Formal Organization: Dimensions of Analysis

Author(s): Peter M. Blau

Source: American Journal of Sociology, Vol. 63, No. 1 (Jul., 1957), pp. 58-69
Published by: The University of Chicago Press
Stable URL: [Link]
Accessed: 13-08-2015 00:59 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at [Link]
info/about/policies/[Link]
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@[Link].

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to American Journal of
Sociology.

[Link]

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

FORMAL ORGANIZATION: DIMENSIONS OF ANALYSIS1

PETER M. BLAU

ABSTRACT
Impersonal controls in formal organizations, such as assembly-line production or evaluation on the basis
of performance records, tend to affect the structure of work groups and reverse the flow of interaction
between superior and subordinates. Three methodological problems are discussed: The effects of group
structure can be isolated by determining the relationship between X and Y for groups while holding constant
the independent variable X for individuals. The quantitative study of complex configurations of interdependent elements involves the internal and external elaboration of a relationship between two major
elements. Dialectical processes characterize organizational change, and these processes may be empirically
investigated by adapting the panel method to organizational research.

It has been only within the last decade or

two that the precise methods of social research developed in interviewing surveys
and in observation laboratories have been
applied to the study of military services, factories, government agencies, and other formal organizations.2Often, however, the research techniques have been adopted without first having been adapted to a new field
of inquiry. Quantification,so important for
providing evidence in support of generalizations, has often producedan artificialatomization of the organized social structures
under investigation. Not that the members
of the organizationare conceived as Robinson Crusoes on isolated islands. Quite the
contrary, the emphasis is all on human relations, but as atoms somehow suspended in
free space. Specifically, human relations are
treated in the analysis as though they were
I I am indebted to Joan W. Moore, who helped
me with the survey of the literature and contributed
important ideas to the analysis, and to the Social
Science Research Committee of the University of
Chicago, which provided funds.
2 The term "formal
organizations" is used here to
refer to social organizations that are formally
established for explicit purposes but to include the
informal as well as the formalized aspects of such
organizations. Herbert A. Simon, Donald W. Smithburg, and Victor A. Thompson define formal organization as "a planned system of cooperative effort
in which each participant has a recognized role to
play and duties or tasks to perform. These duties are
assigned to achieve the organization purpose"
(Public Administration [New York: A. A. Knopf,
1956], p. 5).

attributes of individuals, and the group

structures of which they are component
parts as well as the larger organization of
which these groups are parts are neglected.
These tendencies are the result of a fundamental methodologicalproblem. Since an
empiricalstudy is usually confinedto one or
two organizationsand the investigation of a
large sample of organizationsis hardly feasible, quantitative evidence for generalizations must be based on the observation of
regularities among individual members or
subgroups. But, by treating individuals, or
even subgroups, as independent units of
analysis that can be classified and reclassified accordingto any one of their charactertistics, this procedurenecessarily ignores the
unique constellation of relationships between groups and individuals in the organization-its [Link], on the other hand, the
analysis is focused on the organizedwhole of
interdependent elements, it deals only with
a single case and provides no empirical evidence for generalization, no matter how
many individuals are observed.
This paper is an attempt to explore this
dilemma in the study of formal organizations. For this purpose a secondary analysis
of some research findings will be presented
first, which is largely concerned with the
effect of impersonal mechanismsof control
upon the structure of work groups and the
flow of communication in the hierarchy.
Four dimensionsin the analysis of formalorganization are suggested.

58

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

FORMAL ORGANIZATION
WORK GROUPS ON THE ASSEMBLY LINE

The distinct difference between having

established personal relations with several
co-workersand being a member of a work
group is clearly illustrated in Walker and
Guest's study of assembly-lineworkersin an
automobile plant. Despite the noise and the
fact that hardly any jobs on the assembly
line require co-operation between workers,
most workers have regular social contacts
with a few others stationed near by. Indeed,
over three-quartersof them consider friendly contacts with fellow workers one of the
things they like best about their job.3
Since the workersare strung out along the
line, however, the set of interpersonalrelations of each differs somewhat from those of
everyone else. Tom and Dick are friends,and
both have frequent contacts with Harry,
who stands between them; but Tom also
often talks to three fellows on his right, who
are out of Dick's earshot, and Dick has
friendly ties with two men on his left, whom
Tom hardly knows. There is no common
network of social relationshipsthat unites a
number of workers and distinguishes them
from others by furnishinga socially agreedupon definition of the boundariesof the ingroup. Notwithstanding regular patterns of
informalinteraction, therefore,work groups
do not seem to exist on the assembly line.
This is not merely a matter of arbitrary
definition. In the absence of a sharedset of
social relations and a common boundary,
there is no single groupwith which a number
of individuals can identify themselves and
which, in turn, provides them social support. Perhaps this lack of group support is
one of the reasons why assembly-line workers become so quickly and strongly identified with their union.4 It may also play a
major role in the prevalent dislike of work
and the high rates of turnover and of absenteeism on the assembly line.5 Indeed,
when the same research team in a second
3 Charles R. Walker and Robert H. Guest, The
Man on the Assembly Line (Cambridge, Mass.:
Harvard University Press, 1952), pp. 67-68.
4 Ibid., p.

132.

59

study discovered that work groups sometimes do become establishedon the assembly
line, they also found that absenteeism declined in them.6
The men on the assembly line are divided into sections under different foremen.
Merely having the same foreman does not
give rise to a group structure in the section:
the foreman must help to create it. Most
important, the foreman must identify himself with his men as a group and "think of
himself as also a memberof the group";7 he
sticks up for his men, treats them as equals,
and delegates responsibility to them. These
things make a foreman a symbol of identification uniting the members of his section.
Some foremen, moreover, institute periodic
meetings of the entire section. Of particular
significance is the establishment of "informal systems of job rotation,"8 which not
only reduce monotony and make men more
satisfied with work but also help to create
group [Link] the men in a section intermittently change their positions on the
line, their social situation is no longer very
different from that of other work groups:
each, sooner or later, finds himself close
enough to every one of the others for informal contacts. Opportunitiesfor recurrent
interactionamong all membersof the section
promote a common network of social relationships and a cohesive group.
But how could foremenbecome identified
with the workers in their section and still
discharge their managerial responsibilities?
Walker, Guest, and Turnerargue that a successful foremanmust play a dual role, representing both his men and the management.
It may also be, however, that assembly-line
production itself has a bearing on the problem. The unrelentingmovement of the conIIbid., pp. 62-63, 116-17, 119-20. See also Ely
Chinoy, Automobile Workers and the American
Dream (Garden City, N.Y.: Doubleday & Co.,
1955), pp. 62-72.
6 Charles R. Walker, Robert H. Guest, and
Arthur N. Turner, The Foreman on the Assembly
Line (Cambridge, Mass.: Harvard University Press,
1956), pp. 132-33.
7 Ibid., p. 135.
8 Ibid., p. 134.

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

THE AMERICANJOURNALOF SOCIOLOGY

60

veyor constrainsworkersto a certain output,

relieving the foreman of responsibility for
their productivity. But, of all his duties, it
is only the exercise of control over subordinates that benefits from social distance.
Thus, the fact that the conveyor system substitutes in part for the foreman as a mechanism of control makes it possible for him to
identify himself with the workers without
impeding operations.

To be sure, most of the foremen questioned did not think that assembly-lineproduction facilitates their job. The foreman
still has to supervise quality, keep the line
manned, and tackle problemsof adjustment
and morale inevitable in repetitive work.'
There is quite a difference, however, between the problems the assembly line creates and those the foreman would have to
face in its absence. High rates of turnover
and absenteeism make the training of new
workersand temporaryreplacementsone of
his major [Link] ease the extra
burden the absence of a workerplaces on the
rest of the section, the foremanmust be skilful in redistributing the work load and in
negotiating with managementfor a quick replacement. The foreman must try to reduce
turnover and absenteeism by making the
work itself less arduous and the situation as
satisfactory as possible. In discharging his
responsibility for maintaining quality, he
sees to it that the workers'tools are kept in
good repairand that the materials they need
are delivered to them at the propertime. All
these tasks involve helping subordinates
rather than making demands on them. The
major exception is checking on the quality
of performance,but even the significanceof
such checks is altered by the powerful constraint of the moving line.
IMPERSONAL CONSTRAINTS AND THE
FLOW OF DEMAND

The impersonal constraints exerted by

production-linemethods change the flow of
demand in the organization. The concept of
flow of demand is derived from the concepts
of originationof action and flow of work de9Ibid., pp. 31-32.

veloped by Arensberg,Whyte, and Chapple

in their studies of patterns of interaction
among the members of an organization.10
Whyte shows, for example, that demands in
the restaurant flow not only from management through supervisors down to waitresses and cooks but also from customersvia
waitresses and pantry personnel to the
cooks. The fact that demandsare made from
two different sources often precipitates
problems and conflicts, particularly when
the person asked to do something considers
himself superior to the one making the
request.
Usually, demands flow primarily down
the hierarchy from management through
supervisorsor foremen to workers,although
staff experts provide an alternative route.
The superiordirects operationsby giving his
subordinatesinstructionsand checkingtheir
work. Studies reveal, however, that frequent
and detailed instructionsand close checking
of the subordinates' work is not the best
method of supervision; on the contrary,
such close supervision actually reduces productivity." In other words, the flow of demand down the hierarchy, even if there are
no conflicting streams, seems to impede effective operations.
Assembly-lineproductionreverses the direction of the flow of demand. It is the conveyor that assures co-ordinationand a certain level of productivity, not the directives
10 See especially Conrad M. Arensberg, "Behavior and Organization," in John H. Rohrer and
Muzafer Sherif (eds.), Social Psychology at the
Crossroads (New York: Harper & Bros., 1951);
William F. Whyte, Human Relations in the Restaurant Industry (New York: McGraw-Hill Book
Co., 1948); E. D. Chapple and Conrad M. Arensberg, "Measuring Human Relations," Genetic
Psychology Monographs, XXII (1940), 3-147;
and Joan W. Moore, "Restructuring Demands
in Formal Organizations" (unpublished manuscript, University of Chicago Department of
Sociology, 1956).
11See Robert L. Kahn and Daniel Katz, "Leadership Practices in Relation to Productivity and
Morale," in Dorwin Cartwright and Alvin Zander
(eds.), Group Dynamics (Evanston, Ill.: Row,
Peterson Co., 1953), pp. 617-19. Cf. bibliography
of the empirical studies on which this summary is
based, pp. 627-28.

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

FORMALORGANIZATION
of foremen. And, where an impersonal
mechanism makes most necessary demands
on workers,the major task of the foremanis
no longer to issue directives but to be a
trouble-shooter-to come to his subordinates' aid when they have difficulties.
Hence, the typical interactionis initiated by
a worker's demand for the foreman's help
rather than by a demand by the foremanon
the [Link] reversalis also manifest on
the next level in the hierarchy. It has been
pointed out that staff experts not merely
advise management but, in effect, give
orders to foremen and operators."2Foremen
often feel that staff officials interfere with
their work by making unreasonable demands,"3but the majority of the foremen in
the assembly plant studied considered staff
and service personnelhelpful and felt free to
call on them. When management and experts exercisecontrol by planning assemblyline production, there is no need for issuing
many directives to the foreman,and so most
of the contacts between the foreman and
superiors or staff officials come in consequence of his requests for help.14
The fact that the foreman is cast in the
role of adviser and assistant to his subordinates affects interaction even on the occasions when he makes demands on them. He
could not maintain this role if, upon discovering imperfections,he would curtly ordera
worker to improve the quality of his work.
Moreover, since the foremanknows that the
standardized quality requirementsare generally accepted by workers-he thinks they
want to do a good job-he is likely to lay the
blame for failures upon the changing speed
of the line or the worker's inexperienceand
12 See Victor A. Thompson,
The Regulatory
Process in OPA Rationing (New York: King's
Crown Press, 1950), esp. pp. 430-33.

61

not on lack of effort. Thus demands for improvement are likely to take the form of
guidance and training rather than commands. Workers,finally, are not so prone to
blame the foreman if they have difficulty
meeting standards as they might otherwise
be, because the speeding line absorbs the
brunt of their aggression.
The change in flow of demand engendered
by the constraints of the assembly line distributes discretion more equitably between
superiorand [Link] the demands of superiors are often worded as requests, it is difficult to refuse them-much
more difficult than it is to refuse requests of
subordinates. Demands that flow downward, no matter how polite, control the conduct of subordinatesand restrict their freedom of action. To be sure, the foremanexercises control over workers even if demands
flow upward; his counsel and guidance, in
effect, influencethe conduct of [Link]
when the flow is upward, workers decide
when to call upon the superior;not so when
the flow is downward. The superior continues to exercise considerable discretion
over granting requests for assistance and
over what guidance to furnish, but instead
of monopolizingdiscretion he shares it with
subordinates.
There is a fair amount of evidence that
the exercise of discretion and responsibility
increases satisfaction at work. Thus, Hoppock finds that, the higher the level of skill
and responsibility, the greater is job satisfaction.",Katz and Kahn confirm this finding, and they also show that on a given occupational level individuals whose superiors
permit them to exercise discretion are more
satisfied and less often absent from work
than others.16Feeling free to bring problems
to the supervisor,for one thing, is inversely
related to absenteeism. Coch and French

13
See, e.g., Melville Dalton, "Conflicts between
Staff and Line Managerial Officials," American
Sociological Review, XV (1950), 342-51.

15Robert Hoppock, Job Satisfaction (New York:

Harper & Bros., 1935).

14 The only foreman for whom Walker, Guest,

and Turner (op. cit., p. 91) report quantitative
data initiated fewer contacts with his subordinates
than they did with him, and many more contacts
with superiors and staff personnel than they did
with him.

Kahn, "Some Recent Findings in Human-Relations

Research in Industry," in Guy E. Swanson, Theodore M. Newcomb, and Eugene L. Hartley (eds.),
Readings in Social Psychology (New York: Henry
Holt & Co., 1952), pp. 657, 663-64.

16 Op.

cit., p. 618;and Daniel Katz and RobertL.

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

62

THE AMERICANJOURNALOF SOCIOLOGY

discovered that factory workerswho participate in deciding on a change in production

accept the change more readily than those
without a voice in making it, performmuch
better once the change is made, and are less
likely to quit."7These data support the conclusion that the reversal in the flow of demand, since it increases the workers' discretion, contributes to their satisfaction.
The impersonalconstraints of the assembly line decrease the worker's discretionon
the job, but the upward flow of demand encouragedby these very constraintsincreases
his discretion in his relationship with his
foreman. Hence, assembly-line operations
reduce work satisfaction at one point while
enhancingit at another. This illustrates that
a correlationbetween two "terminal" variables-a formal condition and its ultimate
effect in the organization-can be misleading unless the intervening social processes
are [Link], since it is unlikely
that demand flows upward in all sections on
the assembly line, it is essential to determine
the other conditions in the organization on
which the reversal of flow depends. In short,
to analyze complex configurations,relationships between two variables must be elaborated externally by inquiringinto additional
necessary conditionsas well as internally by
examining intervening variables.

can serve to control qualitative as well as

quantitative standards of performance,
since a variety of errorsand successescan be
counted and recorded.
A study of two government agencies reveals that statistical recordsof performance,
too, reverse the flow of demand between
supervisor and operating officials.'8The direct influence the records exert on the performance of officials and the exact knowledge of accomplishmentsthey furnish make
it unnecessaryfor the supervisorto check on
subordinates frequently and permit him to
let them come to him for advice when needed. In fact, the more a supervisor relied on
statistical records in his evaluation, the
larger the proportion of interactions between him and them that they ratherthan he
initiated.'9 Even when a supervisor talks to
a subordinateabout improving his performance, statistical evidence transforms the
significanceof their conference;what might
have been a much-resentedcritical opinion
becomesan offerof help to make a better record. Evaluation on the basis of a recordalso
makes it possible to give officials considerable discretion in the discharge of their
duties. A quantitative record of performance and a conveyor belt are each impersonal mechanisms of control which, be the setting the semiprofessional work in the two
government offices or the semiskilled work
MULTIPLE CONSEQUENCES AND CHANGE
in the automobileplant, appear to have simIN ORGANIZATION
ilar effects upon interaction between superior and subordinates.
the
is
not
only
production
Assembly-line
Quantitative evaluation has a series of
impersonal constraint. Evaluation on the
basis of published statistical records of per- consequences in an organization. Its introformance is another: precise knowledge of duction in one government agency, for exhow his work compares with others' con- ample, raisedproductivity, improvedthe restrains every employee to try to improve lations between interviewers and their suand so exercises control over operations pervisors, and promoted a detached, imparwithout any direct intervention by superi- tial attitude toward clients. But it also fosors. Indeed, the statistical method of evalua- tered competitive tendencies which intertion is a more adaptable mechanism of con- fered with operations,and, in responseto the
trol than the assembly line. It lends itself to new operating problems, new practices and
being used not only for manual but also for patterns of interaction were developed.20
clerical and even professional work, and it
18 Peter M. Blau, The Dynamics of Bureaucracy
17Lester Coch and John R. P. French, Jr.,
"Overcoming Resistance to Change," Human Relations, I (1948), 512-32.

(Chicago: University of Chicago Press, 1955), pp.

33-40, 101-5.
19 Ibid., p. 232, n. 7.
20 Ibid., pp. 44-68.

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

FORMAL ORGANIZATION

The same aspects of the organizationthat

make essential contributions to operations
frequently also create conflicts and problems, and the unanticipated consequencesof
the adjustments instituted to cope with
them may, in turn, produce further problems. Hence, there is a continual process of
change in the organization.
DIMENSIONS OF ANALYSIS

The methodologicalproblems posed here

may be dealt with by distinguishinga structural, an organizational,and a developmental dimension in the analysis of formal organizations. There is at least one other dimension, which can be called "environmental": the analysis of the relationships between formal organizationsand other social
institutions, for example, of the connections
between the economic or political system
and formal organizations; of the relations
between the organizationof unions and that
of companiesin an industry;or of the role of
the culture in personalconduct in the organization. Obviously, restrictingthe discussion
here to the three intraorganizationaldimensions does not mean that the environmental
one is any less important.
Structural dimension.-The fact that it
makes a differencewhether workers on the
assembly line merely have friendly relations
with several fellow workers or whether a
group structure has developed among them
raises the question of how the distinct significance of social structure can be taken
into account in systematic research. Ever
since the early writings of Durkheim, and
even though he modified his own position
later, it has been recommended that the
study of social structure confine itself to indexes that are independentof the behaviorof
individuals, such as the laws in a society, or
the group's resistance to disruption.2'This
procedure, however, entails the danger of

63

reifying the concept of group structure and

ignoring the fact that it refers to a network
of social relations between individuals which
finds expression in their interaction. In any
case, it is possible to investigate the effects of
social structure by an alternative method.
This method consists of three steps. First,
empirical measures are obtained that pertain to those characteristicsof the individual
members of the groups that have direct or
indirect bearing on their relations to each
other, such as group identification, sociometric choices, initiation of interaction, or
promotions. Second, the measures that describe individuals in one respect are combined into one index for each group, and this
index no longer refers to any characteristic
of individuals but to a characteristic of the
group.22Examples of such group attributes
are the proportion of members identified
with the group, the average number of ingroup sociometricchoices, the degreeof variation in rates of interaction, and homogeneity of interests. Third, to isolate a structural
effect, the relationship between a group attribute and some effect is determinedwhile
the correspondingcharacteristicof individuals is held constant. An illustration will
make this clear.
To test the hypothesis that the free flow
of communication within a work group [Link] performanceof its members, two
kinds of data have been collected in fifty
work groups of about ten members each:
measuresof performancefor every individual and the frequency with which he discusses his problemswith another member of
his own work group. We could investigate
whether frequency of discussion and quality
of performanceare correlated in the entire
sample of five hundred. But, if they were, it
would show only that individualswho readily discuss their problems with others per-

22On the distinction between empirical measures

pertaining to individuals and the corresponding
ones pertaining to groups see Patricia L. Kendall
and Paul F. Lazarsfeld, "Problems of Survey Analysis," in Robert K. Merton and Paul F. Lazarsfeld
Cohesiveness," American Journal of Sociology, (eds.), Continuities in Social Research (Glencoe, Ill.:
Free Press, 1950), pp. 187-92.
LVII (1952), 546-54.

See ?Emile Durkheim, Rules of Sociological

Method (Chicago: University of Chicago Press,
1938), pp. 44-46. For a recent discussion of the issue
see Neal Gross and William E. Martin, "On Group
21

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

THE AMERICANJOURNALOF SOCIOLOGY

64

form better, not that the network of communication in a group influences performance. A second test would be to divide the
fifty groups into those with many and those
with few memberswho readily discuss their
problems and to determine whether average
performancein the first category of groups
is superior. A positive finding in this case,
however, might merely be a reflection of a
high correlation between the individual's
readinessto discuss and his performanceand
thus still would not supply unequivocal evidence of the significance of the network of
communication for [Link], on the
other hand, individuals are first divided on
TABLE 1
PERFORMANCE SCORES BY RATE
AND FREQUENCY OF DIsCUSSION
(HYPOTHETICAL EXAMPLE)
INDIVIDUALSWHO
DIScusS Tm1R
PROBLEMS

Often ..........
Rarely .........

GROUPS MOST OF WHOSE

MEMBERS DISCUSS
TH1iR PROBLEMS
Rarely
Of ten

.85
.70

.65
.40

the basis of their frequency of discussion,

and it turns out that within each category of
individuals about equally ready to talk
about their problems those who belong to
groups where frequent discussion is prevalent perform better than those in other
groups, then it is demonstratedthat the network of communicationitself influencesperformance (see Table 1, where the differences
between columns indicate the structural
effect of the network of communication on
performance).This findingwould show that,
even when the effect of the individual's discussion rate of his problemson his performance is eliminated, just to be in a group
where communicationflows freely improves
performance-other things being equal.
What would account for such a finding,
were it obtained?The fact that an individual
discusses his problems with others has consequencesfor them as well as for himself. He
may get specific advice, and, even when he
does not, the discussion may clarify his
thinking. At the same time the others may
learn something from his discussion which

they can use in their own work, or their selfconfidencemay be raised by his often coming to them for advice. Moreover, the observation that others have problems, too,
and that they feel free to consult one another probably reduces an individual's anxiety over his own problems even before he
starts discussing them.23It is, therefore,hypothesized that ego's discussionsof his problems contribute to the performanceof alters
as well as to that of ego. (Of course, every
member might alternate between playing
the role of ego and that of alter, which
means that reciprocity prevails in discussions.) If this is correct, and only if it is, one
would actually obtain the finding described,
that is, a relationshipbetween the frequency
of discussion in the group and performance
when the individual's rate of discussion is
controlled.
The general principle is that if ego's X
affects not only ego's Y but also alters' Y, a
structural effect will be observed, which
means that the distribution of X in a group
is related to Y even though the individual's
X is held constant. Such a finding indicates
that the network of relations in the group
with respect to X influences Y. It isolates
the effects of X on Y that are entirely due to
or transmitted by the processes of social
interaction.
A somewhat different structural effect is
reported in a study by Stoufferand his colleagues.24Soldierswho have been promoted
have morefavorableattitudes towardchances
of promotion in the army than those who
have remained privates. However, soldiers
in outfits a large proportionof whose members have been promotedhave less favorable
attitudes toward chances of promotion than
others of equal rank in outfits with fewer
promoted members. Thus, the frequency of
promotions in a group has an unfavorable
effect on these attitudes, while the individual's own promotion has the opposite effect.
Being promoted raises the individual's
status, but the promotion of many other
See Blau, op. cit., pp. 105-16.
Samuel A. Stouffer et al., The American Soldier,
Vol. I (Princeton, N.J.: Princeton University Press,
1949), pp. 250-54.
23

24

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

FORMAL ORGANIZATION

members of his group depresses his status

relative to theirs. In short, the promotionof
a numberof egos decreasesthe relative status of alters and increasesthe relative status
of egos. Since ego's X has the opposite implications for alters' Y and for ego's Y in
this case, the structural effect and that of
the correspondingcharacteristicof individuals are in opposite directions.25
Social norms also have structural effects.
Workerswho firmly believe that it is wrong
to be a "rate-buster"are probablyless likely
than others to exceed informal standards of
output. Even workers who see nothing
wrongwith rate-busting,however,may work
slower than they otherwise would if most
members of the group believe rate-busting
[Link] reasonis, of course, that prevailing social standardsare enforcedthroughout
the group and, therefore, influence the conduct of those who do not fully accept them
as well as of those who do. Again, ego's X influences both alter's Y and ego's Y. If the
pressureof the group is successful,however,
one-time deviants will not only conform to
the expectations of the majority but sooner
or later incorporate them in their own
thinking. Once this happens and virtually
all members of some groups condemn ratebusting while hardly any of others do, it is
no longerpossible to use the proposedmethod of determining structural effects, for it
requires a sufficient number of individuals
who reject the norm in groups which, on the
whole, accept it and of individuals who accept the norm in groupswhich, on the whole,
reject it. This indicates an important limitation of the method. It reveals only the present, not the past, effects of the normative
structureof [Link] prevalenceof a normative orientation in a group may have
three effects upon deviants: intensify their
deviant conduct as a reaction to being alienated from the majority, constrain them to
conform against their own convictions, or
convert their very thinking. Although all
25 Another instance: competitive interviewers in
an employment agency produce more than others,
but the prevalence of competitiveness in a group
reduces its productivity (see Blau, op. cit., pp.
61-65).

65

three are effects of the social structure, the

last would not find expression in what has
here been called a structural effect.
A different structural effect is illustrated
by a findingof Lipset, Trow, and Coleman.26
If the membersof a small printing shop are
in substantial agreement on political issues,
they are more prone to be active in union
politics than if there are considerablepolitical differences among them. Not whether
the members of a shop are liberal or conservative but their consensus is significant,
whatever their political opinion. Wide divergences in political viewpoint among the
members of a work group incline them to
avoid political topics of conversation to
avoid arguments. Sufficient political consensus to provide a basis for cordial discourse, on the other hand, encourages political talk at work, and this stimulates interest in the political activities of the union.
In formal terms, if the joint occurrenceof
X, or of non-X, in ego and alter affects Y,
the variance of X in the group will have an
effect on Y which is independentof any possible relationshipbetween the individual'sX
and his Y.
Organizationaldimension.-The structure
of work groups is, of course, profoundly influencedby the formal organizationof which
they are parts. Although the analysis of formally established organizations is generally
concernedwith larger social units than work
groups, the distinction between the structural and the organizational dimension is
analytical, not one of size. To speak of the
interrelations within a social system may
refer either to the social relations between
individuals or groups or to the interdependence of abstract elements in the organization, say, the relationships between personnel policies, supervisorypractices, and interThe term "organizaaction amongworkers.27
tional dimension" is used to denote the lat26 Seymour M. Lipset, Martin Trow, and James
Coleman, Union Democracy (Glencoe, Ill.: Free
Press, 1956), pp. 165-67.
27 For a more general distinction between "part"
concepts and analytical elements see Talcott Parsons, The Structure of Social Action (New York:
McGraw-Hill Book Co., 1937), pp. 30-40.

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

THE AMERICAN JOURNAL OF SOCIOLOGY

66

ter type of analysis, whose focus is the configurationof interdependentelements in the

organization.
As has been pointed out, the establishment of a relationship between two characteristics of organizationsis merelya starting
point for elaboratingit internally as well as
externally. Internal elaboration involves a
search for intervening variables, without
which the finding that an antecedent condition has a certain ultimate consequencecannot be interpretedand may even be misleading. For example,when it was found that the
impersonal control exerted by statistical
records of performance gave rise to more
impersonal and impartial treatment of clients in an employment agency, one might,
on first impression,conclude that impersonality is simply transmitted from the exercise
of control down to the contact with clients.
Actually, complex intervening processes
were responsible for the relationship. Performance records stimulated employment
interviewersto concentrateon makingmany
placements and induced them to eliminate
all considerations that had no bearing on
making placements-in short, all personal
considerations-in their treatment of clients. This disinterested approach often created conflictswith clients, which made it difficult for interviewers to remain detached
and neither become angry at clients nor
modify decisions in order to pacify them.
However, the practice developedof relieving
the tensions generated by these conflicts by
complainingor joking about clients in informal discussions with colleagues. These
friendly, not impersonal,interactionsamong
interviewers rendered conflicts with clients
less disturbing,so that it was easier to maintain an impersonal attitude toward them
even at the risk of conflict.28
In analyzing an organization, the major
independentvariables are the formalinstitutions in terms of which social conduct is organized: the division of labor, the hierarchy
of offices, control and sanctioning mechanisms, productionmethods, officialrules and
regulations, and personnel policies, and so
on. The major dependent variables are the
28

Blau, op. cit., pp. 73-95.

results accomplishedby operations and the

attachment of its members to the organization, as indicated by productive efficiency,
changes effected in the community (say, a
decline in crimerates), turnover,satisfaction
with work, and various other effect criteria.
To explain the relationships between these
two sets of abstract variables, it is necessary
to investigate the processes of social interaction and the interpersonal relations and
group structures.29In dealing with these
patterns of conduct, psychological processes
cannot be entirely ignored. To be sure, the
student of organization is not concerned
with the effects of psychological characteristics but with those of conditions in the organization on social conduct. However, psychological processesare the interveningvariables drawn on to explain why social conditions give rise to certain patterns of conduct.30 Statistical records brought about
more impartial treatment of clients, for example, because they motivated interviewers
to exclude all irrelevant personal considerations from their decisions in making placemerits. In sum, intervening psychological
variables explain why the conditions in the
organization lead to given processes of social interaction, and these social processes,
in turn, must be examinedto account for the
relationships between conditions in the organization and the results they accomplish.
The external elaborationof a proposition
that one factor influencesanother involves a
search for the other conditions necessaryfor
the observed effect. Granted that impersonal mechanismsof control tend to reverse
the flow of demand, on what other conditions does the reversal depend? Concern
with the significance of a combination of
conditions introduces the conception of organization as a configurationof interdependent elements. It is often assumed that the
29 For a method of testing the hypothesis that a
given intervening variable or set of intervening
variables accounts for the relationship between two
variables see Kendall and Lazarsfeld, op. cit., pp.
147-51.
30 This is the main implication of Weber's concept of Verstehen (see Max Weber, The Theory of
Social and Economic Organization [New York:
Oxford University Press, 1947], pp. 87-107).

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

FORMAL ORGANIZATION

67

concept of Gestaltdefies quantitative analy- alizations about the total configurationcansis, but this assumption seems unwarranted. not be supported by evidence collected in
If the concept of Gestaltmeans that the or- only one organization. A second substitute
ganized arrangementof elements in a larger method, which permits making tentative
whole has a significance of its own, not at- generalizations about total organizations,
tributable to the specific character of the might be the secondaryanalysis of a number
elements, then analysis of variance furnishes of empirical studies of formal organizaa method for its systematic investigation. tions.,2
Thus it is possible to test the hypothesis
Developmentaldimension.-Change in the
implied by Weber that administrative effi- organization is the result of the very interciency is the result of a combinationof vari- dependence between elements that is often
ous characteristics in a bureaucracy,31pro- assumedto imply a stable [Link]
vided that empirical data on these charac- if there were a perfect organizationwith no
teristicsand on efficiencycan be obtained for problems,changes in its environmentwould
a large sample of bureaucraticorganizations. soon create some. But internal as well as
Although the empirical measures describe external conditions generate change in the
only the elements and not their configura- organization,since innovations instituted to
tion, the significanceof the latter would be- solve one problem, as already mentioned,
come apparent in the analysis of variance, have a variety of repercussions, some of
for this statistical method would supply in- which are likely to produce other problems.
formation not only on the contribution of
A mistakewe often tend to makeis that the
each characteristic to efficiency but also on world stands still while we are going through
the additional contribution made by their the process of a given adjustment. And it
combination. The so-called interaction ef- doesn'[Link],wemustkeepup with the
fects would furnishquantitative measuresof facts; keepingup with the facts changes the
the significanceof the Gestaltby abstracting facts.... When we think that we have solved
the effects due to the configuration itself a problem,well,by the very processof solving,
from the sum of the effects of its compo- new elementsor forcescome into the situation
nents. While the cost in time and money of and you have a new problemon your hands
such a project is virtually prohibitive, it is to be solved.33
nevertheless important to realize that there
Interdependence entails dilemmas: effiare no inherent obstacles to the systematic cient operation in a large organization deinvestigation of the complex configurations pends on many differentconditions, and the
in formal organizations.
practices instituted to establish one of these
Practical problems, however, cannot be conditions do not remain solely means for
brushed aside. Since it is rarely possible to this end but have implications for others;
establish generalizationson the basis of evi- and, since the conditions required for optidencefrom a representativesampleof formal mum operations are diverse, the measures
organizations, substitute methods have to to improve them are often incompatible.
be developed,one of which might be internal
For this to be fruitful, however, research [Link] great variations existing in cedures would have to be better standardized or, at
large organizationshave not been sufficient- least, more accurately reported than they usually
ly exploited for systematic [Link], are. Research centers that regularly conduct studapparently, by a mistaken notion of Gestalt, ies of formal organizations, such as those at the University of Michigan and at Yale University,
many investigators are concerned with the special opportunities for co-ordinating varioushave
in"typical" foreman or the over-all pattern of vestigations in the interest of deriving generaliza"human relations" in a company instead of tions about organizations.
33Mary Parker Follett, "The Process of Conderivinglimited generalizationsfrom the differences in the organization of the various trol," in Luther Gulick and L. Urwick (eds.),
on the Science of Administration (New
divisions or [Link] any case, gener- Papers
York: Institute of Public Administration, 1937),
32

31 Ibid.,

pp. 329-41.

p. 166.

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

THE AMERICAN JOURNAL OF SOCIOLOGY

68

Thus, effective administrationis contingent

on uniform adherenceto regulations as well
as on adaptability to a variety of specific
situations, but bureaucraticpressurescompelling strict conformity to rules also give
rise to rigidities that interfere with the
adaptability needed to handle special
cases.34Although evaluation on the basis of
accomplished results encourages the responsible performanceof complex tasks, it
simultaneously engenders anxieties which
impede decision-making.35 Assembly-line
methods, while increasingproductivity, lead
to absenteeism and make operations particularly dependent on regular attendance,
to boot. Incompatibility of means, not simply lack of administrative foresight, is responsible for recurrent problems requiring
adjustment in the organizationand thus for
its continual development.
The pattern of change in formal organizations can be describedas a dialectical development. The process of solving some problems while frequently creating others is also
a learning process in which experience is
gained. On the one hand, efforts at adjustment shift from one problem to another as
new difficulties arise when old ones are resolved. After assembly-line production has
been instituted,reductionof absenteeismand
turnover replaces technical questions of coordination as the major area of concern. On
the other hand, as one type of problem recurs, it does not remain the same type of
problem, since cumulative experience
changes the orientation of the members of
the organization toward it. This is so of
problems confronting work groups as well
as of those confrontingmanagement. Work
groupsexperiencedin maintainingsolidarity
against excessive demands of superiorswill
be less threatened by a new and unreasonably demanding foreman than groups that
never had to cope with the problem, just as
management will find it easier to combat
absenteeismif it has successfullydone so on
nrPlrini~ACC ac>.irrns
34 See Robert K. Merton, Social Theoryand Social
Structure (Glencoe, Ill.: Free Press, 1949), pp.
153-57.

These processes of development are further complicated by changes in personnel

which result in the loss of accumulated experience despite the profusion of written
records typical of bureaucratic organizations. Of particular importance is the fact
that knowledge of the networks of informal
relations can be acquired only through direct experience. Gouldnershows that a new
manager's inevitable ignorance of informal
relations constrains him to resort to formal
proceduresin discharginghis responsibilities
even if he is convinced that informalprocedures are more effective.36Similarly, turnover of personnelon lower levels in the hierarchy undermines the cohesiveness of work
groups and threatens informallyestablished
co-operative practices.
Conflicts of interests between management and non-managerial personnel, and
between other groups, are an additional
source of dialectical change. What constitutes adjustmentfor one groupmay be quite
the opposite for the other, since different interests serve as criteria for defining adjustment, and, when issues between workersand
management have been resolved on one
level, new ones on a different level often
arise. After satisfactory working arrangements have been agreed upon, management
introducesnew machines, which then create
new problems of adjustment for workers.
After the union has achieved the right of
collective bargaining, it uses it to raise the
issue of pensions. Indeed, independent of
conflicts between union and management,
the successful attainment of an objective
stimulates efforts to make further improvements and seek new fields to conquer, and
this successionof goals as they are achieved
by more ultimate ones is still another force
that produceschange in the organization.
In sum, dialectical organizationaldevelopments are generated by differentpatterns
of change superimposedupon one another.
35Blau, op. cit., pp. 103-4.
36Alvin W. Gouldner, Patterns of Industrial
Bureaucracy (Glencoe, Ill.: Free Press, 1954), pp.
70-101.

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions

FORMAL ORGANIZATION

The process of adjustment in the organization changes the kind of difficultiesthat demand foremost attention, since new ones
arise as old ones are resolved, and, simultaneously, experience alters the orientation
with which problems are approached.
Hence, as efforts at adjustment are turned
from one problemto another, their effectiveness tends to increase. There are, however,
several different spirals of adjustment of
this sort rather than a single one, because
conflicts of interests between various groupings in the organization produce diverse
conceptions of adjustment. When issues
created by these conflicts are settled, the
conflicting developments continue and new
issues emerge. Thus, the existence of divergent dialecticalprocessesgives rise to yet another dialectical process of adjustment-a
spiral linking the other spirals, as it were.
The complexity of these developments is
further increased by two conditions.
Changes in personnel,which are disruptive,
notwithstanding attempts to preserve continuity through written recordsand formalized procedures,reduce the level of adjustment. And the additional demands made on
the organizationas the result of striving for
new objectives once old ones have been successfully attained create new problems of
adjustment.
The systematic study of these processes
of development requires that the time dimension be taken into account in the investigation of formal [Link] be sure,
there are a number of empirical studies of
change in organizations;for example, Richardson and Walker trace the repercussions
of a change in productionmethods in a factory;37Gouldnerexamines the consequences
of a change in managementfor an industrial
organization;38Selznick analyzes modifications in an organization resulting from its
37 F. L. W. Richardson, Jr., and Charles R.
Walker, Human Relations in an Expanding Company (New Haven, Conn.: Yale Labor and Management Center, 1948).

38 Op. cit.

3 Philip Selznick, TVA and the Grass Roots

(Berkeley: University of California Press, 1949).

69

adaptation to a hostile environment;39and

Michels deals with the changesgeneratedby
the very establishmentof a formal organization.40But the empirical data for most of
these studies were collected at one time,4'
and the patterns of change had to be inferred subsequentlyeither from written records or from other evidence. Although preferable to ignoring change completely, the
procedureis far from ideal.
The adaptation of the panel technique to
the study of formal organizations would
yield reliable evidence on developmental
processes. It would involve systematic observation as well as interviewing in an organization at repeated intervals, perhaps a
year or more apart. Precise indications of
both informal and formal changes would be
suppliedby this method, and intensive interviewing about the changes discovered and
analysis of pertinent recordswould provide
information on the social and psychological
processes leading to them. Although continuous observation for several years, permitting the investigation of changes as they
occur, is preferable to collecting data at
periodic intervals, it is rarely feasible.
Studies based on repeated interviews with
the same respondents have greatly contributed to accurate knowledge about
chanige in opinions and attitudes.42 This
panel design, properly adapted to research
on formal organizations, may well prove
equally fruitful in the systematic study of
organizationaldevelopmentsand, specifically, in testing the hypothesis advanced here
that such developments are dialectical.
UNIVERSITY OF CHICAGO
40 Robert Michels, Political Parties (Glencoe,
Ill.: Free Press, 1949).

41My own study is no exception. The investigation by Jaques should be mentioned as a notable
exception; change in a factory is analyzed on the
basis of continuous observation for two and a half
years, although no quantitative data were collected (see Elliott Jaques, The Changing Culture of a
Factory [New York: Dryden Press, 1952]).
41See, e.g., Paul F. Lazarsfeld, Bernard Berelson,
and Hazel Gaudet, The People's Choice (New York:
Duell, Sloan & Pearce, 1944).

This content downloaded from [Link] on Thu, 13 Aug 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions