STABILITY OF ORGANIZATIONAL STATUS STRUCTURES* Morris Zelditch, Jr. Joseph Berger Bernard P. Cohen This is an expanded version of a paper read at the National Joint Conference of The Institute for Management Science and the Operations Research Society of America at Minneapolis, Minnesota, October 8, 1964. Research for this paper was conducted with the support of NSF grant #G-23990 for investigation of authority and evaluation structures, and a Ford Foundation grant to the Stanford Graduate School of Business for investigation of conflict in Staff- Line structures. The three authors bear equal responsibility for the ideas expressed here, the order of names on various reports of the two projects having been determined by a random device. We would like gratefully to acknowledge the help given us by Bo Anderson, Sanford M. Dornbusch, and W. Richard Scott. May, 1965 Technical Report No. 10 (Revised)
44
Embed
STABILITY OF ORGANIZATIONAL STATUS …wf168gw3603/TR10...STABILITY OF ORGANIZATIONAL STATUS STRUCTURES* Morris Zelditch, Jr. Joseph Berger Bernard P. Cohen This is an expanded version
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
STABILITY OF ORGANIZATIONAL STATUS STRUCTURES*
Morris Zelditch, Jr. Joseph Berger Bernard P. Cohen
This is an expanded version of a paper read at the National Joint Conference of The Institute for Management Science and the Operations Research Society of America at Minneapolis, Minnesota, October 8, 1964. Research for this paper was conducted with the support of NSF grant #G-23990 for investigation of authority and evaluation structures, and a Ford Foundation grant to the Stanford Graduate School of Business for investigation of conflict in Staff- Line structures. The three authors bear equal responsibility for the ideas expressed here, the order of names on various reports of the two projects having been determined by a random device. We would like gratefully to acknowledge the help given us by Bo Anderson, Sanford M. Dornbusch, and W. Richard Scott.
May, 1965Technical Report No. 10 (Revised)
STABILITY OF ORGANIZATIONAL STATUS STRUCTURES
1. Introduction
Sixteen years ago, Hughes could say that
Jobs and departments in an industry are rated by everyone concerned.We expect that. Less attention has been given to the fact that the kind of people hired for a given job determines to some extent the job's prestige. (Hughes, 1949, 120)
He would not have said the same thing in 1965. By "the kind of people
hired" he meant particularly their sex, age, race, and ethnicity. We shall
call such factors external status-characteristics. Partly because of his
own work, partly because of the pervasiveness of the process, there now is
a fairly large, fairly consistent body of knowledge about the wajs in which
external status-characteristics invest washrooms with invidious significance,
or determine sponsorship, career, and the price of labor, or activate ex
pectations, the violation of which creates upheavals in organization.*
The principal kind of finding is that
...if women are hired for a job that only men have done, the men may take the hiring of women, not as proof that women are rising in status, but as proof that the job's status is threatened. We have heard of one industry in which Italians, who had been limited to poorer jobs, were annoyed when Negroes were hired to work alongside them; not because they disliked Negroes particularly, but on the ground that--since they knew what people thoughtof Negroes־■ the hiring of them was additional evidence that management had a low opinion of the Italians (Hughes, 1949, 120-121).
Among the earliest contributions were Collins, 1946 and Hughes, 1946. Since then we have had Dalton, 1950, 1951, 1959; Gardner and Moore, 1955; Hall, 1946, 1948; Homans, 1953; and Hughes, 1949. Related problems are considered by Becker and Geer, 1960, and Gouldner, 1957, 1958.
As a consequence, organizational functions become "segregated" (Hughes,
1949) by age, or sex, or ethnicity; to maintain this structure, discri
mination is practiced in both recruitment and advancement, made possible be
cause both depend on a process of sponsorship largely controlled by the
"inner fraternity" of an organization (Collins, 1946; Dalton, 1951; Hall,
1946, 1948). If the "wrong" people do happen to enter or advance into segre
gated positions, informal rejection restores the system to its usual segre
gated state (Collins, 1946). Sometimes contradictions in status are un
recognized or ineradicable; in that case, one observes conflict and tension
(Dalton, 1950; Whyte, 1948).
Collins' paper (1946) is fairly typical of this literature. He reported
that in a New England factory the managerial jobs were mainly held by Yankees,
first-line supervisors were mainly Irish, and workers were mainly south and
east European . Employees tended to regard this as normal and right. Trouble
threatened when management wanted to appoint a Yankee as foreman of a newly
reorganized janitorial service, trouble not only from Yankees, but from fore
men and workers. It was averted by upgrading the job, calling it "sanitation
engineer" and defining it as managerial.
Behavior of the kind Hughes observed is explained if we assume two
properties of status structures: a balance property, a tendency for two or
more evaluated states to tend to have the same value, as when the executive
washroom is more highly evaluated than the worker's washroom; and a diffuse
property, a tendency for at least some characteristics to be relevant in very
many contexts, as when sex or ethnicity tend to order behavior in many social
■2■
situations. We propose to construct a model of status structures having
these two properties, and apply it to problems of the status structures of
organizations.
The model we will construct is a balance model, in which the central
problems are, first, to decide what entities and relations are sufficient
to describe a given system, and second, to decide what states of the system
should be classified as balanced. These decisions are made in such a way
that one may relate the balance of a system to its stability and the imbalance
of a system to its tensions and pressures to change. A final task is to de
fine the scope of the theory; that is, to define the situations in which it
may be applied.
In applying the model to the status structures of organizations, we
must identify observables of the model with some empirical events, determine
the value of its underived parameters, and determine typical initial and
boundary conditions of the field of application. The most difficult part
of this task commonly derives from the complexity of the world as compared
to the model. What the task really comes to is that concrete empirical
things such as jobs must be analyzed into more simple notions, notions that
permit "jobs" to be coordinated with concepts in the model. Thus in organizing
the presentation of the material, we first formulate the balance property in
a highly general way and, before developing the model further, show its ap
plication to a simple and well-known example which does not yet involve the
diffuse property. Our purpose is both to illustrate the balance idea and to
introduce those new notions necessary to apply it to organizational status
-3־
structures. We then formulate the diffuse property, after which we show
its applications to a problem which does not yet involve the external status-
characteristics with which Hughes was most concerned. This both illustrates
the diffuse property and introduces some new ideas required for applications.
Returning to the empirical uniformity from which we started, the com
monly observed influence of external status-characteristics on the structure
of organizations, our problem is not only to use the model to explain what
is already well-known but also to show that it explains exceptions. Put
another way, the model provides knowledge of how organizations can be con
structed that are not influenced so much by external status-characteristics.
It is easy to find existing organiz ations that satisfy the required con
ditions, though none of them are industrial organizations. But it will turn
out that the conditions are rather extreme. Finally, in summarizing the
model an opportunity arises to show the entire set of conditions it yields
for stability of the structure of an organization.
2. The Balance Property
Consider two actors, p and o, to whom we attribute any number of charac
teristics (such as red hair, energy, insight into human character, or mechani
cal ability) and who desire any number of goal-objects (such as the esteem
of others, money, or a key to the executive washroom). A characteristic will
ordinarily be denoted by the letter C, or, when necessary to distinguish among
two or more of the, by C^, C^...; and a goal-object by GO, or GO^,GO^..,,
where necessary. We look at characteristics and goal-objects from p's point
of view and o, in fact, is not properly an actor at all; o is solely a
referent, an object of p's orientation. Characteristics and goal-objects
are said to have states (such as high and low in the case of mechaoical
ability, or black, brown, blonde, red in the case of hair color; or, if the
type of washroom in an organization is one of its goal-objects, "executive"
and "not-executive" might be its states) . States of characteristics will be
noted by C(x), x =• a, or b, a and b being the different states of a single
characteristic; states of goal-objects by GO(x), In this paper we limit
our discussion to characteristics and goal-objects with only two states; for
example, hair color, the states of which are black, brown, red, blonde,
would be treated as if the states were black and not-black, or red and not-
red.
Among the elements of the theory there are two kinds of relations:
evaluation and relevance. States of characteristics and goal-objects are
evaluated by p if he thinks of some state as better and another as worse.
Observe that this is not the same as seeing some states as high and others
as low; it is possible to devalue the "higher" state of a characteristic, as
some people give a low value to high aggressiveness. Characteristics and
goal-objects are re levant to other characteristics and goal-objects when:
(1) given that p or o possess a state of one, he also possesses or expects
to possess one, but not both states of another; or (2) given that p or o
possesses a specific state of one, he also possesses or expects to possess
2To characterize a system of actors a transition must be made from this p-centric level of analysis. This is accomplished by taking each actor (or class of actors) as p, in turn, predicting the response of each one to the situation in which he and the others are found.
־5־2
a specific state of another. While we use "relevance" to cover both cases,
when there is reason to distinguish them we will call the second strict and
the first not-strict. Thus, where relevance is not strict, p merely under
stands that there is some relationship established between elements, while
in strict relevance p understands also how the two elements are related,
what states are related to what other states. In this paper we deal largely
with strict relevance.
Taken together, these elements and relations form a status structure, S,
from p's point of view. It is completely described if we stipulate what
characteristics are evaluated, what values are given to their states, what
goal-objects are valued, what values are given to tbeir states, and, finally,
what elements are relevant to what other elements.
The most important property the structure exhibits is the balance property
(see Heider, 1946, 1958; and Cartwright and Harary, 1956). By balance we
mean
Definition 2.1. (Balance). Two relevant elements of a status structureare balanced if and only if they have the same evaluation.
Definition 2.2. (Balanced status structure). A status structure is balanced if and only if all sets of relevant elements in it are balanced.
Homans (1953), for example, reported that two positions in a telephone
company billing office, called the ledger clerk and the cash poster, were
differentially evaluated for the skill they required but are paid the same
wage. This example is developed in the next section. All we observe for
the moment is that the characteristic and goal-object are imbalanced (see
Figure 1).
־6־
־7־
Figure 1. An imbalanced status structure. C(x), x = a, b, represents a state of a characteristic; GO(x), x * a, b, a state of a goal-object. Each straight line is an evaluation by p of a state. A "+" sign signifies a positive evaluation, a sign, a negative evaluation. Braces represent relevance as seen by p; that is, a brace from C(a) to GO(b) means that p sees C(a) as relevant to GO(b). Because relevance need not be symmetric, direction of relevance is shown by arrows. (Relevance bonds are always assumed to be positive and, since negative relevance is not defined in our theory, signs are omitted for the relevance relation.) In the status structure shown here, p positively evaluates C(a) and negatively evaluates C(b) and GO(b). Both states of the characteristic are seen as relevant to GO(b) [which would be the case, for example, if it were distributed equally to both]. Because the relevant states C(a) and GO(b) have different evaluations the status structure is imbalanced.
By the "balance property" we mean that balanced structures are stable,
while imbalanced structures create tension and pressures to change in the
direction of balance. Homans' ledger clerks, for example, ought to be up
set and agitate for an increase in wages so as to create an appropriate wage
differential. But we must be given that the states are relevant and dif
ferentially evaluated. Given such conditions,
Assumption 2.1. (Stability). A status structure is stable if and only if it is balanced.
By "stable" here, we mean that the status structure will not change as
a result of any pressures within the structure itself.
Assumption 2.2. (Tension), If it is imbalanced, there is strain generated within the status structure.
Assumption 2.3. (Path). If it is imbalanced, there are pressures from within the status structure to change it in the direction of balance.
The "pressures" referred to in 2.3 need not be understood by actors in
the system as explicit efforts to change the status structure; they are not
even necessarily conscious.
It is important not to misunderstand the implications of the balance
assumptions. We d£ not say that all or most or even any imbalanced status
structure eventually must be balanced. In many empirical systems other
processes often combine to prevent balance from developing. However, the
theory does say that tension will result in such systems as a consequence of
the imbalance.
Like other published balance theories, ours is indeterminate in at
least two ways. It does not predict which of several possible balanced
patterns will be the terminal structure, if any, after a process of change;
nor does it predict which of many possible paths will be the particular path
to that terminal structure.
Finally, note from the way balance is defined that only if elements of
S are relevant do we expect to observe the balance property. If e^ and e
are any two elements of S (where e stands for any sort of state or object)
that are differentially evaluated but not relevant, they may be either the
-9־
the same or different in evaluation; in neither case would there be any
issue of balance or imbalance, hence no issue of stability or instability.
Perhaps e^ is a university professorship and e is high athletic ability;
both of which are presumably evaluated in a university community but not seen
as particularly relevant. Of course, if they can be made relevant; for ex
ample, if in some university appointments come to depend on athletic ability;
then the situation is quite different. If high athletic ability and a pro
fessorship are both positively evaluated to about the same degree in that
university, then they are balanced (Definition 2.1) and hence stable
(assumption 2.1); there will be no pressure from within S to change either
evaluations or relevance relations. If the states that become relevant have
different evaluations (as would be the case if the professorship were more
positively evaluated than high athletic ability), then they will be imbalanced
(Definition 2.1), there will be strain (assumption 2.2), and pressures will
develop to change the structure (assumption 2.3). That is, there will be
pressures from within 3 to change either evaluations or relevance relations.
But in the absence of any relevance relation no such pressures arise. In
the same way, the balance property is observed only if the states of each
of the relevant elements are differentially evaluated. If hair color is,
for some reason, supposed by actors in S to be relevant to handedness--per-
haps black hair is associated with being left-handed--both hair color and
handedness must be differentially evaluated. If either hair color states
or handedness states are not differentially evaluated, there is no reason
to think of the stability or instability of this relation in terms of the
balance theory.
Up to this point we have considered cases where the elements are al
ready differentially evaluated in S and then made relevant to each other.
But what can we say of an element which is not already evaluated in S? What
can we say of an element that is initially not relevant?
First consider an element e. that is not evaliated, and some other ele-lments, e , e^..., that are; and suppose that e^ is made relevant to e , e^...,
Perhaps e^ is a key to an organization's washroom, e is executive status in
the organization, e^ is a high salary, and so on; and perhaps only executives
use the washroom and they all have high salaries. What can we eay about
such a case? Essentially we believe that e^ acquires the evaluations of e
and e, ; that, for example, the key to the executive washroom will become aКpositively evaluated goal-object (see Figure 2). We do not believe this
will happen, however, if to transfer evaluations to e^ would create imbalance.
For example, in imbalance would have been created if e and e^ were two
differentially evaluated status elements in S (e.g., if both executives and
workers had access to the same washroom. In this case we would not expect
e^ to become evaluated either positively or negatively; while it might have
other bonds of value, it would not acquire status value within S (see
Figure 2c).
Assumptions 2.4. (Spread of Value). Let e. be a non-evaluated element of of S and e., e^..., be any evaluated elements of S. If e^ is made“* re levant to e j, e^,..., or e j , ej¿, ..., are made relevant to e¿, e acquires the evaluation of the evaluated elements if and only if ej, e^, ..., all have the same evaluation.
-10-
o'
-11-
p-
Figure 2b. Assumption 2.4 claims that C(a) will acquire the sign of the evaluated object to which it is relevant.
Figure 2a. A characteristic that is initially not evaluated may be made relevant to an evaluated goal■ object. (We use an 0 to represent the absence of evaluation.)
Figure 2d. Note also that if two apparently distinct states both become linked to the same evaluated state, they come to have the same evaluation in the status structure.
Figure 2c. If, however, the nonevaluated state becomes two states that are different in evaluation, it remains non-evaluated.
Now consider an element e^ that is relevant to a second, say e ,
but not to a third, e, . And suppose e. is relevant to e , though initiallyk j ke^ is not. Perhaps skin color is not initially relevant to income, but is
believed to be relevant to executive ability, and executive ability is
relevant to income. In this case, we can say that skin color will become
relevant to income; for we believe that relevance is transitive (see figure. 3)
־12־
Third, consider a structure in which there was no evaluated element; or,
though there were evaluated elements, no element was relevant to any other
element. In neither of these cases would we want to talk of the balance
property; neither relevant non-evaluated elements nor evaluated non-relevant
elements can be said to be balanced or imbalanced.
One field of application of the balance theory is to the distribution
of a wage to positions within the organization. We will use a very simple
model as an illustration. The purpose is to exhibit the meaning of the
balance concepts. No new light on wage allocation is to be expected. Our
reason for choosing this particular illustration will be more evident as we
Assume that we are given the simplest possible organization, one that
Figure 3b. Assumption 2.5 claims that C.(a) will become relevant to GO(a).
Figure 3a. The characteristic state C\(a) is initially not relevant to the goal- object state GO(a), but is relevant to a second characteristic state C.(a) that is relevant to GO(a). ־*
3. Application: Simple model of organizational wage structure.
develop it
consists of only two positions that together work on the single task T.
Suppose that this task has just two outcomes, say T(a) and T(b) . Further
more, which outcome occurs is determined by a single ability, C which has
only two states, C(a) and C(b). (We will say that C is instrumental to T if
possessing C(x) increases the likelihood of achieving one task-outcome and
decreases the likelihood of achieving the other.) Finally, performance in
this organization is rewarded by only one kind of goal-object, GO, having
only states GO(a) and GO(b). Observation of this organization shows us, say,
that the evaluation of C made by actors in S assigns a positive value to the a
state and a negative value to the b state; a similar evaluation is made of
GO, that is, the a state is positive and the b state negative; and, finally,3the state T(a) is the more valued outcome of the task. To summarize: (1)
Two positions form an organization S with a single task T; (2) the task T
has two outcomes, T(a) and T(b) , the outcome T(a) being more valued in S;
(3) which outcome is more likely to occur is determined by a single charac
teristic, C, states of which are possessed by actors in the positions in S;
C has two states, C(a) and C(b), the former having the greater value in S;
(4) a single kind of goal-object, GO, is awarded actors in S, having only
the states a and b, the former of which is more valued in S.
Now suppose that one position is identified with state C(a) while the
other is identified with state C(b). We will compare three wage policies
3While it is convenient, in illustrations to let a always be the positively evaluated state, there is no reason in the theory for this always to be so; b could just as well be the positively evaluated state and a the negatively evaluated state.
-13-
•14־
each of which is a way of assigning goal-objects to the two positions. The
first wage policy, called a uniform wage policy, is to award the same goal-
object, say GO(b), to actors in both positions. The second wage policy,
called differential-imbalanced, is to award the negative goal-object to the
positive state of the characteristic and the positive goal-object to the
negative state of the characteristic. The third wage policy, called dif-
ferential-balanced, is to award the positive goal-object to the positive state
of the characteristic, and the negative goal-object to the negative state
of the characteristic (see Figure 4).
C(a) GO(b) C(a) GO (a)
Figure 4c. Differ* ential-Balanced Wage Policy
Figure 4b. Differential- Imbalanced Wage Policy
Figure 4a. Uniform Wage Policy
If we adopt a uniform wage policy, we have by definition 2.1 an im
balanced status structure. By assumptions 2.1-2.3 such a structure will
create tensions and conflict in the organization and if possible will change
in the direction of balance. The effect of a uniform wage policy will be
either to force states of the characteristic to be more equally evaluated,
or to convince actors in S that the characteristic is not relevant to allo
cation of goal-objects. Managers will regard neither of these outcomes as
desirable ־־perhaps because they have a theory that if state C(a) is instru
mental to outcome T(a), and T(a) is the desired task outcome, the positive
value of C(a) ought to be maintained; and perhaps also because they have a
theory that if GO is awarded for performance of the task, it ought to be
seen by actors in the system as given on the basis of just that characteristic
that determines outcome T(a). What the balance theory claims is that a uni- ״־i
form wage policy has only two stable states, neither of which is desirable
if a manager feels bound by these two constraints.
There is no point in showing that the differential imbalanced wage
policy is not desirable either. The only stable state that satisfies the
constraints we have introduced is found in a system with a differential
balanced wage policy.
Now let us take a further step and illustrate the wage model, in turn,
with a concrete case. Homans (1953) reported on the status structure of a
telephone billing company that had three jobs: ledger clerk, cash poster,
and address-file clerk. The three jobs differed in responsibility, skill,
seniority of the girls who held the jobs, autonomy, and variety. The ledger
clerk was most responsible, skilled, senior, and so on, the cash poster next,
and the address-file clerk least. The address-file clerk was the lowest paid
job, but cash poster and ledger clerk were paid the same wage. Ledger clerks
agitated in their union for a wage increase. Cash posters sometimes refused
what was regarded as a promotion to ledger clerk. Address-file clerks were
not too happy with their low-ranking job, but the other two kinds of clerks
were hostile to management and regarded their wage as unjust while the address-
file clerks did not.
-15-
The illustration is well-known and the solution not in doubt. But
Homans1 billing office does suggest some new issues. The most important
is that organizations, or jobs in them, or the rewards they use, are much
too complex for our theoretical analysis. Though managers and workers may
think of concrete entities such as jobs, shops, or companies, even the sim
plest job has many characteristics, the simplest shop many tasks, the simplest
company many status structures, and the simplest incentive system many kinds
of goal-objects. We must break such compound entities down into simpler
elements. This is illustrated, for example, in the way we have looked at a
"job." A job is seen here as a set of states of characteristics and goal-
objects. Not that there are not other aspects of a job, but they are neg
lected in applications of our theory.
We do not intend to suggest that, with this particular way of looking
at complex entities like jobs, application of our theory always becomes
straight-forward and simple. For one thing, characteristics, particularly
performance characteristics, are typically not directly observable, so that
identification of the characteristics of which a job consists is not at all
straight-forward. Nor is interviewing actors in the system an infallible
method of identifying them. The dimensions usually obtained in such inter
viewing are often themselves compounds, not only of several characteristics
but also sometimes of both characteristics and goal-objects. For example,
in Homans' billing office, autonomy and variety are probably partly goal-
object, partly characteristic. The only fact that eases the problem of
application a little is that such compounds appear (and can be shown to
do so from the balance theory) to behave like simple characteristics.
S'
4. Diffuse Status-Characteristics
Sex, race, ethnicity, or social class are often called status־•
characteristics by sociologists. In this section we will define that term
and attribute to it certain important properties.
We want our definition to accord with what most sociologists mean by a
status-characteristic, though this may be difficult to accomplish since they
have often meant so much. Whatever a status-characteristic is, it is not
something intrinsic to the characteristic itself. An exceptionally light
skinned Negro known to be Negro is treated like a Negro; if he passes he
will be treated very differently, though physicially he may be no different.
That he is a Negro, we conclude, is a matter of beliefs about him, not skin
color itself. But if he i¿ believed to be a Negro, many other beliefs are
often held of him for no other reason than that he is Negro. Some white
employers, for example, believe Negro employees learn more slowly, are less
industrious, less dependable, less intelligent, but better able to stand the
heat, than white employees (Wilson and Gilmore, 1943). Apparently one can
activate this set of beliefs simply by identifying an actor as Negro, with
out direct knowledge of his behavior. The comparison, furthermore, is al
ways invidious; that is, the belief that one is Negro or white, and also all
the beliefs associated with this characteristic, are evaluations as well as
beliefs. It is better to be white, better to be industrious and dependable,
and so on.^
^ Typical discussions of the subject of course see many other correlates of a status-characteristic, such as differential legal privilege, differential reward, differential life-chances, and so on (Cf Hughes and Hughes, 1952). We propose to define a status-characteristic more narrowly and account for these other features of status structures as consequences of our theory about the balance of status-characteristics with other status elements.
-17־
But not only do actors associate specific beliefs and evaluations with
the status-characteristic; often they also generalize from them to judge
actors as wholes. The status-characteristic, we will say, becomes diffuse--
a synonym for global, indefinite, or generalized. Instead of thinking of
"whites" specifically as careful about time, careful about money, industrious,
or whatever is to be associated with this status term, actors within the
system come to think of "whites" as "able", or "moral", or just "superior."
The particular respects in which they are able, or moral, or superior be
comes indefinite; what is paramount is that, if "white," the actor as a whole
is thought of as good. Status terms such as "gentleman" or "peasant," or
similar terms that occur in many status languages, are examples of labels
that tend to carry such diffuse connotations. What we have in mind here has
often been called a "halo" effect.
Formulating these ideas more explicitly: First, let C be any fairly
specific characteristic, such as thrift, or mechanical ability. We assume
that its states are differentially evaluated; that is, one is positively
evaluated while the other is negatively evaluated. A state C(x) may be
attributed to o directly or because he possesses a state of a second charac
teristic, say, D(x) Here we are interested in the latter case. What is
meant is that if D(x) is attributed to o, then an associated state C(x) is
also attributed to o independent of any knowledge of o personally. The
We speak here of p attributing states to o. P also may attribute states to himself, since he is not only an actor in the theory but also a possible object of orientation to himself. To distinguish p as actor and p as object, if it becomes necessary, we may denote the former as p and the latter as p 1 .
-18-
distinction is between the situation in which, knowing o, others say "o is
musical" and that in which, knowing that o is Negro, actors say, "o is Negro,
Negroes are musical, therefore o is musical." Second, we must consider the
fact that there are probably many specific states associated with D(x). Let
be a set of such states. It seems reasonable to require that if one
state of a given characteristic is in the set ) f &t then the other state of
that characteristic is in f (We will say that D(x) is relevant to
meaning that D(x) is strictly relevant to all elements in ^ f •) Third, we
have the idea that some features of f t* * f°r example the evaluations of its
elements, come to be generalized; so that, associated with each state of D
is one of two general expectation states, GES(x), which are attributed to p1
or o on the basis of their state of D. GES(x) might be "competent," "in
competent" or "superior," "inferior," or "gentleman," "peasant," or some
thing of the sort.
Now we define a diffuse status characteristic in S in terms of the ideas,
-19-
Definition 4.1. A characteristic D is a diffuse status characteristic in S if and only if
(1) The states of D are differentially evaluated, and
(2) to each state x of D there corresponds a set } f states of characteristics specifically associated withD(x), such that if C(x) is in y its complement is in¿ V and 6 &
(3) to each state x of D there corresponds a different general expectation state GES(x).
The expression "in S" is important. If S^ is the status structure of
a factory located in a community with status structure S^, it is possible for
D to be a diffuse status characteristic within S. but not in S.. Perhapsi J
within the factory a marked difference is seen between sales and research di
visions, one which generalizes to many apparently non-relevant characteristics
of members of the two divisions, without the distinction becoming significant
in the same way in the community. Many distinctions within organizations do
become the basis of diffuse status characteristics in the community, but how
this comes about must be explained by some explicit assumption.
A common property that we do not want to attribute to D is that it must
be ascribed. Sex, ethnicity, race, the most common examples we have used,
happen all to be ascribed, but status-characteristics need not be in general.
Any status^ will have attached to it some beliefs about actors and these may
be attributed to the status-occupants as characteristics which are activated
in other situations. Coates and Pellegrin (1957), for example show that the
occupational statuses "executive" and "supervisor" are associated, in the minds
of both, with a very long list of specific differentiating characteristics,
such as energy, altertness, aggressiveness, ability to manipulate people,
magnetism, tact, determiniation, and so on. Investigations such as Strodtbeck's
mock jury studies show that such beliefs tend to be attributed to actors not
only in the occupational setting, but also in non־occupational settings
(Strodtbeck, James, and Hawkins, 1957; see the discussion in Berger, Cohen,
Zelditch, 1965).
A status, as distinct from a status-characteristic, is a position in a social system, such as "President of General Motors", rather than a characteristic attributed to persons.
-21־
When q and GES(x) are attributed to actors who possess the state D(x)
in some specific situation, we will say that D is activated in that situation
If D is activated, and C is some characteristic instrumental to whatever task
outcome is required in that situation, we note that there are four possible
ways in which C may be related to D (our purpose here is to fix terminology)
(1) Either states of C are specifically associated with states of D [if D(a)
is "comptroller of a corporation" and D(b) is "shop foreman in a factory",
both probably believe that higher ability at keeping books is associated with
D(a) than D(b)j, (2) Or states of C are similar to states of some other
characteristic that is specifically associated with states of D [if the
comptroller and shop foreman must solve a problem in abstract algebra, they
probably associate the higher state of C with D(a) and the lower state with
D(b) since bookkeeping too is "mathematical'!. (3) Or states of C are
specifically dissociated from states of D--that is, they are explicitly under
stood, as part of beliefs already established about D, to be irrelevant to
its states [if both the comptroller and the shop foreman are on the company
bowling team, probably they understand as part of well-established social
conventions that C is independent of D]. (4) Or, finally, states of C are
not previously associated with states of D, in the sense that no social be
liefs have yet been formed about the relation of C to D [if the comptroller
and shop foreman must write a joint projective story about an ambiguous
stimulus, probably no beliefs about how C is related to D exist at all].
It is important to distinguish cases 3 and 4, for in the former, a system of
specific beliefs does exist about how D and C are related while in the latter
no social beliefs have yet formed about C in relation to D. On the other
hand, in the following sections, cases 1 and 2 will be treated as essentially
the same; we will refer to both of them as specifically associated.
-23־
5. Status classes in organizations
As preparation for the task of relating external status-characteristics
to organizational status structures, we must first consider how jobs are
assigned to levels within an organization. We observe first that Dot only
are particular jobs rated by everyone within organizations, but jobs tend
to be grouped into rather broad classes. Of a small ship for example, Homans
observes:
Like other societies, large and small, the crew of a ship does not consist of an undifferentiated mass of men, but is segmented, largely on the basis of rank and job. In the class structure, if we may call it that, there are three main levels: the commissioned officers, the chief petty officers,...and the rest of the sailors, with a less important distinction among these others between the 1,rated" and the ,,nonrated11 men (Homans, 1962, p. 52).
These classes can be regarded as status classes; that is, as classes
of actors who have the same state of a diffuse status characteristic D.
That these classes are diffuse has been observed, for example, by Gardner
and Moore (1955, pp. 103-116). In organizations, the term "superior," they
have argued, has two senses. One is the chain of command sense, in which a
specific foreman, say, gives orders to a specific worker. The other has to
do with a general extension of the rank of superior, "that does not involve
the right of command," but that "intrudes itself into all sorts of situations
and in innumerable guises," and that makes "all foremen considered superior
to all workers," just as in an army, officers are assumed to be in most
-24-
contexts superior to enlisted men. Furthermore, Gardner and Moore observed ,
there are other such characteristics in an organization, such as the shop-
dffice distinction. How the shop-office distinction functions they illustrated
with the following typical quotation:
Isn't it funny the way office people treat factory people? I don't see any difference between them myself, but the office people think they are so much better than the girls who work in the factories....They seem tothink that factory girls are loud and rough.,.,It just seems that having an office job makes them feel that they're better than we are (Gardner and Moore, 1955, p. 27).
Now, suppose that we have an organization with an already well-developed,
stable, status structure S, in which there are two status classes A and B
[corresponding to the states D(a) and D(b)]. We will consider only organi
zations in which there is a non-empty subset of goal-objects, x^, uniquely
associated with the status class X. By "uniquely associated" we mean that:
in S it is understood that if p1 is in status class X, the goal-objects he
possesses are in the set U"x» a״d, conversely, if p has goal-objects that
are i n V , he is in status class X. In other words, D(x) and ״UT are sym- x *־־* xmetrically relevant. For example, salaries generally reflect the levels
within the organization fairly uniquely. Often the method of payment (hourly
wage, weekly wage, yearly salary) will also reflect them. On the other hand,
the organization may have a system of bonuses that is not associated with
its status classes.
And suppose that a new position is created in S, requiring the ability-
state C(x) in order to perform the task T. Ordinarily the organization will
offer in payment some goal-object already significant in the status structure
S; let us assume that the goal-object awarded for performance of the job is
GO(x) and that GO(x) belongs to which is uniquely associated with the X
class.
How will the new position fit into the existing structure of status
classes? What we will show first is that
5.1. The state of a new characteristic C(x) becomes relevant to a state D(x) of a diffuse status characteristic in S if
(1) a goal-object state GO(x) is allocated to actors possessing state C(x), and
(2) goal-object state GO(x) is a member of a set "UT of goal- object states that is uniquely associated with $(x).
That C(x) is relevant to GO(x) is given. Furthermore GO(x) is a member of
X*/׳x 0^ hypothesis) and w is relevant to D(x) (because they are uniquely ״
associated). Therefore GO(x) is relevant to D(x). If C(x) is relevant to
GO(x) and GO(x) is relevant to D(x), C(x) is relevant to D(x) (2.5).
There are now two possible cases to consider. First, C(x) may be anA . . .element initially not evaluated in S (in which case we will denote it C(x)).
Second, C(x) may be an element that has already acquired status-value by some
process external to and independent of S, so that it is initially evaluated.
In the first case, what status-value (x) acquires will depend solely
on its relations to other elements of S. And the most important of these
will be GO(x), to which the organization itself has established its relevance,
and D(x), to which it is relevant because GO(x) is (5.1 above). Now observe
that the GO(x), D(x) unit is itself balanced (by hypothesis --see above).ATherefore, if C(x) is assigned to a state of D that is opposite in value to
-25-
its relevant state of GO, it acquires no status-value in S (by assumption 2.4)
If it is assigned to a state of D that is the same in value as its relevant
state of GO, it acquires the common value of D(x) and GO(x) (2.4). Hence,
we may say
5.2. C(x) will acquire status-value in S if and only if it is associatedwith that state of D which is balanced with its relevant state of GO.
In the second case, where C(x) has already acquired some status-value,
from the definition of balance and assumption 2.1 we have
5.3. If the state of a characteristic C(x) is associated with a state D(x) of a diffuse status characteristic in S in a balanced manaer, the D,C structure is stable in S.
By the association of the states C(x) and D(x) we mean the individuals
who possess the state C(x) also possess the state D(x) in S, and thus are
members of status class X;^the D,C structure refers to the evaluations
attached to each state x of D and C and the relevance relations between states
and stable means tba t no pressures arise from within the structure itself to
change evaluations attached to states or the relevance bonds which hold
^While the relevance of D(x) and W is symmetric, relevance of C(x) to D(x) is not. Possessing C(x) may imply ihat one possesses D(x) without the converse being true. If two distinct task-roles in S require two distinct characteristics Ci and , we can say that the positive state of each is associated with the positive state of D; we can even say that, if p possesses the positive state of D he will possess the positive state of either C! or Cj, whichever characteristic is instrumental to his particular task-role; but we cannot predict from D alone which characteristic is his, C¿ or Cj. In general associated with each state of D will be a number of distinct characteristic states Ci(x), Cj(x),..., each with the same evaluation, and the relation of D(x) to such states will be one of many.
between states.
If for some reason C(x) becomes associated with a state D(x)v that is its
opposite in value, the states are imbalanced (definition 2.1) and from
assumptions 2.2 and 2.3, we have
5.4. If the state of a characteristic C(x) is associated with the stateD(x) of a diffuse status characteristic in S in an imbalanced manner, there is tension in S and the D, C structure in S is unstable.
Finally: if the D, C structure is imbalanced, one way to restore balance
would be to break the relevance bonds between D(x) and C(x), that is, to
dissociate them. But to accomplish this requires that we also dissociate the
GO(x) state allocated to C(x) from the goal-object states in the set
For if GO(x) is allocated to C(x), and at the same time remains in W - »
attempted dissociation of C(x) and D(x) cannot be stable. C(x) will continue
to be relevant to D(x) because it is linked to a goal-object state G0(x) that
is uniquely associated in S with D(x) (5.1). Hence to become and remain dis
sociated from D(x), it must also be dissociated from all those other elements
in S through which it might remain linked to D.
A good illustration of the process of allocating a new position to a
diffuse status class is a study of marine radiomen by Record (1957). In
1910, while radio was still in its infancy, the United States government re
quired it as equipment on all passenger liners. At first there was great
variation in the way radiomen were treated. On some ships they were treated
as officers, on others as deck crew. This was shown by the kinds of privileges
granted or withheld from them, such as their mess and quarters. The radio
man could not find his proper niche, becoming a "kind of seagoing platypus,
-28-
straddling the great divide between officers and crewmen..." (Record, 1957,
p. 354). But by the 1930's the radiomen had a union capable of winning them
officer status. The union's first effort was devoted to building officer
status into contracts negotiated with employers, in clauses such as:
Each Radio Officer shall receive the same courtesies, privileges, and food uniformly accorded Licensed Watch Officers.
On cargo vessels, the Radio Officers shall eat in the Officers' mess. On passenger vessels, when separate Officers' mess is not provided, the Radio Officers shall be accorded the same privilege of eating in the Salon as is accorded Licensed Watch Officers.
The rooms, facilities maintained therein, the painting and general appearance, condition, and comfort of the room shall be maintained at a level comparable with the quarters of Licensed Watch Officers..." (Record, 1957, pp. 354-355).
Such clauses appear superficially to be allocations of goal-objects to a
characteristic, rather than of characteristics to a status-class. But ob
taining the goal-objects, which were, in any case, largely symbolic, was not
sufficient for the radiomen. They were still legally no better than the
unlicensed common deck sailors, firemen, or stewards. Therefore, they turned
their efforts to lobbying in Congress. From 1937 to 1948, when they were
finally successful, the union sought a bill that would license them as line
officers and would accept nothing less, no matter what compromises were
offered. Since 1948 the status class to which the radioman has been assigned
has remained relatively stable, though Masters and Engineers have not
necessarily accepted the radioman as fully an officer yet.
6. External Status-Characteristics
Organizations are typically part of larger systems (communities, societies)
which themselves have status structures. We will say that, if two systems, g
and Sj have some actors in common, is a subsystem of if and only
if: (1) members of differentially evaluate the states of , a status
characteristic in S^, in the same manner as members of S.; (2) members of
differentially evaluate the states of lj^ of goal-objects associated
with Dj in the same manner as members of S^; (3) at least one goal-object
differentially evaluated in Sj is allocated in S^. The average industrial
concern that pays a wage is an example of a subsystem within a community.
It is not sufficient that be contained ecologically in S , since it might
still carry a wholly deviant culture or have a wholly independent reward
system. A status characteristic in is, from the point of view of its
subsystems, called an external status characteristic. How can one account
for the fact that external status characteristics seem often to be significant
elements in the status structures of subsystems? Is it possible to con
struct.a status structure that is independent of the effect of external status
characteristics?
A . Utilitarian Organizations
First, let us consider an organization in which: (1) Goal-object
GO(such as money) differentially evaluated and allocated in S , is differ
entially evaluated in some larger system of which is a subsystem.
(2) There is a diffuse status characteristic (such as white collar, blue
collar) within S^, states of which are uniquely associated with sets of
goal-object states V", that are differentially evalua ted in S., (3)1 j X X
Members of also hold beliefs about a diffuse status characteristic gSince their status structures are the only relevant features of social
systems analyzed in this paper, we will let the same symbol stand for both systems and their status structures.
־29־
(such as color or sex), significant in the larger system Sj, states of which
are associated with sets of goal-object statesVf. that are differentiallyJ >xevaluated in S . And let us suppose that (4) goal-object state G0(x), dif
ferentially evaluated and allocated in S., is an element in both TJT andX 3. j X
This seems to us a plausible description of the typical organi
zation that pays a wage. Following Clark and Wilson (1961) and Etzioni (1961)
we will call such organizations utilitarian.
Now suppose that in there are individuals who differ in their states
of D^, the status characteristic significant in the larger system. (In such
cases we will say that discriminates between p' and some o in S^.) How
will their differences in states of D. be related to their differences inJstates of D^?
It would seem most straightforward to assume that: if S. employs an
actor p who has the state (a) of an external diffuse status characteristic,
and if in Sj everyone believes that superior intelligence, industry, or
what have you is associated with (a) ; and if, finally, within the organi
zation the members of the A class as supposed by everyone to be superior in
intelligence, industry, and so on; then it will be natural to assign p to
the organization's A class. For, by hypothesis, actors within the organi
zation have the same beliefs about D^.
But we are not given that the various specific characteristics in
are already specifically associated with the external status characteristic,
and for the sake of greater generality we will suppose that initially they
are not. Nevertheless, in terms of our theoretical formulation it can be
shown that
-30-
6.1. An external status state Dj (x) becomes relevant to a status state D^x) in every subsystem of Sj in which
(1) states of Dj discriminate between p1 and o in S , and
(2) GO(x) belongs to U*". x, where x is a set of goal-object states in St that is’uniquely associated with D^(x), and
*(3) GO(x) belongs to U~. , where Tjj~ is a set of goal-objectstates in S. that is,associated with D.(x),J j
Essentially, what 6.1 asserts is that, given different states of Dj for p1
and o, the fact that there are states of GO associated with the different
states of Dj that are at the same time uniquely associated with the different
states of D^, makes the external status characteristic relevant to the states
of the status structure within the organization. For, by hypothesis, Dj(x)
is relevant to a set of goal-object states Xj/I > of which GO(x) is a member.J >xTherefore D.(x) is relevant to GO(x) . But GO(x) is a member also of ,j 1 ,XandXt/^ x is relevant to D^x). Therefore GO(x) is relevant to D^Cx). If
Dj(x) is relevant to GO(x) and GO(x) is relevant to D^(x), then Dj(x) is
relevant to D^x) (2.5).
Furthermore, we have from the definition of balance and assumption 2.1
that,
6.2. If the state of the external status characteristic, Dj (x) is associated with the state of the internal status characteristic, D^(x) in a balanced manner in S , then the D^(x), Dj(x) unit in is stable.
and from the definition of balance and assumptions 2.2 and 2.3 that,
-31-
6.3. If the state of the external status characteristic, D.(x) isassociated with the state of the internal status characteristic,
D^(x), in an imbalanced manner in S , the D.(x), D^(x) unit in is imbalanced, there is tension associated with this unit in
S^, and there will be pressures to change it in the direction of balance.
In other words, if p״ or o, possessing given states of D., come also to
possess given states of D, in S , such that the evaluations of their
and states are the same, there will be no pressures from within the ,
structure itself to change either the evaluations or relevance bonds of
that structure. If p' or o come to possess states of such that the
evaluations of their and states are different, they will experience
strains arising from the D ^ s t r u c t u r e and there will be pressures to
change either its evaluations or relevance bonds.
From the point of view of the organization, perhaps the most important
implication of this argument is that a subsystem cannot entirely control the
conditions of its own stability. It will be vulnerable to imbalance in the
larger system, which very often will be contagious. If, for example, the
status distinctions between Dj(a) and (b) are undergoing change in
while they are being maintained in S^, the D ^ s t r u c t u r e will come in
time to be imbalanced and will continue imbalanced until the status structure
also changes. For example, we might have an organization that allocates
all Negroes to the least valued jobs. But it might be located in a community
that is undergoing great change in race relations, in the process of which
there are all sorts of discrepancies in the way income, residence, style of
life, and so forth are related to being Negro, If indeed the result of this
process is a change in the status structure of the community, the organization
will eventually be imbalanced with the larger status structure. As a result
־32־
there is likely to be agitation for change in the organization itself.
B. Insulation of Organizational Status Structures
־33־
What struck Hughes most about external status characteristics was their
pervasive capacity to penetrate almost everywhere in organizations. Dis
tinctions of sex, color, class, and ethnicity became part of daily inter
action; were embodied in basic status structures; and were part of the funda
mental dynamics of recruitment and career. But there will be many organi
zations in which such a result is highly undesirable; they will want to in
sulate themselves from the effects of a community's stereotypes. Is there
no way in which this can be done?
In our formulation there are three sets of conditions that determine
the relevance of D to Di• The first is simply that
I. (a) there is a status characteristic D. and sets of goal-object states Wj differentially evaluate¿ in S^, and
(b) there is a status characteristic D. and sets of goal-object states ur\ differentially evaluates in S .
In other words S^ and S each have status structures. The second set of
conditions determine that S is a subsystem of Sj, namely
II. (a) the members of S. differentially evaluate the states of D in the same manner as the menbers of S^, and ^
(b) the members of S^ differentially evaluate the sets of goal- object states i j / ' . in the same manner as the members of S., and J J
(c) S allocates at least one goal-object state valued in S .
Finally, a third set of conditions determines the way in which D becomes
related to D^, and is simply the set of conditions set out in 6.1,
III. (a) D. discriminates p and o in S^, and
(b) there is a set of goal-object states ~i>f: x that is uniquely associated with D^(x), and ,x
(c) there is a set of goal-object states t¿ that is associated with Dj (x) , and ־* ,x
(d) there is a goal-object state GO(x) that is a member of both^ l.x and WJ.x•
Now suppose that the first set of conditions is given, namely that there
are status structures in and [1(a) and 1(b)], but that we want to in
sulate the status structure of from that of S^. Four possible policies
suggest themselves. The first two involve fairly revolutionary changes in
the character of the subsystem's relations to the larger system. For one
policy is to simply change in such a way that It is no longer a subsystem
of Sj. This could be accomplished by constructing a deviant subculture in
which 11(a) and 11(b) were not true. The members of would hold different
evaluations of and than other actors who were members of Sj , but not
S^. For example, one might find in a Communist Party cell that race, sex,
and occupation were not significant in its internal status structure-־pro-
viding, that is, that the Party was in fact successful in maintaining deviant
values. A second fairly revolutionary policy would be to alter the basic
status conditions of the organization or the larger system. This could be
accomplished by constructing an independent goal-object strocture, one in
which either 111(b) or III(c), or both, would be untrue. For example, while
the members of would differentially evaluate the sets of the goal-object
states , they would not see them as uniquely associated with ; or per
haps 1׳- could be made independent of D .
-34-
But suppose we wish to insulate without developing a deviant sub
culture and under fairly typical status conditions. We would like, that is,
to break the bond of relevance between and D ,, while nevertheless it is
true that: and UJ־ have status value in [1(a)] ;D and ־ZuT have status
value in S^Iib)]; Si is a subsystem of [11(a) , 11(b), 11(c)], Our result
6.1 asserts that if it is not true that (x) is relevant to D^ix), then either:
(1) there is not a set of goal-object states icK associated with D.(x) inj *x J
S.; or (2) there is not a set of goal-object states ’U /'. uniquely associated J •*־ *xwith D. (x) in S ; or (3) D. does not discriminate p and o in S.; or (4) there* 1 Jis not a goal-object state GO(x) that is in fact a member of both U ~ andJ >xZjJ ' . . But conditions (1) and (2) are common status conditions that we wouldi ,Xnormally expect to obtain in and S^. And if we rule them out; if we assume
that there i¿ a set of goal-object states associated with (x) and a set of
goal-object states uniquely associated with D^(x), then where (x) is not
relevant to D^x), either (3) or (4), or both, must hold. In either
does not discriminate between p and o or there is no goal-object state allo
cated in S. that is a member of both Z j * . and TaT . . If we look at insulated i j,x i,xsubsystems is this in fact true?
One kind of organization that commonly insulates itself is the organization
that socializes, where the principal purpose is to train actors in new norms
and values. Every organization socializes to some extent, but where new norms
and values differ׳ greatly from old norms and values, norms and values into
which everyone is socialized by the everyday operation of family and school,
it often requires a specialized organization to accomplish value change.
-35-
Such organizations include professional schools, rehabilitating prisons,
military training installations (basic training, boot-training, etc.)» a°d
one may add mental institutions, where the purpose is more resocialization
to old conventions than the creation of new ones.
Not all such socializing organizations are effective. Those that are
appear to have certain properties in common (Dornbusch, 1955; Goffman, 1957,
1962; Lifton, 1957; Merton, Reader, and Kendall, 1957; Schein, 1956). In
all of them there is an orderly sequence of statuses, each progressively
higher in rank than the earlier status in the sequence. The socializee enters
at the least valued status, progresses through the sequence, and his goal is
to reach the most valued status. This status corresponds to having learned
the new set of values, and his progress must depend solely on his acquiring
these and on nothing else. Measurement of his progress, furthermore, must
depend wholly on the decisions of an agent of the organization who is al
ready socialized, whose approval the socializee must value. If more than
one socializee enters the system, all newcomers must start equally at the
bottom and any differentiation between them must be wholly with respect to
the norms and values which it is the purpose of the organization to transmit.
While from the point of view of the organization all socializees must
be alike except for differences that the organization itself creates by
training them, from the point of view of the actors themselves they often are
divided by distinctions of sex, race, color, or class. Furthermore, in the
very nature of the organization, what is important is the relevance of what
takes place in the organization to status in the larger society. The various
ways in which actors may be "graded" within S. will be relevant to the kind
of place they occupy outside the organization. We thus seem to have three
conditions that make relevant to D : (1) p' and o are discriminated by
Dj in S ; (2) a state D^(x) (such as "upperclassman" in a military academy)
is uniquely associated in with a goal-object state GO(x); and (3) a state
Dj(x) (such as being "white collar") is associated with the goal-object state
GO(x) in the larger system.
In facing this problem what socializing organizations typically do is
to suppress the external status characteristic, that is, prevent it from dis
criminating among actors in S^. This is done in several ways. Partly it
is accomplished by actual physical isolation of S^, so that members of
will not react to, and hence activate, D^. Partly it is accomplished by a
disengagement, or stripping process, in which everything possible is done to
divest newcomers of any signs of their external status as early as possible.
All their clothes , possessions, even the beliefs that particularly identify
them are if possible taken away. Partly also new members are made uniform
literally, by giving them common clothing. Not all the relevant mechanisms
depend only on what agents of socialization do, incidentally, because often
taboos spontaneously develop among socializees against any discussion of one's
external status characteristic. In the case of indelible characteristics,
like race and sex, which are not so easily made invisible, segregation is used
if selection cannot eliminate differences in D^.
Instead of stripping p and o of their external status characteristics,
one might as an alternative reward them only with local goal-objects. Local
-37-
goal-objects are goal-objects differentially evaluated only within organi
zations like S^. We cannot make the same argument about local goal-objects
that we have for, say, the "wage" in a utilitarian organization. We cannot,
as we have up to this point, say that if D^(x) is a basis within S. of allo
cating GO(x), and Dj(x) is in the larger system associated with GO(x), then
imbalance of with will upset the sense of a "just" distribution of goal-
objects in the larger system. There would be no basis for regarding (x)
as relevant to a purely local goal-object. It is true that for S. to be a
subsystem of S^, it is necessary that at least one goal-object be differentially
evaluated in both. But it is certainly possible to allocate other, local
goal-objects in that are not relevant to D . Thus, in subsystems that
attempt to break the relevance bond between (x) and D^(x) we should find
that there is a particularly strong emphasis on local goal-objects and that
only local goal-objects are uniquely associated with states of D^.
For example: Suppose it is true, as Coleman has argued (1961), that the
adolescent society is distinct from the adult society in its activities and
values. This distinctive society is, Coleman argues, organized in and sym
bolized by the high school. Though it does not symbolize the values of adult
society, as adults might ask it to, the school does symbolize the solidarity
of adolescents and therefore they are very identified with their schools. It
follows, Coleman claims, that athletics are important to adolescents, because
athletic competitions are the only distinctively school identified activities.
Now it seems plausible to suppose, though Coleman does not agrue it, that
athletic ability is dissociated from social class in the larger community;
-38-
and to suppose too, as Coleman apparently does, that interscholastic athletic
prizes are significant primarily to the pupil status system rather:than•the
adult status system. If these assumptions are plausible, then we should.find
that the more important athletic ability to the pupil status structure, the
less important social class in determining the pupil elite. If states of
the internal status characteristic are independent of states of the external
status characteristic,we should expect that goal-objects allocated in and
uniquely associated with are significant primarily in the sybsystem rather
than the larger system. Had interscholastic athletic prizes been among the
goal-objects associated with social class in the community, however, we should
have found (given the other status conditions of 6.1), that position in the9pupil status structure was correlated with class.
7. Summary and Conclusion
We have tried to formulate a set of assumptions that would account for
the relation of an organization's status structure to external status charac
teristics, The two key assumptions in this theory are the balance property,
attributed to status structures generally, and the diffuse property, attributed
to certain status characteristics. What the theory really does is to summarize
the behavior of structures having these two properties. Vie have not tried
to develop many ideas of what mechanisms might restore balance, nor what
symptoms might express strain; but we have tried to state the conditions under
which the status structure of the organization might be expected to change.9In fact, Coleman's data (1961) appear to suggest that the more important
athletic ability to pupils, the less the pupil status structure is determined by social class; while the more important social skills and academic excellence to the pupil status structure, the more it is determined by class.
-39-
-40-
From the point of view of an organization what we have done amounts to
identifying three status structures which are relevant to its stability.
These are:
1. The structure of specific statuses, expressed in our theory by specific performance characteristics and specific goal-objects.
2. The structure of organizational levels or status-classes, expressed in our theory as a diffuse status characteristic operant within the organization..
3. The structure of external status characteristics, significant in the larger community or society of which the organization is a part.
We have seen that the problem of applying our theory to this rather complex
set of structures is one of showing when the various parts become relevant
to each other.
Investigating the behavior of structures which have the balance and dif
fuse properties, our principal results are:
1. Stability of the organizational status structure is achieved only if each status element is balanced with every other relevant status element of the organization. The relevant status elements will typically include external status characteristics, as well as status-classes, specific characteristics, and goal-objects. If external status characteristics are relevant, the organizational status structure is vulnerable to imbalance in the community and society.
2. Only under very narrow conditions is it possible to make the status structure of the organization independent of that of the community and society, without deviating from the larger society or changing its status structure. One such condition is that either differentiation of the external status characteristic naturally does not occur or else is suppressed by the organization. Another such condition is that the rewards relevant to a specific characteristic be independent of those that are linked to the larger system. The latter condition would be difficult to realize in an organization that induces performance by awarding goal-objects valued in the larger society.
These results, of course, will be true only if the assumptions of our
theory are true. That is, we offer them as validly deduced from the balance
and diffuse properties, without comment at this time on their empirical
verification־־though obviously we regard them as consistent with results so
far obtained by empirical investigations. The more difficult problems of
the theory we expect will be resolved only by empirical investigations that
are specifically addressed to testing our theoretical formulations.
־41־
BIBLIOGRAPHY
Becker, H. and B. Geer, "Latent Culture: A Note on The Theory of Latent Social Roles,V Administrative Science Quarterly, 5 (1960), 304-313.
Berger, J., B. P. Cohen, and M. Zelditch, Jr., "Status Characteristics and Expectation States," Laboratory for Social Research Technical Report No. 12, Stanford University, 1965.
Cartwright, D. and Frank Harary, "Structural Balance: A Generalization of Heider's Theory," Psychological Review, 63 (1956), 277-293.
Clark, P. and J. Q. Wilson, "Incentive Systems," Administrative Science Quarterly, 6 (1961), 129-166.
Coates, Charles H., and Roland J. Pellegrin, "Executives and Supervisors: Contrasting Self-conceptions and Conceptions of Each Other," American Sociological Review, 22 (1957), 217-220.
Coleman, J., The Adolescent Society (New York: The Free Press of Glencoe, 1961) .
Collins, 0., "Ethnic Behavior in Industry: Sponsorship and Rejection in New England," American Journal of Sociology, 51 (1946), 293-298.
Dalton, M., "Conflicts between Staff and Line Managerial Officers," American Sociological Review, 15 (1950), 342-351.
_________ , "Informal Factors in Career Achievement," American Journal ofSociology, 56 (1951), 407-415.
_________ , Men Who Manage (New York: John Wiley, 1959).
Dornbusch, S., "The Military Academy as an Assimilating Institution," Social Forces, 33, No. 4 (1955), 316-321.
Gardner, B., and D. Moore, Human Relations in Industry, 3rd ed. (Homewood,111.: R. D.'Irwin, 1955).
Goffman, E., "The Characteristics of Total Institutions," in Symposium on Preventive and Social Psychiatry, Washington, D.C.: Walter Reed Institute of Research, 1957.
Hall, 0., "The Informal Organization of the Medical Profession," Canadian Journal of Economics and Political Science, 12 (1946), 30-44.
_______ , "Stages of a Medical Career," American Journal of Sociology, 53(1948), 327-336.
Heider, Franz, "Attitudes and Cognitive Organization," Journal of Psychology,21 (1946), 107-112.
____________ , The Psychology of Interpersonal Relations (New York: JohnWiley, 1958).
Homans, George C., "Status Among Clerical Workers," Human Organization, 12 (1953), 5-10.
________________, The Small Warship," American Sociological Review, 11(1946), 294-300. Reprinted in Homans, G. C., Sentiments and Activities (New York: Free Press of Glencoe, 1962), 50-60.
Hughes, E. C., and H. M. Hughes, Where Peoples Meet (Glencoe, 111.: Free Press, 1952), Chapter 7.
Hughes, E. C,, "The Knitting of Racial Groups in Industry," American Sociological Review, 11 (1946), 512-519.
Hughes, E. C., "Race Relations in Industry," in Whyte, W. F., Industry and Society (New York: McGraw-Hill, 1946), 107-122.
Lifton, R., "Chinese Communist Thought Reform," in Schaffner, B., (ed.),Group Processes. (Proceedings of the Third Conference of the Josiah Macy, Jr. Foundation) (New York: Josiah Macy, Jr. Foundation Publications, 1957).
Merton, R. K., G. Reader, and P. Kendall, The Student Physician (Cambridge, Mass.: Harvard University Press, 1957).
Record, June C., "The Marine Radioman's Struggle for Status," American Journal of Sociology, 62 (1957), 353-359.
BIBLIOGRAPHY
-3-
Schein, E., "The Chinese Indoctrination Program for Prisoners of War," Psychiatry, 19 (1956), 149-172.
Strodtbeck, Fred L., Rita M. James and Charles Hawkins, "Social Status in Jury Deliberations," American Sociological Review, 22 (1957), 713-719.
__________________, and Richard D, Mann, "Sex Role Differentiation in JuryDeliberations," Sociometry, 19 (1956), 3-11.
Torrance, E. Paul, "Some Consequences of Power Differences on Decision Making in Permanent and Temporary Three-Man Groups," Research Studies, State College of Washington, 22 (1954), 130-140.
Whyte, William F., Human Relations in the Restaurant Industry (New York: McGraw-Hill Book Co., 1948).
Wilson, L., and H. Gilmore, "White Employers and Negro Workers," American Sociological Review, 8 (1943), 698-705.
Zalesnik, A., C. R. Christensen, and F. Roethlisberger, Motivation, Productivity, and Satisfaction of Workers (Cambridge, Mass.: Harvard University Press, 1958).