "zone of indifference" as to what
to do with him? it is possible that the court procedures
No File
Incorrigibility
Petition
Pile
(a2)
Recommend
Detention
(bl)
Petition
70
Pile
(a3)
Hold in
Ab^rajice
No
Recommend
(b2)
Recommend
Release
(b3)
Attend Not Attend
fdl ) .
Judicial Processing Not Guilty
Guilty
Home Home Placement Suitable Stayed Camp
with with Placement Camp
Suitable Relative Order Order
Placement or out of
Order State
(41) (42) (d3) (d4) (45)
(46)
Calif.
Youth
Auth.
(47)
j?IG. 6-1
DECISION TREE
71
may mean more inconvenience than the apprehension is
worth, for instance. In this case, the agency informs the
unit of the infraction, and leaves the decision — to file
a petition or not — up to the PO. To file in such a case
clearly requires some initiative and investigation on the
part of the PO; while not filing falls into the time-
buying category we have described.
b) Incorrigibility petition.— In cases where the
child has committed no specific crime, but is beyond the
control that society demands, a PO may file a petition on
the grounds that the juvenile is incorrigible. The deci
sion is a personal one initiated by the PO, and requires
much commitment to the belief that there is no further
hope: for pressure from relatives, and perhaps a skep
tical judge, can make this a very difficult procedure.
Needless to say, it is rarely used. Not to file in such
a case is not really a decision, since this is the state
in which PO and child normally live; no intervention in
the environment is required.
c) Petition request.— This is sent to the PO by
the police agency in response to arrest of the individual.
Since the arrest puts the child formally into the system,
the PO has only two choices: to file a petition, or to
72
hold it in abeyance. While the second is again a time-
buying maneuver, it is not as attractive as not filing
from a referral, because the police department is now
formally committed to a course of action, and contravening
this lays groundwork for later poor relations between
agencies. Moreover, it goes against the SIMBAD maxim
that the PO is always required to do something about the
problem.
Having filed a petition, the PO has three straight
forward recommendations to make to the court. These are:
to recommend detention, to make no recommendation, or not
to recommend detention. The issue is clear-cut; the
choices are comprehensive and mutually exclusive.
Should the officer feel strongly about urging a
recommendation, he or she may attend the hearing and per
sonally speak to the court. This set of decisions, to
attend or not to attend, is included, but only for the
sake of completeness: for so many variables enter into
the choice that it has proved impossible to stabilize.
The geographical distance to the court, the press of other
commitments, or even something as mundane as vacation, can
affect the choice. In effect, the environment sets the
choice for the PO, and hence it cannot legitimately be
considered a personal choice. Its analysis will not be
73
included in the discussion which follows.
The actual hearing can occur in two forms. If
the child has been kept in the Juvenile Hall since his
arrest, then a detention hearing will be held. If he has
been allowed to return home, he will appear at an arraign
ment hearing. The type of hearing is not important to the
PO's decisions. At this point, however, the child is
either judged not guilty and released, or guilty and de
tained for recommendations on disposition. Given the
latter, the judge will obtain suggestions from the PO re
garding the fate of the offender. At the time of the
survey, the PO had a set of seven choices, but in fact was
having to work from a subset of five. The choices are
listed along a spectrum of severity as it appears to the
child — i.e., in terms of separation from his society.
While they roughly correspond to the seriousness with
which the court views the offender's behavior, this linear
placement of choices is not comprehensive, as we shall see.
a) The child may be placed under home supervi
sion. This is little more than strict family control,
with periodic visits by the PO.
b) Home with a suitable placement order is a
slightly greater threat, in that the option of taking him
out of his home and placing him in a stricter environment
74
is retained by the court, and can be used if he does not
"behave." While this may seem a useful option, in fact
the problems involved — both in terms of waiting for the
child to do something sufficiently bad to warrant return
ing him to the court, and in terms of the bureaucratic
procedure involved in getting him first to court and then
into a suitable placement — makes it a complicated deci
sion with more clear-cut alternatives.
c) Placement with relatives, or placement out of
state, is such an alternative. While the child may view
this with alarm, it does remove the environmental effect,
and may also provide new authority figures to whom the
child can relate. It is largely dependent, of course, on
the availability of people who are personally concerned
with the child, and who are willing to take him in.
Choice, then, is subject to external constraints, and the
decision is not fully in the PO's hands.
d) The suitable placement order requires that the
child be placed in some corrective environment, such as a
home. This is a more formal atmosphere for the child,
and a certain amount of bureaucratic procedure is required
to find a place which is really suitable, but otherwise
the decision seems fairly straightforward. It also allows
the PO a longer time over which to observe and control
75
the child; again, it is a time-buyer.
e) The stayed camp order appears in the larger
set, but is almost never used. Again, it is a threat held
over the child, but, while apparently buying time, it
actually creates so much personal, interpersonal, and
organizational disruption that it appears to have disap
peared from the working list of choices. It is included
for completeness, but will not be used in the analysis.
f) Assigning the child to camp is normally the
most severe punishment the court will consider. In camp
(from which, it may be noted, the child has already come,
apparently with little effect), there will be an obliga
tion to socialize, play games, etc., under the control of
a deputy probation officer. The grounds of such camps
are fenced, but this is more symbolic of the control the
Department has over the life of the offender than any
serious attempt to prevent "breaks."
Since camp is provided for within the probation
system, it is bureaucratically an easy decision to make;
further, it allows the philosophy of the department to be
carried on with some continuity. Its drawbacks are two.
First, the child may view camp as something separate from
real life, something "unnatural" that he is being forced
to do. In such a case, the child will merely wait out
76
the program. (Compare this with the more indefinite
schedule of the suitable placement.) Second, if a mis
take has been made in the assessment of the child, and
he is basically good, then camp may have a deleterious
effect on him.
Of course, the same rationale applies to the bad
child who is allowed to stay home with his mother. Clear-
cut decisions, then, may be organizationally simple, but
may entail large risks if basic character assessment is
faulty. More complex decisions may carry a certain amount
of inbuilt insurance.
g) There is a possible step past camp, and that
is assignment to the California Youth Authority. However,
this is an abnormal step, dependent on the criminal his
tory of the juvenile. It is, therefore, not one of the
usual range of options, and is largely out of the hands
of the PO. Again, we include it for completeness, but
will omit it from our analysis.
Not included in the above list are the vast range
of informal techniques at the disposal of the PO: the
family visits, tutoring sessions, and outings. While
these can occur on any and every day of the PO1s working
life, they are not organizational, in the sequential,
structural sense. Further, they are not the crucial
77
events in the child's progress through the system. We
will, therefore, recognize them for the important activ
ities they are, but will regretfully omit them from our
discussion.
The nomenclature for these decisions is simple.
A letter, such as "a," refers to the level of the decision
(file/do not file) , while a number, such as "1," refers to
one of the choices available at that point in the process,
(do not file). A minor complication lies in the fact that
certain decisions are repetitive, but lie at the same
level. Thus, a PO may file from either a referral or a
petition request, but both will not occur from the same
arrest. Decision al can only be compared to a2, and
similarly, a3 can only be compared to a4. However, the
nature of the decisions is the same in organizational
terms, and it seems misleading to differentiate them
further.
CHAPTER VII
Where there is experience, there is
a living being.
— Dewey
The criteria for our sample are relatively few:
1. It should be a member of a public organization.
2. It must have some decisionmaking capacity.
3. There must be some choice in the types of decisions
offered for any one situation.
4. It must have no prior experience in decision theory.
The first is merely a limitation imposed by the
discipline within which we are writing. The second re
quires something more than the mechanical ability of a
computer; for we are trying to avoid the idealized rules
that a logician might build into such a machine.
The third is partly specified by the organization
structure. Simon74 believes that the structure provides
a common set of presuppositions for certain classes of
decisions, which will be carried out by a defined class
of members. He divides organizations into three layers.
At the lower level, we have basic work processes of
78
79
production. In the middle, we have programmed decision
processes, which maintain the system. At the top, we
have non-programmed processes, where the system is contin
ually designed and re-designed. This, of course, is
simplistic, and we might expect middle echelons to main
tain a certain amount of creativity.
Indeed, some constraints of structure are desir
able from our point of view. If we were to concentrate
solely on upper echelons, we might find ourselves dealing
with situations which were political without being organ
izational. The analysis of the upper management of
Rolls-Royce is such an example. The inspectors of the
Department of Trade and Industry, inquiring into the
demise of that venerable institution, stated that: "[the
chairman and managing director] failed properly to dis
charge the responsibilities of stewardship which rest upon
the directors of a public company."75 Their sharpest
words were reserved for: "the deal that management struck
with the government at home after it had Lockheed's order
in the bag."
The behavior of the Rolls-Royce upper management
was contrary to the survival of the system. We cannot
allow the possibility of such latitude at this stage;
therefore, we shall take our sample from the area of the
80
middle echelons.
Finally, the sample must have no knowledge of
game or decision theory, so that no second guessing can
occur. On the basis of these requirements, a PO was
chosen from one of Los Angeles County's Aftercare Units.
Both the PO and her superior were given assurances re
garding the confidentiality of the work; no monetary re
wards were offered, though the results of the analysis
were to be made available to both subject and Supervisor.
The actual officer was chosen in a somewhat non-random
manner by the Supervisor, who believed that she would be
less defensive in her answers than other possible sub
jects. This was certainly true in the absolute sense:
the subject was, throughout the experiment, open and com
municative on all matters relevant to her decisions.
The approach used was essentially a phenomeno
logical one, designed to pick out the variables which the
PO regarded as important, or operative. Open-ended ques
tions were first used to delineate the Dewey-type ques
tions as they might appear to a member of the organiza
tion:
a) What is the goal of your organization?
b) By what processes do you aim for this goal? (Or,
more simply, what do you do?)
81
c) Which decisions would you choose, given a certain
situation?
Much of the background material mentioned in the
previous chapter was gained by this means. The third
question was only amenable to investigation by monitoring
the daily routines of the subject.
In fact, a minute-by-minute survey turned out to
be unnecessary. The relevant decisions have to do with
significant changes in the processing of the delinquent.
Because of the relationships between the courts, the
police, and the camps, decisions on any one case occurred
at irregular intervals of a week to several months. An
informal agreement was therefore arrived at, whereby
interviews were carried out roughly once a week, or when
ever an important decision had been made.
The various analogies familiar to the PO then
included:
State of nature.— This is a relatively simple con
cept for classical decision theory; yet it proved sur
prisingly difficult to translate to the present case.
Ideally, it refers to occurrence of one state or another;
rain and no rain, for instance. For us, it could mean
success or failure: achievement of goal, or non-achieve
ment of goal.
82
There is, first of all, the ambiguity of what the
goal is; and second, the ambiguity of whether we are re
ferring to the PO's long-term estimate of success, or the
particular estimate for a certain juvenile. The first
goal will vary according to the child and his offenses.
The second could conceivably be connected in the Bayesian
manner of general and specific estimates, but we cannot
permit unverified causal connections between each juvenile
and the Los Angeles community. There is, then, no con
sistency in either criterion — and, since we are allowing
the possibility of complex variables, we must base their
legitimacy, not on their elemental nature, but on their
consistency. The concept can, however, be incorporated
in the next section.
Frequency of observation.— Again, in classical
examples, Z, the data we derive from experiments, is a
simple concept: the reading of a barometer regarding
rain, for instance. But to obtain estimates of frequency
of occurrence of Z for different states of nature in a
real situation is an unnerving experience.
The data Z may occur in n different ways for our
experiments. The frequency of occurrence of the various
phenomena must total unity for any one state of nature.
The question we should ask then becomes:
83
"Assuming there is an underlying ethos of success
in your efforts, at what frequency will the various forms
of data (police reports, family reports, school reports,
etc.) occur?"
And then we ask the same question, assuming fail
ure instead of success. Any chance of a useful — and
respectable — answer to such questions would be low in
deed. To her credit, the subject treated exploratory
questions in this vein with an inscrutable expression and
polite statements to the effect that she did not know what
I was talking about.
The data frequency array therefore had to be at
tacked in a different way. The reason for the subject's
lack of understanding lies in the fact that classical
theory assumes the static existence of unreal, perfect
states of nature. In our world, a permanent state of
success or failure does not exist. As Dewey would say,
the successes use themselves upy the situation is a dy
namic one, with a constantly changing environment.
Further, all the data are never available, so
there is no way of knowing all the probabilities which
can lead to the perfect state; thus there is no way of
accumulating a sum of unity for probabilities under any
one state of nature.
84
What the data array does in real life is propor
tion out the amounts of data related to any one situation.
We, therefore, divided the data first into two categories,
good and had. This is arbitrary, and any other workable
solution is acceptable. This data comes from "outside,"
in the sense that it is about the juvenile, and not of
him; it consists of police reports, school reports, etc.
Next, we treated the data which is of the child:
synthesized opinions as to the child's innate abilities.
This we considered his "chance of success" (COS) — his
chance of emerging from the system before "retirement."
It should be evident that a child can thus be "good," in
the sense that the police have no hard data on him, but
yet have a low COS, because the PO, the family, and the
police know from experience what he is "really" like.
Because it contains an estimate of the child's chances of
achieving success, it is related to the concept of state
of nature.
One might expect that, given an aware officer,
the spread of good/bad data (data allocation, or DA)
would approximate the COS, or be related linearly. The
difference in a time sense is that, while the DA deals
with the past, the COS represents the PO's view of the
future. The combination of the two should provide us
85
with a synthesized idea of the PO's view of a case. Each
concept, alone, may have its uses too, depending on the
type of manipulation.
Both DA and COS were measured as a percentage,
since this seemed an easy scale for the PO to deal with;
a COS of 10, then, means that the child is given a 10
percent chance of escaping the system; a DA of 10 means
that only 10 percent of the data received about the child
is good. Combination of the two is carried out in a sim
ple four-cell table, with the percentages treated as
probabilities and combined in the manner of a contingency
table. No attempt was made to break the data down into
categories (police, school, etc.); questioning established
at an early date that such categorizing was both highly
variable and placed a strain on the PO's way of viewing
things.
Payoff function.— Because of the bureaucratic
nature of the work, it seemed more appropriate to convert
the normal payoff into a loss function. The subject was
merely asked to assign a degree of difficulty to the
relevant decision, in terms of the inconvenience, per
sonal trouble and risk that the decision entailed. As
we shall see, this did not adequately describe the entire
utility function, because it did not incorporate the posi-
86
tive effects of time-buying, and may or may not adequately
include the effects of prior decisions by an outside
agency which force the PO into certain processes. Un
fortunately, the effects of time-buying were not realized
(indeed, the policy itself was not recognized) until the
experiment was well under way. (The reason being that
time-buying maneuvers predominated in the final, disposi
tional stages of the procedure: often a good two months
from the initiation of a case.)
The incompleteness of the function is not as im
portant as it might seem, however. Exceptions are quite
evident, as we shall see, and the magnitude of the loss
function is not often crucial for our level of theorizing.
The allocation of values to this loss function
was difficult, even in this simplistic model. Using a
scale of 0-100 proved useless, in that the subject con
sidered such a scale "artificial," and, indeed, could not
accurately repeat loss values for any one decision within
15 points. A more flexible scale was then introduced,
whereby the subject was given four alternatives to de
scribe any decision: very difficult, difficult, easy,
and very easy. Unknown to her, these categories were
then given values of 4, 3, 2, and 1. These kept values
consistent to within a point.
87
The only precaution necessary was to ensure that
the subject was being specific — natural shorthand seems
to leave out such qualifiers as "very," in favor of added
stress on the operative word; if, therefore, "very" was
not used, a follow-up question had to be used to make cer
tain it had been left off deliberately. Of course, this
opens us to the charge that we might be insisting an ex
treme value on the subject. However, the logic of the
context generally made the answer clear.
In conclusion, we must admit that this accommoda
tion to the subject's convenience may have sacrificed
some of the accuracy we may have under conventional ap
proaches, but it seems to us that such accuracy is illu
sory: the result of logic external to the subject.
How much should it influence our results? Only experience
can telly but it seems to depend on what we are trying to
find out. Measuring the height of the tide with a mi
crometer is pointless, though accurate, because the varia
tion of the waves may be too coarse both in action and
origin. We are using two different paradigms, one for
the process, and one for the measurement of the process.
Planetary motions such as these must be related to plane
tary origins for them to make sense.
Similarly, we may attempt to measure with great
accuracy this degree of difficulty, but if the accuracy
88
is not as important as merely being an indication of some
larger trend, then there is little point in pursuing it.
To put the experiment in motion, loss measurements
were taken in two ways. First, a list of possible deci
sions was drawn up. The PO was then asked to consider
the child's rehabilitation, both when DA and COS were 0
and 100, and estimate the loss for each decision in each
case. She was then asked to repeat the procedure, while
considering the effect on the community of each decision.
The results are shown in Table 7-1. Though a Chi-square
test was inappropriate, one can still observe that the
differences were minor.
Since, for this PO and her charges, there ap
peared to be little goal ambiguity in this regard, the
information was treated as if it were a single case, and
the differences averaged. Next, actual decisions were
accumulated over a period of six months, and the losses
for these obtained. Eight cases were followed, and
twenty-five decisions recorded. From both sources,
graphs of Loss against COS and DA were drawn.
A look at these profiles on the following pages
is useful. The DA profiles match the COS profiles in
nearly all cases, indicating that the PO is closely re
lating what has gone before (DA) to what will come after
Decision
Loss
Rehabilitation Community
CO
o
CO
s
100
DA/COS
0
DA/COS
100
DA/COS
s1 4 4 3 4
a1 1 1 1 1
a2 4 4 4 4
a3
1 1 1 1
a4 4
1 4
1
b1 1 4
1 4
b2 2 1 2 1
b3 3 1 4
1
d1 3 1 4
1
d2 4 3 4 3
d3
2 1 3
1
d4 3
2 1
d5
1 2 1 2
d6 1 3
1 3
d7
1
4
1 4
TABLE 7-1
IDEAL DECISION LOSSES
(COS). Because of this, we will make the following dis
cussion shorter by concentrating on the COS profiles only.
The various decision groups were then bombined:
al/a2, a3/a4, bl/b2/b3, and dl/d2/d3/d4/d6. The reader
will realize that si cannot be compared to the non-deci
sion s2; and d5 and d7 are also omitted, because they are
not in the PO's operating subset. Prom these, a set of
envelope functions were obtained, which represent the
minimal losses that can be expected for varying COS. The
fact that these functions were obtained partly from actual
results does not necessarily mean that any conclusions
drawn will be circular in nature: there is room for con
tradictory results within the scheme, as we shall see.
4
Loss
0
0 20 40 60
DA/COS
FIG. 7-1
LOSS FUNCTION, DECISION a1
80
COS
DA
2
1
100
Loss
4
3
1
0
0 20 40 60
DA/COS
COS
DA
80 100
FIG. 7-2
LOSS FUNCTION, DECISION a2
4
Loss 2
COS:
DA
0
0
_ J______________ I _____________ I ________
20 40 60
DA/COS
DIG. 7-3
LOSS FUNCTION, DECISION a3
80 100
vO
00
4
Loss
3 “
2 -
1 -
0
COS
DA
40 60
DA/COS
100
DIG. 7-4
LOSS FUNCTION, DECISION a4
0 0 20 40 60 80
DA/COS
FIG. 7-5
LOSS FUNCTION, DECISION L1
__J
100
Loss
4
COS
3
DA
2
1
0
100 80
20 0
FIG. 7-6
LOSS FUNCTION, DECISION L2
lO
O'
4
COS
DA
3
2
1
0
100 60 80 40 20
DA/COS
PIG. 7-7
LOSS FUNCTION, DECISION b3
4
3
DA
2
1
0
60 100 80
20
DA/COS
PIG. 7-5
LOSS FUNCTION, DECISION di
40 60
DA/COS
PIG. 7-9
LOSS FUNCTION, DECISION d2
Loss
4
3
1
0
80 100 20
COS
DA
DA/COS
DIG. 7-19
LOSS FUNCTION, DECISION d3
100
4
COS =--
DA =--
3
2
1
0
40 DA/COS 60
100 20 80 0
PIG. 7-11
LOSS FUNCTION, DECISION d4
101
4
3
2
1
0
40 DA/COS 50
0 20 80 100
PIG. 7-12
IOSS FUNCTION, DECISION d6
102
4
3
2
1
0
40 COS 60
100 80 20 0
PIG. 7-13
MINIMUM LOSS ENVELOPE, a1 /a2
103
4° Q0S 60
80 100 20 0
PIG. 7-14
MINIMUM LOSS ENVELOPE, a3/a4
104
4
3
2
1
0
4° oos 60 80 100 20 0
PIG. 7-15
MINIMUM LOSS ENVELOPE, 131/1)2/1)3
105
0 20 40 cos 60 80
FIG. 7-16
MINIMUM LOSS ENVELOPE, d1/d2/d3/d4/d6
__I
100
106
CHAPTER VIII
Knowledge is always a matter of the use
that is made of experienced natural events.
— Dewey
The envelope functions which we have drawn up can
be looked at in two different ways. First, they are a
visual representation of the probability rule, which, as
we have stated, selects as the best decision that with
the greatest expected value for payoff. (Since loss is
actually a negative value, in our case this means the
minimum expected value for loss.) Formally, the expected
value for any decision equals the value of COS, p(c), as
a probability, times the loss for that value of COS, L.
E(V) = p (c) • L
Table 8-1 shows the comparison of the decision
that would be chosen by the probability rule against the
decision that was actually taken. Even for so small a
number of decisions, the agreement is satisfying.
Only the two d4 decisions contravene the rule.
In the first of these (Subject 2, Decision d4), the juve
nile was already in a d4 state when the petition request
107
Decision
Observed
Frequency
"Correct”
Frequency
a1 2 2
a3 8 8
b1 8 8
U2 1 1
d1 1 1
U 2 0
d6
3 3
Total
25 23
TABLE 8-1
PROBABILITY RULE
DECISION FREQUENCY COMPARISON
109
was received. Our loss profiles are based on an initial
state of zero: the juvenile is presumed to be at home,
or in some other non-decision state, where no supervision
is taking place. Since the child was already in a suit
able placement, and the feedback from this source was
generally positive, the child’s arrest was considered an
unfortunate lapse in otherwise satisfactory treatment.
In such a case, it was easier to continue the d4 state.
The wisdom of this can be seen in the fact that the juve
nile reached a COS of 50 percent after a month under the
new ruling, and 100 percent, in the PO's eyes, a few
weeks after that.
In the second case (Subject 8, Decision d4), the
shortcomings of our simple utility function are evident.
The PO did indicate that d6 was an acceptable decision
(the COS for the case was 0 percent), but refused it on
the grounds that camp treatment was too short, and she
would like to keep her eye on the child. Thus, she was
either changing her assessment of losses on the basis of
the time-buying advantages of d4, or she was rearranging
the dl-d7 continuum to some scale other than the juve
nile's perspective of severity. In either case, the
degree of difficulty (4), reflects her struggle with the
problem.
110
Alternatively, we can view our loss profiles as
classic representations of the Bayes no-data solution:
except that, instead of no data, we are assuming data
which is 50 percent good and 50 percent bad. In other
words, the data reaching the PO regarding priors leaves
her knowledgeable, but indifferent. The data array which
we obtained by multiplying the frequencies of DA and COS
is then not the initial, but the experimental array. In
fact, the PO does not need to be a Bayesian thinker until
the experiment has actually taken place; a reasonable
statement, since the prior experiment (crime and arrest)
is not performed by her anyway.
But speculation along this line is unnecessary.
The data strongly indicate that the PO is, first, ra
tional, at least when making independent organizational
decisions; and, second, that her decisionmaking behavior
can be simply explained by the probability rule, thus
rendering elaborate Bayesian analysis superfluous.
Further, a workable utility function can be easily con
structed (assuming we can include the time-buying mecha
nism in some way) without isolating a large number of
unidimensional variables. It is suggested that attempts
to find a highly-sophisticated function may actually be
putting the cart before the horse; that, in fact, the PO
Ill
works quite well with our simplified function up to the
point where complicating elements enter the picture. At
this point, the utility function is made more complex to
suit. Using the complex function in simple situations,
we think, may be misleading.
A matter related to this concerns the apparently
blithe disregard we showed for accurately determining our
loss values. While individual values may have showed
"unacceptable" variation, the data taken as a whole show
a logical consistency.
And finally, we should point out once more the
inapplicability of the familiar statistical world, in
which simple independent random sampling plays such a
large part. Here, we are dealing with one person, and
each decision is inextricably related to every other
decision. Causal factors, autocorrelations, abound.
This is a result of our orientation inward, toward the
individual; compare this with the techniques necessary —
as in the SIMBAD study — when the orientation is outward,
toward the environment.
The data we studied in the previous section made
two important, and related, assumptions. First, the deci
sions analyzed were independent: which is to say that no
outside agency was in a position to bias the outcome.
112
Second, the use of the minimum loss envelopes implied an
overriding assumption that the PO would at all costs
avoid conflict. In the a3/a4 group, for instance, a3 is
the dominant strategy for the entire range of COS; there
fore we would assume that the PO would never choose a4.
Neither of these assumptions is justified outside
our simple model; and, in fact, both assumptions were
violated in one significant case. (That it has not been
mentioned till now has not been an attempt to bias the
results; our conclusions so far remain as sound as they
might be; the case to be discussed is in the manner of an
added complexity.) In the history of Subject 6, police
action limited the alternatives available to the PO. The
juvenile was arrested but not detained, and the PO was
accordingly forced to file decision b3 (recommend no de
tention) . This she did with much difficulty (4). Upon
receiving information on a second arrest, the PO immedi
ately filed bl (detain), thus correcting the earlier
decision. However, the first decision was incorrect,
according to the probability rule, and we are left to
consider the implications of this. If the PO violates
the rule, the difficulty factor tells us, conflict re
sults. Moreover, an apparent lack of rationality results.
But this behavior is not "irrational," as we have seen;
113
it is due to the influence of other power sources, such
as the police, family, etc., and appears to represent a
more complex rationality: the compromise present in any
complex decision.
It would seem, then, that the decision theory
used by the PO is merely a discreet expression of a
larger dynamic; and the violation of this dynamic is ex
pressed through some sort of conflict mechanism. This
is most important. For at this stage, we ourselves
must decide where we want to go. We have set up the
phenomenological bases which justify the rationale used
in the SIMBAD model. In the classical sense, the SIMBAD
experiment would be the next logical step for us to take:
given that our individual is a rational decisionmaker,
let us present the universe in such an organized manner
as to fully utilize these characteristics on a large,
more general scale. Standard statistics could be used
to keep our work within logical limits. The development
is linear: we are generalizing the decision theory model.
The appearance of the anomalous decision, however,
presents us with a clue to an alternative. If decisions
are merely cross sections of some motivation, it would
be worth building on our present results to construct
114
some hypotheses for further experiment. In the next
chapter, then, we will begin an attempt to relate our
decision data to a more complex conflict model.
CHAPTER IX
The use of force to secure narrow because
exclusive aims is no novelty in human affairs.
— Dewey
The concept of conflicts ranks high on the pri
orities of most organizational theorists; for none of us
can live long in contact with other humans without being
touched by it. Yet it has also been one of the most dif
ficult to measure, largely because of the refusal of
most bureaucracies to admit that it exists.
Literature on the subject is as extensive and
historic as that of any subject in sociology. Coser
has noted;
In all social thought derived from Hegel,
particularly in Marxian thought, conflict is
the key explanatory variable. The same is
the case with social thinkers directly or in
directly inspired by social Darwinism, such
as Herbert Spencer. . . . The struggle for
power and influence is one of the themes of
Pareto's theories, as well as those of Mosca,
Michels, and Sorel. Similarly, in the clas
sical tradition in German sociology, from
Toennies to Simmel to Weber, conflict was
considered a major social p h e n o m e n o n .76
So the direction of our thought joins once more with the
classical thinkers. Two points have been central to
115
116
their expositions of the subject.
First, there is the acceptance that conflict is
an interaction between individuals and groups.77 For
Kant, antagonism between men formed the dynamic; for
Simmel, conflict fed on reciprocal action. But there was
another possibility, primitively developed by Marx —
which was that men did not necessarily fight each other,
but, in their battles to transform themselves, they might
do battle against nature itself.
The two forms are possible. Yet the second, in
which there are no antagonists, only differing goal di
rections, has been systematically excluded from conflict
theory, and has been segregated under the name of compe
tition. The reason for this obsessive and exclusive
classification is unclear. Dahrendorf admits two kinds
of struggle;78 there are those brought in from the out
side, and those born within the system. Boulding79 rec
ognizes that the world may now be so complex that people
may be in competition without realizing it; in an organ
izational world, the same should be true of alienated
conflict.
Still we have Schmidt and Kochan®® denying this
as a form of conflict. They quote Fink®-*- as believing
that competition involves a respect for rules, while con-
117
flict is unregulated, and involves violation of rules.
The difference is between parallel striving (competition)
and mutual interference (conflict).
It seems to us that such a differentiation might
be satisfactory to a sociologist in general, but it is
unreal in organizational terms. Conflict in organizations
can appear in the very act of following rules, especially
those which might be ambiguous, as we have noted in the
case of probation policy. Under the circumstances, dis
turbances are all the more destructive because of the
bureaucratic inability to interfere in another's business.
There is still the awareness that future desirable posi
tions might be made impossible to occupy, if present
decisions are made in a certain way: yet the rigidity of
rules allow for nothing else to be done. Our suggestion
is that concentrating on interaction within organizations
unnecessarily limits the study of conflict, and that an
equally important area of study lies with the decision-
making sector.
The second factor in classical thought deals with
the implications of such excessive and academic classi
fication. We have Fink dealing with Sorokin's classifi
cation of parties who may be involved in conflict —
races, families, etc. Mack and Snyder®^ classify accord
118
ing to internal characteristics — motives, information,
size, etc. Waldo®3 classifies on the basis of moral and
immoral societies. Gouldner®^ classifies on the basis of
bureaucratic typologies. Katz and Kahn®® classify by
role typologies. Etzioni®® starts from the other end and
classifies by coerciveness and compliance. House and
Rizzo®^ classify by employee satisfaction, supportive
leader practices, etc. Boulding's work, already men
tioned, classifies according to sets of human time and
energy. And Galtung8® halves Boulding's eight-set model
to produce four sets, based on the nature of the parties,
and the structural relations between the parties. Essen
tial to these models is the familiar (a) construction of
the ideal universe, into which examples are then fitted;
and (b) the freezing of the model at that point, so that
the cells (but not the transition) can be discussed.
The results of this have been clear to more than
this author. Fink himself notes that:
Even a master list of all the distinct types
of conflict that appear in these several
schemes taken together would not solve the
problem.89
Fink sees the way barred by the different frame
works which the various authors use. We see this as
illusory: what Fink seems to be asking is for everyone
to get together to produce one last, ideal classification.
119
The problem is more fundamental: we need the extra intui
tion that Dewey is talking about to turn the aggregate
into a higher level of conceptualization, and no amount
of reclassification will achieve this.
An alternative approach has been seen by at least
three different writers. First, Gross^O notes the close
relationship between conflict and its opposite, coopera
tion. He sees no resolution as necessarily final:
indeed, the resolution may itself be the prelude to new
conflict. Combine this with his emphasis of the "vector"
nature of conflict, and we can see that his thinking is
both dynamic and interactive in the pragmatic sense.
Further, he sees conflict as complex, embedded in a matrix
of influences: a description similar to our organiza
tional position.
Though their definition of conflict differs in
this respect, Schmidt and Kochan also attempt a dynamic
picture by placing conflict on a three-dimensional axis,
with goal incompatibility, activity dependence, and shared
resources as the three dimensions. Interesting as the
method might be, however, we can see no immediate appli
cation of mathematical techniques, and, more importantly,
their assumptions about the nature of the conflict make
attempts to adapt their model unwise.
120
Though it is six years older, the final work by
Sisson and Ackoff^l comes closest to our views. In this
paper, the authors simulate a conflict situation, using
a standard decision theory technique, complete with prob
abilities for actions, outcomes, and utilities. Players
gain experience which guides their later actions, and,
while the technique may not be strictly Bayesian, the
ability to learn is used to modify utility functions.
The theory is inclusive, in that it applies to both
"conflict" and "competition," and the writers seem to be
aware of complex situations in which both are inextri
cably present.
Their criterion is the degree of cooperation, or,
with a negative sign, degree of conflict. It is obtained
by subtracting the expected values for the two players
for a combination of the probability of selecting a cer
tain procedure, the probability that a certain outcome
will eventuate from this procedure, and the relative
utility of that outcome to the player.
We can find very little wrong with this approach.
It is a social conflict model, but modification to an
organizational mode would require only minor adjustment.
Perhaps our most serious objection is that the model is
still heuristic at a point where it should be postulative;
121
by this we mean that it simulates without knowing why the
difference in expected values should correspond to a de
gree of conflict. The theory is empirical in the Greek
sense, rather than the pragmatic sense; the "office of
reason" is not defined.
As a result, the Sisson and Ackoff work is con
sidered a step in the right direction; but something
more must be added.
The reader may recall that, in our first chapters,
we lingered on the possibilities of innovating through
metaphoric transfer. By this method, we hoped to see un
familiar problems in a familiar light.
Within sociology, such grafting from the natural
sciences has always been a temptation hard to resist.
Miller^ notes that as far back as 1910, Haret was basing
a law of attraction of people on the inverse square gravi
tation law. The important thing is not the use of such
a law, of course, but its use without empirical justifica
tion.
Lewin's field theory hardly does better.93 The
concept has difficulty making up its mind whether to be
magnetic or chemical; and in either case, the use of
valency to express attraction and repulsion implies under
lying structures which Lewin failed to verify. (Does
122
Lewin intend us to believe that there is a 2, 4, 8, etc.,
atomic structure in personal interaction?)
It is interesting to follow the fashion in socio
logical analogy and compare it to progress in the contem
porary natural sciences. In general, sociologists have
tended to fall upon those aspects which have the most
immediate effect, but the least long-range value. Haret's
interest in gravitation came when physics had passed on
to the deeper problems of relativity. It was Newtonian
physics at a time when Einstein was the physicist.
Lewin's theories relate directly to the work of Dalton,
at a time when Chadwicke was smashing the atom. One
cannot help but wonder if systems theory is doomed to the
same fate.
The problem with such theories is that they seem
to have been artificially introduced by some theorist who
has been struck by the similarities between the total
theory (e.g., the law of gravitation) and his conceptual
view of the world (human interaction). The vital link
between thought and empiricism is ignored. Another way
of saying this is that it is uncommon for abstract theory
and sociological conceptualization to evolve together;
for in its modern form, such conceptualization demands
contemporary investigation.
123
Whether by accident or design, decision theory
and its esoteric offshoot, information theory, have
achieved such a parallel development. This seems to have
been due at least in part to the diversity of origins? one
is almost tempted to ascribe a Marxist ethos to the manner
in which several disciplines have seen their problems in
a similar light.
Thus Garner^ points out that one origin can be
traced to Samuel Morse, who related his code to the fre
quency of use of various letters? that, around 1924,
Nyquist and Kupfmuller developed an interrelation between
the number of alternatives per event and the number of
successive events. Hartley developed the log system of
measuring the number of alternative possible sequences
within any one message in 1928, and in 1946 Gabor at
tempted to define an elementary quantum of information
called the logon. In 1948, Wiener's work on cybernetics,
and Shannon's work on communications, were published.
Garner points out that Shannon's contributions were,
first, to treat the problem as a statistical one, and
second, to address himself to the concept of uncertainty
in a message.
But the statisticians, too, were involved in this
problem of information. From 1922 to at least 1934,
124
Fisher was trying to view the efficiency of a statistic
as a function of the available information, and was trying
to measure the concept in an additive way. In 1950, McKay
went into the classification business by distinguishing
logon content, which was a priori information, from metron
content, which was experimental'information. The linking
between Gabor's earlier work, and the later statistics of
Savage, are evident.
But it is important to us to note how quickly the
psychologists found uses for Shannon's worky Miller and
Frick were to introduce the concept to psychology barely
a year after Shannon published his paper. In Broadbent's
words:
S-R theorists have used as their central variable
the occurrence of stimuli, past and present.
Gestaltists have considered the pattern of stimuli
present at any one time. But neither of them have
talked about the effect at any moment of the stim
uli which might have occurred, but did n o t . 95
Broadbent's allusions are in fact describing the
change that has taken place in empiricism. We have gone
from the Greek and Lockeian modes of empiricism to a new
method, in which we are not at all sure where we are
going. In their attempts to describe this insecure ex
perience, the various fields have managed to describe one
common aspect: the uncertainty of it all. It is no co
incidence that Kaplan has pointed out that information
125
theory is fundamental to the pragmatic approach.
Tribus^S tells us that it was at the suggestion
of John Von Neumann, the great game theorist, that
Shannon called this function of uncertainty, entropy.
In 1957 E. T. Jaynes demonstrated that the
function had deeper meaning than had been
supposed and gave a demonstration of its
use in statistical mechanics. In 1961 it
was demonstrated that the two entropies
were, in fact, examples of the same idea
and therefore were not merely analogies.97
It is thus a distortion to believe that entropy
is a simple borrowing of a thermodynamic term used in the
mid-nineteenth century by Clausius and Boltzmann. Indeed,
it is even possible that such concepts as systems theory,
which rely for much of their strength on the thermo
dynamic legitimacy of the entropy function, may in fact
be putting the cart before the horse. Thermodynamics
uses entropy in a specialized way peculiar to itself, not
vice versa. Entropy can thus survive, even if what hap
pens in thermodynamic theory does not describe what
happens in organization practice.
In any case, entropy, as used by Shannon, has
possibilities for us which extend into the very way we
make decisions. Unfortunately, theoretical papers have
tended to dwell on the mathematical aspects, while
126
empirical work has been limited to engineering problems
and matters of choice in experimentation. Its applica
tion to organization behavior is a challenging task; one
to which we shall apply ourselves in the next chapter.
CHAPTER X
Experience in its vital form is experimental,
an effort to change the given; it is character
ized by projection, by reaching forward into the
unknown; connection with the future is its sali
ent trait.
— Dewey
Within any organization, goal conflict entails
three things:
1. It involves a sequential decisionmaking process,
which is affected by the communication of events
and perceptions of the environment.
2. It involves a restriction of choices, in order that
the achievements brought about by constraints in our
behavior will pass through a certain goal point, or
nexus.
3. It involves differences in perception of the deci
sionmaking role by different participants.
To begin, then, we have two basic points — the
nexus, and the point at which the nexus is perceived.
But to place these two points in some context, we must
define the coordinates of a system. In goal conflict,
there are three:
127
128
1. The difficulty of the goal.
2. The time taken to achieve the goal.
3. The amount of Whatever policy is required to achieve
the goal.
So far, this is not a goal conflict system, but a
goal achievement system. The moment of perception of the
nexus is the point (0, Q, 0), and the nexus is the limit
ing point of the system (Nd, Nt, Np), beyond which a new
goal must be defined — or, in the case of a continuing
decision procedure, such as we have here, the coordinate
system remains the same, but the origin migrates to point
N at the arrival of each new decision.
Our task now becomes that of defining the locus
of ON. The simplest case is that of a frictionless
machine doing work in a perfect and frictionless world.
In such a situation, no effort is required to achieve the
goal, and ON becomes a straight line in the difficulty/
time plane. The achievement of the goal, in other words,
only requires the time taken to get there. This could
have been predicted from Newtonian mechanics, for the
equation is the simple vector
V = i • j
defined in the special case as
Work = Force x Distance
129
or
Achievement = Difficulty x Time
In practice, some friction would occur, and the
locus would become a straight line with a slope into the
policy-effort planes. In organizational terms, both
these cases correspond to simplistic views of decision
making. It is only necessary to define the goal, and
then logically divide the work up into efficient units.
A hypothetical illustration will be dealt with later.
Now, as we enter the real world, and humans re
place machines, the situation changes. Mechanical effort
becomes decisionmaking behavior, and the causal relation
between 0 and N becomes more complex.
Let us first take our assumption that goal
achievement involves the gradual restriction of choices.
At the moment before we enter the system, our choices are
unlimited; we may even, if we wish, ignore the system al
together, and pursue an infinity of other goals. But to
enter the system will reduce our choices by at least one;
the value set (utility functions, etc.) , which will go
to make up our decision behavior will have been defined.
As we move along the decision tree from 0 to the final N,
freedom of choice becomes increasingly restricted until,
130
just before we enter N (i.e., as the problem tends to the
limit), we are left with a (0, 1) choice: either we
enter N and end the effort, or we do not enter N and the
goal is left unachieved. At this point, freedom is at a
minimum, and our choices relative to the goal are defined
with a closeness approximating certainty.
This simplistic world, in which randomness is
reduced to certainty has been familiarly described as a
reversed entropic process; or, to put it another way, an
entropic process 180° out of phase. Basic information
theory tells us that the locus of such a process is
H = -Sp± • log p±
where H is the entropy of the system, and p is the prob
ability that the value set will be in i space— that is,
a value set whose probability of occurrence is p.
Our next point involves the perception of roles
by decisionmaking participants. There are two ways of
looking at this. First, the fewer choices we have left,
the more ingenuity we must exercise. If this ingenuity
is hampered, frustration results. If we perceive another
person doing the hampering, conflict erupts. This is the
case dealt with by Kochan and Schmidt. But achievement
is also a projective process; we each have our own ideas
131
as to how a goal should be reached. If the person actu
ally involved in the decisionmaking is going by a differ
ent decision tree to the one we would take, then this is
similar to being made to take that path ourselves. This,
too, is a restriction of choices, but a worse one, in that
the decisions taken may not even be ones we would normally
consider. In this situation, the differences in the
decision trees — the one being taken, and the one we
would take if we were doing the job — is an indication
of the frustration and eventual conflict.
To take this the logical step further, the dif
ference in decision trees can also be expressed as the
difference in entropy curves. This difference will show
the potential for conflict that exists between partici
pants. It will probably not be a direct measure of con
flict, because the amount of tolerance that exists be
tween the parties will depend on other elements of the
organizations in which they exist — the authority rela
tionship, for example. The possible existence of a
modifying parameter would have to be tested empirically.
However, it is possible that such a parameter
would have only minor significance in open conflict
situations. The entropy curve is exponential. This
means that, the further along the curve we move — the
132
closer we get to the goal — the steeper the curve be
comes: small changes in time will produce large changes
in decision behavior. This is what Dahrendorf®® meant
by indicating that the speed of development, and the in
tensity, of conflict, depend on the changes that occur.
The effect of the parameter as a deflator or inflator
would be small relative to the exponential changes which
were occurring.
Figure 10-1 shows a possible description of this
process. The goal difficulty-dimension is assumed con
stant for ease of presentation, and the curve is shown in
two dimensions. Phase I describes the normal, almost
mechanical conflict which exists in any organization be
cause of procedural differences (red tape). This is
latent conflict; under normal circumstances, goals are
achieved before the curve starts to rise. Phase II
represents a point at which the decisions become strate
gic: both sides are aware of the other’s actions, but
common norms prevent interference. This is potential
conflict. Finally, Phase III represents open conflict,
in which all pretence is discarded, and interference is
mutual and direct.
The exact cutoff points for each phase are, of
course, unknown. The time for cooperation is probably
133
ENTROPY
CURVES
CONFLICT
PHASE III
PHASE II
PHASE I
0
Time ^
PIG. 10-1
CONFLICT PATTERN
FOR DIFFERENCES IN
GOAL TIME PERCEPTION
134
in Phase II; motivation is too low in Phase I, and the
utility value of Phase III, when perceived, must be very
small. Three ways out of this problem can be seen. The
first is to simply reverse the initial decision, and drop
out. The second is to persist in our procedure, hoping
that, when conflict does break out, the other party will
be forced out by events — a strategy for brinkmen. The
third is, by some bargaining procedure, to change the
goal perception, or the entropy curves of the partici
pants.
Under normal circumstances, how would we expect a
participant to proceed? We are reminded of Simon's
thoughts on the subject.
I might mention that, in the spirit of crude
empiricism, I have presented a number of
students and friends with a problem involving
a multiple pay-off — in which the pay-off
depends violently upon a very contingent and
uncertain event — and have found them ex
tremely reluctant to restrict themselves to
a set of behavior alternatives allowed by the
problem. They were averse to an alternative
that promised very large profit or ruin, where
the relevant probability could not be com
puted, and tried to invent new alternatives
whose pay-pffs were less sensitive to the
contingent e vent .99
The reader experienced in the ways of bureaucracy
will recognize the strategy. It is called "keeping your
options open," and is an attempt to deal out choices
135
sparingly, to keep as many decisions available as pos
sible. In our present language, we are attempting,
through a maximization of possibilities, to maximize our
entropy curves. With this in mind, we will attempt to
apply the model to our existing case.
The theoretical coordinate system is too unwieldy
in its present form, and needs to be expressed in more
useful terms. Therefore, we shall define:
a) goal difficulty in terms of COS — on the theory that
it is more difficult to succeed with a low-COS juve
nile than it is with a high-COS one?
b) policy effort now becomes our decisionmaking strategy,
and can be measured in terms of entropy;
c) time, in the conventional sense, is irrelevant, since
variation is caused as much by judicial procedure as
anything else. A more systematic measure is found by
using the decision path a/b/d, the distances between
each decision point being uniform.
The calculation of entropy in any case will de
pend on how much information we think we have. Maximiza
tion procedures are still new, and simultaneously lacking
in simplicity and flexibility. Tribus^O® is not child's
play (at least not to this author), but he is the most
available, and we shall follow his derivations.
136
Three cases seem appropriate.
1. Case of minimal knowledge (uniform distribution).
This is the classic existential case, in which
all possibilities are equally likely; in other words,
n
S P± = 1
i=l
where pi is the probability of occurrence of the ith
alternative, given that there are n alternatives.
For maximum entropy, the distribution of probabilities
becomes
p± = exp (-wQ)
where wQ is a Lagrange multiplier. This potential func
tion is found from
wQ = In n
and hence we have
1
Pi = n
and entropy expressed as
H = In n Napiers
Decisionmaking of this sort is unlikely, given
the experience of the PO.
137
2. Case of useful knowledge (exponential distribution).
This represents our present case, where the PO
has been making her decisions in a manner which can be
considered rational; in which some estimate of correct
procedure can be made; and in which the consequences of
her decisions are, through prior experience, known to
her. The probability rule is applicable.
The equations of constraint are then
n
S Pi = 1
i=l
p± • v(0±) = (V)
where V(O^) is the utility (in our case, loss) of a given
alternative i, for a COS of 0. (V) represents the desired
expected value for this cluster of alternatives; it is
thus the strategy factor. We decide just how much incon
venience we are willing to put up with in our decisions,
and distribute them accordingly. Another possibility is
that, given a Bayesian process, the (V) may represent the
weighted average risk. It is highly probable that this
factor varies from one PO to another. In any case,
Pi = exp [_w0 -w2 • ]
138
where both wQ and w^ are Lagrange multipliers? for there
must be as many multipliers as there are equations of
constraint. The potential function is found from
wQ = In exp [-w1 • VfC^) ]
with entropy given by
S = wQ + wi • (V)
Some numerical examples will illustrate the pro
cedure. Let us assume that the PO is prepared to put up
with more than a minimal amount of loss; let
(V) = 2
V(0±) = i
which is to say that the losses for each group of alter
natives are uniformly spaced, just as we have arranged
our loss continuum of 1, 2, 3, 4; of course, this could
be extended to higher numbers.
Table 10-1 shows some sample distributions, to
gether with the entropy in Napiers. Note that, for the
three-decision case, choosing an expected loss value
equal to the midpoint is equivalent to a uniform distri
bution; being "indifferent" in this way is equivalent to
having minimal knowledge. It is also clear that having
139
3
Decisions
4
Decisions
5
Decisions
p1
.33
.42 .46
P2
.33
.28 .26
e3
.33
.18 .15
p4
— .12 .08
p5
—
—
.05
Entropy
(Napiers)
1.10 1.25 1 .35
TABLE 10-1
MAXIMUM ENTROPY TABLE
(Y) = 2 , V(ei) = i
EXPONENTIAL DISTRIBUTION
140
a large catalogue of alternatives also makes sense, in
that it increases the possible entropy.
A possible shortcoming in this method is that the
classification is assumed to be an exhaustive set of
mutually exclusive outcomes. The dl/d6 group we used in
our study of the probation system may not fulfill this
requirement; a more rigorous analysis may be required.
3. Case of statistical information (Gaussian distribu
tion) .
It is possible that we may have enough informa
tion to say that the variation in probabilities should
be approximately normal within the limits of our loss
values, and that a certain amount of variance from our
expected value can be specified as acceptable. If this
is so, the equations of constraint become
SPi = 1
2 P i V . = (V)
SPiCVi - (V)]2 = d2
where (V) is the expected value and is the desired
variance. Three Lagrange multipliers are required to
give
p± = exp (-wQ —w-^ • V± -w2 • V 2)
141
from which the entropy can be calculated by standard
means. However, the mathematics of this technique are
quite cumbersome. Tribus does provide a series of graphs
and tables which lead to the desired results.I®* Our
present data, however, do not warrant the use of such a
sophisticated technique.
There are several different ways of viewing the
application of this method. First, we have noted that
the goal definition may lead to ambiguity, so that the PO
may be forced to choose on the basis of rehabilitation of
the child, or protection of the community. Our present
data do not indicate a useful difference for this PO? but
for another, we might expect differing entropy values,
and the differences would indicate a conflict within the
PO herself.
Another possibility exists if the PO is dealing
with a "tough" police department, who believe that maxi
mum detention is the only way to handle offenders. In
this case, the police would always desire her to take
a3/bl/d6, even in cases where the COS approached 100 per
cent. They would perceive the entropy curve as zero for
all values. The indication of conflict would then be
entirely represented by the PO's entropy curve. Several
jpoints are worth noting:
142
a) The entropies for each decision are additive; i.e.,
the nexus for group a becomes the origin for group b.
Thus, entropy can still rise rapidly for unresolved
conflict, even if the number of choices in ensuing
sequences decreases. The total curve is also not the
pure exponential curve we have illustrated in the
previous simple cases.
b) The effects of authority can be illustrated if we
assume that, in the limit, the police policy is
forced on the PO. Then her conceptual losses can
rise to a maximum, and the entropy curve will rise
accordingly. At the same time, her actual decision
procedure will have become so certain that real
entropy drops to zero. Again, conflict is the
deviation between the two.
c) The choice of an expected value will vary according
to circumstances, but can be no more than the highest
loss for any one choice. Ideally, one would hope for
a value of one — the minimum — but empirical work
would be needed to show the acfcugl variation.
d) Similarly, empirical work would have to show the
allocation of probabilities that the PO considered
desirable. Obviously, any other allocation would
143
give a smaller entropy than the maximum used here;
the question remains whether an allocation less than
the maximum is sufficient for an ideal curve, or
whether any deviation from the maximum will produce
conflict.
CONCLUSION
Nothing can eliminate all risk, all adventure;
the one thing doomed to failure is to try to keep
even with the whole environment at once — that is
to say, to maintain the happy moment when all things
go our way.
— Dewey
The purpose of this design has been to make a
distinction between longitudinal, innovative research, and
the more traditional lateral, confirmatory methods. While
the latter are useful in "filling out" the hypotheses con
tained within a certain paradigm, the former technique
attempts to traverse paradigms by emphasizing the experi
mental characteristics of dynamism and probability. The
role of intuition in developing these new conceptual
analogues must be stressed.
The investigation took advantage of the fact that
the repetitive patterns of human interaction which occur
in organizations provide a bridge between the pragmatic/
positivist techniques, and the existential/phenomeno
logical viewpoints. Because of this, the potentially
infinite study of variables can be controlled, and limited
to those which seem relevant and familiar to the subject,
144
145
the interviewer, and the work on hand. The synthesis can
be described in these terms:
1. "Bracket" the problem superficially. Collect a group
of analytic terms which may be relevant.
2. "Bracket" the problem internally. Extract from the
human subjects those concepts which they consider
essential to the problem and its solution.
3. Associate the two lists through intuitive analogy.
4. Empirically determine the relationships between the
analogies.
5. Using these relationships, develop an intuitive model
which describes the transcendental functioning of the
variables.
The method then depends for verification on a
return to stages (2) or (4) for further empirical
analysis. It does not expect that the variables will be
unidimensional, only consistent; that the variables need
be as precise as classical statistics might desire; or
that a final product can be analyzed, and a null hypoth
esis proved or disproved. In Bayesian terms, it proceeds
through a reinforcing (or reassuring) series of strong
priors, which serve to keep our intuition in touch with
reality. In metaphorical terms, it is like walking
through a dark corridor, with only an occasional brush
146
against the wall to give us our bearings.
While this may sound radical, it is far from
being so. The astute reader will have recognized that
we have simply moved the traditional process of scien
tific investigation through one-half phase: experimental
confirmations come in the middle of the design, while
hypotheses only begin to appear near the end. This lack
of a finish may seem unfortunate, but what we lose in
certainty we make up in the added conceptual effort
needed to take what we know from experience to what we
know by reasoning from experience.
The need for this type of investigation lies
deeper than a mere quest for novelty. Two of the greatest
hindrances to scientific advance in sociology lie in: (a)
the mindless accretion of data which continues to over
whelm us past a point of marginal returns. By forcing
the researcher to go one step further and indicate that
his data mean something more than they say, we would
force large numbers of earnest but voluble page-fillers
back to their benches, and mercifully turn the journals
from the function of a data bank to a genuinely informa
tive role. Further, we have (b) the fear that we may say
something wrong. Investigation is a gamble in which we
try to second-guess the unknown. Tedious attempts to
147
block all the escape hatches before we move on may be
academically respectable, but they also reduce us from
discovers to accountants. Mistakes, like our methods
here, are temporary things, to be righted by further
experience. Certainly they are dysfunctional, and to be
avoided where possible. But they are part of the game,
a function of hazardous perception and insufficient in
formation, and are not a general indication of idiocy.
Recognition of this would greatly facilitate the
process of innovation. Refusal to speculate limits us;
realizing that "we know more than we can say," liberates
us.
REFERENCES
REFERENCES
^E. Schein, Address, Division 14, American Psy
chological Association, September 5, 1971.
2R. MacLeod, "The Phenomenological Approach to
Social Psychology," Psychological Review, 54 (1947),
pp. 193-210.
3j. Dewey, Studies in the History of Ideas (New
York: Department of Philosophy of Columbia University,
1935), pp. 3-32.
^A. Kaplan, The Conduct of Inquiry (San Francisco:
Chandler Publishing Co., 1964), p. 36.
5ibid., p. 9.
6h . Simon, Models of Man (New York: John Wiley
& Sons, Inc., 1959), p. 11.
7r . Bendix, Max Weber (New York: Doubleday &
Company, 1960), p. 414.
®D. Schon, Displacement of Concepts (London:
Tavistock Publications, 1963).
9G. Miller, et_al., Plans and the Structure of
Behavior (New York: Henry Holt & Co., 1960), pp. 41-42.
1°J. S. Mill, A System of Logic (New York:
Harper & Row, 1874), pp. 393-394.
H n . Wiener, The Human Use of Human Beings
(London: Eyre & Spottiswoode, 1950), p. 3.
*2A. Kaplan, The Conduct of Inquiry, p. 263.
13r . Bierstedt, Emile Durkheim (London:
Wiedenfeld & Nicolson, 1966), p. 129.
l^E. Durkheim, The Rules of the Sociological
Method, ed. by G. Catlin, trans. by S. Solovay and J.
Mueller (8th ed.; New York: The Free Press, 1962), p. 141.
149
150
^Kaplan, The Conduct of Inquiry, pp. 38-39.
Miller, "Toward a General Theory for the
Behavioral Sciences," American Psychologist, 10 (1955),
p. 517.
■^rbid., p. 519.
1®J. Dewey, On Experience, Nature, and Freedom,
ed. by R. Bernstein (New Yorks The Bobbs-Merrill Co.,
1960), p. 22.
j. Monod, Chance and Necessity (New York:
Vintage Books, 1972).
20r . Bernstein, Praxis and Action (Philadelphia:
University of Philadelphia Press, 1971), p. 169.
2lDewey, On Experience, Nature, and Freedom, p. 19.
p o
Bernstein, Praxis and Action, pp. 174-176.
23Ibid., p. 177.
2^T. Harris, I'm OK — You're OK (New York: Avon
Books, 1967) .
23Kaplan, The Conduct of Inquiry, p. 42.
°J. Coleman, Introduction to Mathematical Soci
ology (Glencoe: The Free Press, 1964), pp. 8-54.
27c. Coombs, "Psychological Scaling Without a
Unit of Measurement," Psychological Review. 57 (1950),
p. 145.
2®F. Stephan, and E. Mischler, "The Distribution
of Participation in Small Groups: An Exponential Approxi
mation," American Sociological Review, 17 (1952), p. 598.
29simon, Models of Man., p. 12.
3®Bierstedt, Bnile Durkheim, p. 94.
3]~Ibid. , p. 125.
32Ibid., p. 129.
151
33c. Broad, The History of Science (Carlton:
Melbourne University Press, 1958), p. 51.
34lbid., p. 52.
35G. Sorel, The Illusions of Progress (Berkeley:
University of California Press, 1969), p. 86.
36lbid., p. 83.
37lbid., p. 87.
38c. Russett, The Concept of Eguilibrium in
American Social Thought (New Haven: Yale University
Press, 1966).
39Ibid. p. 44.
40Ibid. p. 46.
41Ibid. p. 40.
43lbid. p. 32.
43Ibid. p. 40.
44Ibid. p. 50.
45Ibid. p. 42.
4^Ibid. p. 43.
47Ibid. p. 34.
4®P. Honigsheim, On Max Weber (New York: The
Free Press, 1968), p. 114.
49P. Kilby, Entrepreneurship and Economic Devel
opment (New York: The Free Press, 1971), p. 72.
SOfiernstein, Praxis and Action, Part I.
SlHonigsheim, On Max Weber, p. 14.
52m . Weber, The Protestant Ethic and the Spirit
of Capitalism, trans. by T. Parsons (New York: Charles
Scribner & Sons, 1930).
152
33J. Freund, The Sociology of Max Weber (New
York: Vintage Books, 1969), p. 7.
5^Ibid., p. 9.
55r . Mannheim, From Karl Mannheim, ed. by K.
Wolff (New York: Oxford University Press, 1971), p. 56n.
56Honigsheim, On Max Weber, p. 114.
S^Kilby, Entrepreneurship, p. 91.
58Ibid.
59a . Wald, Statistical Decision Functions (New
York: John Wiley & Sons, Inc., 1950).
60H. Raiffa, and R. Luce, Games and Decisions
(New York: John Wiley & Sons, Inc., 1967).
Chernoff, and L. Moses, Elementary Decision
Theory (New York: John Wiley & Sons, Inc., 1970) .
62simon, Models of Man, p. 244.
63W. Barnette, Readings in Psychological Tests
and Measurements (rev. ed.; Homewood: The Dorsey Press,
1968), p. 59.
64ibid., p. 63.
f p. 64.
56a . McEachern, et al., "The Juvenile Probation
System: Simulation for Research and Decision-Making,"
American Behavioral Scientist, Vol. 11, No. 3 (January-
February, 1968), pp. 1-45.
57ibid., p. 28.
68A. Maslow, Toward a Psychology of Being (2d ed.;
New York: Van Nostrand Reinhold Co., 1968), p. 155.
88Dewey, On Experience. Nature, and Freedom.
p. 28
^ M c E a c h e r n , et al., "The Juvenile Probation
System," p. 7.
153
^County of Los Angeles Probation Department
Information Series, December, 1971, No. 8, Probation
Camps, p. 4.
^^McEachern, et al., "The Juvenile Probation
System," p. 13.
73Ibid.
7^H. Simon, The Shape of Automation (New York:
Harper & Row, 1965), p. 63.
75iiRolls-Royce: The Verdict of the Inspectors,"
The Economist (August 4, 1973), p. 68.
76L. Coser, "Social Aspects of Conflict," Int.
Encyc. of Social Sciences (New York: Macmillan, Vol. 3
(1968), p. 232.
77L. Coser, Masters of Sociological Thought (New
York; Harcourt, 1971) .
7®R. Dahrendorf, "Toward a Theory of Social Con
flict," Journal of Conflict Resolution, Vol. 12, No. 2
(June 1958), pp. 170-183.
79k . Boulding, Conflict and Defence: A General
Theory (New York: Harper & Row, 1962), pp. 105-113.
®®S. Schmidt, and T. Kochan, "Conflict: Toward
Conceptual Clarity," Administrative Science Quarterly
(September 1972), pp. 359-370.
81c. Fink, "Some Conceptual Difficulties in the
Theory of Social Conflict," Journal of Conflict Resolu
tion, Vol. 12, No. 4 (1968), p. 443.
®^R. Mack, and R. Snyder, "The Analysis of Social
Conflict," Journal of Conflict Resolution. Vol. 1, No. 2
(1957), pp. 212-248.
83d . Waldo, The Administrative State (New York:
Ronald Press, 1948.
®^A. Gouldner, Patterns of Industrial Bureaucracy
(New York: Routledge & Kegan Paul, 1955), pp. 215-218.
154
85D. Katz, and R. Kahn, The Social Psychology of
Orqanizations (New Yorks John Wiley & Sons, Inc., 1966),
pp. 184-185.
86A. Etzioni, Modern Organizations (Englewood
Cliffs, N.J.: Prentice-Hall, Inc., 1964).
87r . House, and J. Rizzo, "Role Conflict and
Ambiguity as Critical Variables in a Model of Organiza
tional Behavior," Organization Behavior and Human Per
formance, Vol. 7 (1972), pp. 467-505.
®®J. Galtung, "Institutionalized Conflict Resolu
tion," Journal of Peace Research, Vol. 2, No. 4 (1965),
p. 348.
8^Fink, "Some Conceptual Difficulties in the
Theory of Social Conflict," p. 423.
90b . Gross, The Managing of Organizations (New
York: The Free Press, 1964).
91r . Sisson, and R. Ackoff, "Toward a Theory of
the Dynamics of Conflict," Papers, Peace Research
Society (International), Vol. 5 (1966).
92Miller, "Towards a General Theory for the Be
havioral Sciences," p. 521.
93k . Lewin, A Dynamic Theory of Personality:
Selected Papers (New York: McGraw-Hill Book Company,
1935).
9^W. Garner, Uncertainty and Structure as Psy
chological Concepts (New York: John Wiley & Sons, Inc.,
1962), pp. 8-10.
95lbid.? p. 13.
96m . Tribus, Rational Descriptions, Decisions,
and Designs (New York: Pergamon Press, 1969), p. 110.
97ibid.
98Dahrendorf, "Toward a Theory of Social Con
flict," pp. 170-183.
155
^ Simon, Models of Man, p. 253 f.
Rational Descriptions, Decisions,
and Designs, pp. 119-180.
•^•^Ibid., pp. 136-141.
BIBLIOGRAPHY
BIBLIOGRAPHY
Barnette, W. Readings in Psychological Tests and Meas
urements . Homewood: The Dorsey Press, 1968.
Bendix, R. Max Weber. New York: Doubleday & Company,
1960.
Bernstein, R. Praxis and Action. Philadelphia: Univer
sity of Philadelphia Press, 1971.
Bierstedt, R. Emile Durkheim. London: Wiedenfeld &
Nicolson, 1966.
Boulding, K. Conflict and Defence: A General Theory.
New York: Harper & Row, 1962.
Broad, C. The History of Science. Carlton: Melbourne
University Press, 1958.
Chernoff, H., and Moses, L. Elementary Decision Theory.
New York: John Wiley & Sons, Inc., 1970.
Coleman, J. Introduction to Mathematical Sociology.
Glencoe: The Free Press, 1964.
Coombs, C. "Psychological Scaling Without a Unit of Meas
urement," Psychological Review, 57 (1950), 145.
Coser, L. Masters of Sociological Thought. New York:
Harcourt, 1971.
_. "Social Aspects of Conflict," International
Encyclopaedia of Social Sciences. New York:
Macmillan Company, 1968, 232.
Dahrendorf, R. "Toward a Theory of Social Conflict,"
Journal of Conflict Resolution. Vol. 12, No. 2
(1958), 170-183.
Dewey, J. On Experience. Nature, and Freedom. New York:
The Bobbs-Merrill Company, 1960.
_. Studies in the History of Ideas. New York:
Department of Philosophy of Columbia University,
1935.
157
158
Durkheim, E. The Rules of the Sociological Method. New
Yorks The Free Press, 1962.
Economist, The. "Rolls-Royce: The Verdict of the In
spectors" (August 4, 1973), 68.
Etzioni, A. Modern Organizations. Englewood Cliffs,
N.J.: Prentice-Hall, Inc., 1964.
Fink, C. "Some Conceptual Difficulties in the Theory of
Social Conflict," Journal of Conflict Resolution,
Vol. 12, No. 4 (1968), 423-443.
Freund, J. The Sociology of Max Weber. New York:
Vintage Books, 1969.
Galtung, J. "Institutionalized Conflict Resolution,"
Journal of Peace Research, Vol. 2, No. 4 (1965),
348-51.
Garner, W. Uncertainty and Structure as Psychological
Concepts. New York: John Wiley & Sons, Inc.,
1962.
Gouldner, A. Patterns of Industrial Bureaucracy. New
York: Routledge & Kegan Paul, 1955.
Gross, B. The Managing of Organizations. New York:
The Free Press, 1964.
Harris, T. 11m OK — You1 re OK. New York: Avon Books,
1967.
Honigsheim, P. On Max Weber. New York: The Free Press,
1968.
House, R., and Rizzo, J. "Role Conflict and Ambiguity
as Critical Variables in a Model of Organiza
tional Behavior," Organization Behavior and
Human Performance, Vol. 7 (1972), 114-119.
Kaplan, A. The Conduct of Inquiry. San Francisco:
Chandler Publishing Co., 1964.
Katz, D., and Kahn, R. The Social Psychology of Organ
izations. New York: John Wiley & Sons, Inc.,
1966.
159
Kilby, P. Entrepreneurship and Economic Development.
New York: The Free Press, 1971.
Lewin, K. A Dynamic Theory of Personality: Selected
Papers. New York: McGraw-Hill Book Company,
1935.
Mack, R., and Snyder, R. "The Analysis of Social Con
flict," Journal of Conflict Resolution, Vol. 1,
No. 2 (1957), 212-248.
MacLeod, R. "The Phenomenological Approach to Social
Psychology," Psychological Review, Vol. 54 (1947),
193-210.
Mannheim, K. From Karl Mannheim. New York: Oxford
University Press, 1971.
Maslow, A. Toward a Psychology of Being. New York:
Van Nostrand Reinhold Co., 1968.
McEachern, A., et al. "The Juvenile Probation System:
Simulation for Research and Decision-Making,"
American Behavioral Scientist, Vol. 11, No. 3
(1968), 28-46.
Mill, J. A System of Logic. New York: Harper & Row,
1874.
Miller, G., et al. Plans and the Structure of Behavior.
New York: Henry Holt & Company, 1960.
Miller, J. "Toward a General Theory for the Behavioral
Sciences," American Psychologist, Vol. 10 (1955),
517-522.
Monod, J. Chance and Necessity. New York: Vintage
Books, 1972.
Probation Department, County of Los Angeles. Information
Series No. 8, Probation Camps, 1971, 1-11.
Raiff, H., and Luce R. Games and Decisions. New York:
John Wiley & Sons, Inc., 1967.
Russett, C. The Concept of Eguilibrium in American
Social Thought. New Haven: Yale University
Press, 1966.
160
Schein,
Schmidt
Schon, !
Simon, :
Sisson,
Sorel, <
Stephan,
Tribus,
Wald, A
Waldo, ]
Weber, I
Wiener,
E. Address, Division 14, American Psychological
Association, September 5, 1971.
S., and Kochan, T. "Conflict: Toward Conceptual
Clarity," Administrative Science Quarterly.
September 1972, 359-370.
> . Displacement of Concepts. London: Tavistock
Publications, 1963.
[ . Models of Men. New York: John Wiley & Sons,
Inc., 1959.
The Shape of Automation. New York: Harper &
Row, 1965.
R., and Ackoff, R. "Toward a Theory of the Dy
namics of Conflict," Papers. Peach Research
Society (International), Vol. 5 (1966), 46-61.
5 . The Illusion of Progress. Berkeley: University
of California Press, 1969.
F., and Mischler, E. "The Distribution of Par
ticipation in Small Groups: An Exponential Ap
proximation," American Sociological Review, Vol.
17 (1952), 82-88.
M. Rational Descriptions, Decisions, and Designs.
New York: Pergamon Press, 1969.
Statistical Decision Functions. New York:
John Wiley & Sons, Inc., 1950.
> . The Administrative State. New York: Ronald
Press, 1948.
h The Protestant Ethic and the Spirit of
Capitalism. New York: Charles Scribner's Sons,
i93o:------
N. The Human Use of Human Beings. London:
Eyre & Spottiswoode, 1950.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The Image Of The Municipal Service: A Study Of Occupational Values Of Professional Personnel

PDF

Role Conflict Resolution In Complex Social Systems: A Polyarchical Systems Model For Organizational And Social Change

PDF

An Evaluation Of A Communal Organization

PDF

The Administrative Process And Decision-Making

PDF

A Systematic Method Of Personnel Evaluation: Appraising The Job Performance Of Instructors In Higher Education By The Use Of Decision Theory

PDF

Pay Utility In The Public Sector

PDF

Management Of Change In Organizations--A Process Analysis

PDF

Public Policy And Dysfunctional Administrative Mechanisms: The Example Of drunk Drivers And Traffic Accidents

PDF

The Implementation Of Management By Objectives In Public Sector Organizations

PDF

Authority Behavior: An Empirical Approach To Comparative Public Administration

PDF

Change Strategy And Client System: Administrative Reform In Brazil

PDF

An Analysis Of The Role And Communications Dynamics Of The Assistant-To The City Manager

PDF

Juvenile Delinquency: Linkage Between Maintenance And Precipitating Variables As Part Of A System

PDF

Performance Improvement Through Self-Contracting

PDF

Community Economic Development And Upward Social Mobility Of Black Students: A Strategy For Social Change

PDF

Managerial Attitudes As A Variable In Professionalism

PDF

Autonomic and cognitive indices of semantic conditioning and generalization

PDF

An Examination Of Possible Changes In Authoritarianism, Values, And Cognitive Complexity, With Their Implications For Business

PDF

Law Enforcement And The Juvenile: A Study Of Police Interaction With Delinquents

PDF

A Study Of Problems In The Control Of Depreciation Basis By The Government

Asset Metadata

Creator Stubbs, Roy Charles (author)

Core Title Alternatives In Research Design: The Measurement Of Goal Conflict

Contributor Digitized by ProQuest (provenance)

Degree Doctor of Philosophy

Degree Program Public Administration

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,Political Science, public administration

Language English

Advisor Storm, William Bruce (committee chair), Klein, Burton H. (committee member), McEachern, Alexander W. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-845833

Unique identifier UC11356513

Identifier 7421511.pdf (filename),usctheses-c18-845833 (legacy record id)

Legacy Identifier 7421511.pdf

Dmrecord 845833

Document Type Dissertation

Rights Stubbs, Roy Charles

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA