THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA 15
A SYSTEM DYNAMICS APPROACH TO
GROUP FORMATION AND PRODUCTIVITY
Bent E. Bakken
System Dynamics Group, Sloan School of Management
Massachusetts Institute of Technol
Cambridge, Massachusetts 02139, USA
ABSTRACT
Based on previous work on Group formation, in particular on the classic "The Human
Group" by Homans (Homans 1950), we develop a System ics model to portray the
growth of two social subgroups within a larger work group. work-group is the people
working in the bank wiring room of the Western Electric facilities in Hawthorne, Illinois as
described in Homans work (op.cit). Working with concepts taken from Cognitive
Dissonance (Festinger 1957) we describe the dynamic interaction pattern of a synthetic
group consisting of five individuals. Depending on hypothetical time constants describing
the participants willingness to change, we get a distinct subgroup formation process before
the development of a common set of beliefs for the entire group or just development of a
“group culture" without the prior formation of the subgroups. Further research is
suggested, notably the inclusion of antagonism in the model. It is also suggested that such
SD models can be used to integrate knowledge of personalities to simulate the impact of a
person on the performance and behavior of a work-group.
INTRODUCTION
Organization is a generic term for a purposeful entity of people involving more than one
person. An organization usually includes several subgroups. Also characteristic for an
organization is the existence of a goal shared in some way among its members. A sub-unit
of an organization, where the members know of each other, perceive themselves as a
distinct entity and also is perceived by outsiders as a‘distinct entity is called a group. (In an
organization, very often the individuals are unknown to each other.) There are basically
two competing types of groups within an organization. The one is the work group, which
is formally defined. The other is the social group, most often spontaneously defined and
not necessarily serving the goals of the organization. Characteristic for the latter is the
nary interests and the Closeness in social value systems among the individuals belonging
to the group.
The question of how work and social groups are formed has led to a considerable amount
of research in the last fifty years. Common to a major part of these studies is the
acknowledgement of the group formation process as a dynamic one. Even so, most work
has consisted in explaining the necessary conditions for group formation, and very few
have looked into the dynamics of the formation process. Sirnon (1952) is an exception, but
that work is so formal that mainstream organizational research has not continued along the
mathematical path he describes.
The Psychoanalytic/-dynamic group tradition with its ephasts on the projection of
individual needs onto the group and the importance of different instances of personality
(Coleman & Bexton 1975), does not explicitly deal with the group formation process.
Another class of studies takes the group as given and is mainly concerned with the impact
of existing technology, norms and individual values on luction and other aspects of
group output (Kolodny & Kiggundu 1980, Goodman 1986). Yet another class is
16 THE- 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA
concerned with the boundary of the groups (Alderfer, 1976) and interaction between
groups and subgroups.
The need for further investigation into the group formation process has been outlined by
various scholars. Homans (1950) points out that lack of adequate mathematical tools makes
difficult quantitative analysis of the group formation dynamics. His work is however
"verbally quantitative” in the sense that it repeatedly mentions the constructs of mutual and
circular causality. These features excited Simon (1952) to formulate a model of Homans
tulates, but unfortunately Simon only investigates features that are traditionally focussed
bi conte systems research (notably stability). The formation dynamics are only hinted at in
1s WOT!
In System Dynamics (Forrester 1961 and Richmond 1985) one has made available a set of
tools and a paradigm for analyzing dynamic systems. The paradigm stresses that the
underlying structure is a key factor in understanding behavior patterns and in proposing
structural improvements. Combined with improved software, including graphical .
formulation of system structure (High Performance Systems 1985) one can hope for an
increased interest among social scientists in exploring the dynamics of social processes.
This work is an attempt to do so. In addition to show the potential of System Dynamics in
traditional social science, this study has a more concrete goal: To prepare for future studies
where individuals are structurally identical but parametrically different to explore different
work group combinations in order to improve productivity through use of probable social
group synergies.
Our empirical base is Homans (1950) description of AT & T's Western Electric plant in
Hawthorne, Ilinois and its bank wiring room. Whereas the original text deals with group
sociology and considers the group as the unit of analysis, we take a social psychology
viewpoint where we focus on the processes the single person is subject to. The group is
then a consequence of the perceptions we find at the level of each worker. The unit of
analysis are the 20 cooperating individuals in the work environment. We first look into the
institutional setting before we quantify the dynamic behavior pattern. The simulation model
is explained in two steps; We look into the overall structure leading to the behavior mode
and then we describe the formulations. At last we run partial model tests. Later we show
complete simulation runs before we conclude with suggestions for improvements-and
further research nc
INSTITUTIONAL AND ORGANIZATIONAL SETTING
The mentioned Hawthome facilities were some of the largest telephone facilities in the
world when the Harvard-initiated studies took place in the 1930's. Thousands of blue-
collar workers provided an ideal background for organizational research. The initial
purpose of the studies was to investigate to what extent productivity depended on different
task arrangements, in particular physical conditions (light, temperature), reward structures
and benefit schemes. Various tasks were taken out of the ordinary assembly lines and
parallel, investigatory, production lines formed, The famous "Hawthorne effect" is named
after the fact that no matter how the physical environment, the groups taken out of the
normal production lines outperformed the ones remaining. This finding led to a steep
decline in the "Scientific Management" organizational paradigm prevailing at that time and
the rise of a new paradigm focussing on human motivation and interaction; “Human
Relations”.
The bank wiring room consisted of twenty male workers divided into two interdependent
work teams. A couple of quality controllers spent most of their time within the group,
although they formally belonged to another department. The tasks performed required little
THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY, CHINA 17
brainwork but a certain skill was needed to wire copper threads on a spool, or to do various
various assembling and inventory tasks. Both teams performed both kinds of work. There
was a dependency of the group upon outside resources, particularly of unfinished goods
for processing by the teams. Also, the output of the activities in the wiring room was used
elsewhere in the production process. As the production process was tightly scheduled,
there were severe constraints on the production process due to shortages of resources.
The homogeneity of the work group is evidenced by the fact that the individuals were all
male second and third generation americans and only one had a college degree. They
ranged in age from twenty to fifty. The social communication among the younger focussed
around matters like girls and booze, whereas the family fathers seemed to talk about more
homely matters. Sports was a common interest to all of them. Both the work they
performed, the language they spoke as well as most other sociological clues makes the
description "working class" very accurate for all the individuals working in the room.
The participants had volunteered for the experiments, and.as such the group was formed
almost from scratch. When Homans entered the group, the social groups were already
starting to cement, so his account of the formation process is not very substantive. But we
must assume his accounts are based upon the social dynamics that partly finished the
cementation and partly was the normal fluctuations around equilibrium levels of
communications,
REFERENCE MODE!
The reference mode describes important system variables over time. First, the flow of
communication between individuals with similar interests and values grow over time to
reach an equilibrium level. Also, the values they have tend to align more and more.The
characteristics of a social group are then a high level of communication between individuals
as well as shared values among its members, whereas non-members of a group have less
communication with the insiders. The two different subgroups would be referred to as
clusters in sociometric terminology. Some individuals in the work group does not belong to
any group. They communicate little, and can be described as outsiders as evidenced by a
marginal position on a sociometric diagram.
Individuals bring different value systems into the groups. Together with organizational
norms and goals these interact to form a subgroup productivity norm, The subgroup norms
are based on the Perceived Norms at the individual level and deal with both Social and
Professional Values. Two distinct subgroups are formed. They are distinguished by the
content of the social communication but also by the values exposed within the
communication. Both the content and the quantity of communication has an impact on the
Perceived Social Norm at the individual level. This perception in turn colors the Values
held by each member of a social group. In an interactive way, this Value is in tumed used
by other individuals in forming their Perceived Social Norm and subsequent individual
Value.
The young and unmarried individuals start communication because of an initial similarity of
values and also because of a common language. Homans talks about the self-reinforcing
cycle of communication between two friends. If two persons have something in common,
! The reference mode is the characterstic historical behavior of the main variables of the
system. This mode serves as a reference to which the model-based simulation results of the
system structure is compared.
18, THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY, CHINA
they will emit signals of liking and therefore they will spend time together. The time spent
together will lead to additional points of commonalty being discovered. He mentions
fishing trips in particular, where physical conditions might induce further closeness. In a
work environment a similar sequence of activities promoting friendship and social group
formation is assumed to take place.
Homans work (op.cit.) provides no explicit mention of how long the social grouping
requires to take place. Implicit, there are however some clues. The process is not
considered a long one, and we have therefore assumed that the main dynamics take place
within the first year. The following three years are assumed to contain only minor
adjustments compared to the initial ones. The variables are chosen because of their
pecinecoe to the investigated system. They do not correspond directly to the concepts used
yy Homans, but as we shall see - are able to make the connection to social psychology and
sociology theories developed later than his classic work, The main variables are portrayed
in figure 1, and they are:
a) The amount of Social Communication (measured in hours/day). This variable grows
over time as people after an initial search period find out with whom they share interests,
and thereafter develop relationships with those individuals.
b) The Social Values held at the individual level (measured along a one-dimensional axis of
liberal-conservative where ultra-liberal holds the value -5 and ultra-conservative holds the
value 5). Homans refers to different age groups spending more time together than do
people across such age groups. There is also a reference to lunch discussion topics
showing that people tend to gather around areas of common interest and.that common
values in those interests is a predictive factor for determining belonging to a social
group. There is an indication of polarization so that people will hold either one Social Value
or another depending on their initial position along the liberal - conservative dimension.
The shared Social Values constitute the subgroup norms that are active in the work group.
In the model we deal with Social. Values (and -Norms) along two dimensions: Degree of
conservatism regarding leisure activities (Social Value_1) and regarding family activities
(Social Value_2).
c) The Social Norms that are perceived at the individual level. This construct is the
processing variable leading to the Social Values mentioned above and is measured in the
same way. A Norm is the subjective evaluation by each individual of the Social Values that
exist in the environment. Internalizing Social Values is not done by a rational weighing of
what is right and what is wrong, but more by a subconscious assimilation of the Norms
that are perceived.
d) The Value System held by each individual. It is also measured along a liberal -
conservative dimension. This construct is similar to the ethnic and family background.
concepts evoked by Homans. Hence the Value System is concerned with many dimensions
not directly relevant to the social group formation process we are describing. The construct
is heavily anchored in childhood and social learning experiences and filters a lot of the
perceptions of the environment. On the other hand, the experiences that the individuals get
from the environment also influences the Value System variable, but the link is quite weak.
e) The Professional Norm perceived at the individual level. This variable is measured in
terms of deviance from the Production Norm set by the organization.
Figure 1 shows the behavior mode, i.e. how the Norms and Values develop over a time
horizon of 2 1/2 years, We are portraying the behavior of person X who becomes a
member of subgroup B. Note that his values change more than does the average of the
THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA 19
group. His Value System is only marginally adjusted towards conservatism.
Communication increases gradually before it settles to an equilibrium value. Subgroup-A is
totally unaffected by individual X.
Serial Commansantion
Newullay Foetal Vala?
Feseee fal Yor Peteen
Vave Sytem:
4 Conservative - Lident FUBoROD As AVERAGE SOCIAL VALUE
ex: PERCEIVED SOCIAL WORT
x: SOCIAL VALUE
x VALUE systiy
BEHAVIOR STRUCTURE DIAGRAMS
‘The notion of behavior structure diagram is a concept closely related to Morecroft's (1982)
policy structure diagram. The difference being that policy diagrams deal wit
management of physical systems whereas we are concerned with mostly unconscious
processes where the related physical structure is regarded as exogenous to the system. But
as Morecroft, we define the metastructure in terms of the major stock networks. In the
behavior structure diagram we deal with the Social Norms 1 and 2, the Professional Nosm
and the Value System.
There is a mutual interaction between the Value System and the perception of the two Social
Norms. The two Social Values are not interacting in the same way, they are simply first
order smooths of the Perceived Norms they are connected to. In tum, the Social Values
determine the Professional Norms in a unidirectional way. We remember that the subscribts
1 and 2 refers to the topics onto which the Norms and Values apply; 1 meaning family
matters and 2 referring to leisure.
The process determining the subgroup formation is identical for each of the five modelled
individuals. Only parametric differences distinguishes the men we.are dealing with. Figure
2 portrays the meta-structure of that process.
PROFESSIONAL NORM PERCEIVED SOCIAL NORM 2
SECTOR ICTOR
VALUE SYSTEM.
SECTOR
IRCEIVED SOCIAL NORM 1
20-THE 1987 INTERNATIONAL CONFERENCE. OF THE SYSTEM DYNAMICS SOCITY. CHINA
The second behavior structure diagram deals with interaction between individuals. Figure 3
portrays the metastructure behind the fact that when people communicate, they tend to
reveal their social values. These values have an impact on the receiver of the message. Not
shown is the fact that with a substantial timelag this Social Value is processed through a
Social Norm to become internalized as the other persons Social Value and as such it is fed
back to the initial transmitter of the communication. As will be shown later the model is
highly simplified in the sense that impact is solely determined by arithmetic differences in
Social Value. It is however easily conceivable that the Values of certain individuals carry
more weight and impact than that of others so that our principle of total reciprocity does not
necessarily hold.
Figure 3 shows the interaction process dealing with three individuals. The full model has
an identical structure but deals with five individuals. Contrary to figure 2, not all the
variables portrayed in figure 3 are stock networks, the impact circles represents graph
functions.
PERSON X PERSON YS
SOCIAL VALUE 1 SOCIAL VALUE 1
PERSON N'S
SOCIAL VALUE |
Figure 4 illustrates the dynamic formulation of the perception of Social Norms. The initial
Social Norm serves as an anchor, and the deeply rooted Value system is a filtering
mechanism in an “Ancoring and Adjustment’ process. The process ensures that the
perceived Norms eventually will equal the Values that exists in the social environment. The
auxiliaries "Impact Y on X" and "Impact N on X" are taken from figure 3. The adjustment
method is related to the exponential smoothing described by Forrester (1961). Moreover,
the process implies that individuals evaluate new information (consciously or
inconsciously) in the light of both previously processed information and the
multidimensional Value System.
This filtering corresponds to the fact that people only slowly change their views on social
questions, and the further the new ideas are from their basic beliefs, the more internal
resistance there will be in adopting the ideas that exist in the environment. Festinger (1957)
THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA 21
describes this theory in cognitive terms, calling it Cognitive Dissonance. He explains that
people interpret their environment according to existing beliefs, and if there is a discrepancy
(a dissonance) they have to reformulate either the perception of the environment or their
previously existing beliefs. Our formulation represents a process whereby both the
environment and existing beliefs are adapting to new evidence.
= —
PERCEIVED SOCIAL NORM
BY PERSON X
(a level variable)
CHANGE IX
PERCEIVED
SOCIAL NORM 1
BY
delay
Ver, tom,
@
Formulation ion of
THE MODEL
The model describes five individuals. The reduction of the 20 workers down to 5 must be
judged in light of our model's general purpose. We do not want to replicate.the entire
behavior in the original book. Instead we want to portray the key dynamics with as few
variables as possible, though retaining similarity with the original case. We are still
following Forrester (1961) stating that “one must focus upon modelling a problem, not a
system". Since we want to establish two social groups and one outsider, we need at least
five portrayed individuals.
In addition to differences in initial values in the stock variables, the individuals are unequal
as regards both their Resistance To Change (in Value System) and the Impact their
communication have on the surroundings. These two factors are portrayed by parametric
means, by non-linear graph functions. As the individual personality and communication
structures are equal across individuals, we will describe the formulation of only one such
structure. The same argument allows us to narrow the description of the two individual
Social Values to one of them.
Our choice of units is degree of conservatism. Contrary to most Industrial and System
Dynamics work this model is free from countable physical units. Therefore, the unit choice
is non-trivial, and our choice is guided by the original work. There we find descriptions of
individual differences along a multitude of dimensions, it seems however as the scale of
conservatism can be taken as an overlying construct coveringall of them. The degree of
conservatism ranges from -5 (ultra liberal) to 5 (ultra conservative) and is the measurement
unit for Value System, Social Value 1, Social Value 2 as well as Perceived Social Norm 1
and 2. Social Communication is an impact construct. It is hence measured by a closely
related unit: Hours/day spent on social communication. The scale ranges from 0 to 2 hours
per day. The Perceived Professional Norms conceptualized as the deviance from the norm
that the organization places upon the workers, it therefore measures relative deviance from
the production norm and is dimensionless. Figure 5 represents the stock -and flow
22 THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA
‘structure of the Value and Norm sectors. It portrays only the formation of Norm and
Values conceming leisure time activities. Family matters' Norm and Values would have
‘been identical.
ice “aMBt- Court
VALUE SYSTEM SECTOR
The Value System
The Value System is a level variable and accumulates changes that are brought upon it from
the outside world as well as from the internalized perception of surrounding Social Norms
in the group. The reason behind it being conceived of as a level and not a constant, (we will
see that the time constant governing its behavior ranges from 5 to 10 years) resides in the
THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA 23
fact that despite its slow change relative to the interesting dynamics, it is conceptually
inconsistent to portray it as a constant.
The Change in Value System
The Change in Value System is measured in terms of influence from conservatism per time
unit. The discrepancies at any time between actual Value System and Perceived Social
Norm are closed within the horizon of the Cognitive Adjustment Time. Other Factors (the
outside world) also influetice the Value System. Through the Value System these factors
eventually influences the Perception of Social Norm of a given individual. Through the
interaction with other group members the Other Factors eventually spread their influence
upon the whole interacting group.
Other Factors
Other factors are condensed into a single exogenous variable. It too has the measurement
unit of degree of conservatism.
Discrepancy between the Value System and the Perceived Social Norm
This is the basic Cognitive dissonance term as explained by Festinger (1957) and it
influences both the Perception of Social Norms as well as the Value System. Its
measurement unit is difference in degree of conservatism.
Cognitive Adjustment time
This variable is measured in terms of weeks. It portrays the fact that the more conservative
people are, the less able they are to change. (That is the definition of conservatism.) The
variable ranges from 520 to 1040 weeks as degree of conservatism goes from -5 to +5. The
graph function (non-linearity) is shown in figure 6.
wey
COGNITIVE :
ADJUSTMENT
TIME
a
=5.000
VALUE SYSTEM
and measured in weeks.
THE PERCEIVED SOCIAL NORM SECTOR.
Perceived Social Norm
24 THE 1987 INTERNATIONAL CONFERENCE OF ‘THE SYSTEM DYNAMICS SOCITY. CHINA
The Norm perception is previously defined. Suffice it here to underline that the change rate
formulation governing its behavior is the core of the model.
The Change in Perceived Social Norm
ChPSoNo;=Dis,.1/CogAT + (SoCo_2,.1/4 + SoCo_3,./4 + SoCo_4,.1/4 + SoCo_5,.1/4 -
PSoNo,.1)/ResToCh
The measurement is in terms of conservatism per time unit. The first term is describing the
impact of the Value System on the Change. The second is averaging the impact of the social
communication from the 4 actors. This average is the goal that the level is searching for.
Any discrepancies between the goal and current state of affairs are smoothed out with the
Resistance to change parameter.
IN THE COMPUTER MODEL, THE SUBSCRIPTS ARE COMPREHENSIVE. THE SOCIAL
VALUES ARE SEPARATED, SO THAT THEY HAVE THE EXPRESSION SO_VA1 AND
SO_VA2 RESPECTIVELY. THERE ALSO IS A RECORD OF WHERE INFORMATION IS
COMING FROM AND WHERE IT IS DIRECTED TO. SO LEISURE TIME SOCIAL
COMMUNICATION FROM PERSON # 1 TO PERSON # 2 HAS THE SUBSCRIPTS: SOCO1_12
Social Communication (impact) variables
These are described in the next section.
Resistance to Change
This construct is conceptually a time constant. It regulates the time it takes for individuals to
adjust to new information. It is a graph function of the Value System. The more
conservative, the more stubbornly one is holding onto ones old perceptions. Figure 7
shows that the parameter goes from 26 to 52 weeks depending on the Value System of an
individual.
THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA 25
Social Value
Social Value is measured in terms of degree of conservatism. This construct is a simple
first-order smooth of the Perceived Professional Norm.
Change in Social Value
This rate measures the speed of change in the Social Value. According to the original case
description, we have chosen a time constant of one year - assuring that after 52 weeks 70
per cent of any initial discrepancy is smoothed out.
THE PROFESSIONAL NORM SECTOR
The Professional Norm is a construct similar to the Social Norm. But whereas the Social
Norm is an active component (in the sense as it determines other variables as much as it is
determined by them), the Professional Norm is not determined in such a circular way. This
simplification of real-life complexity (obviously the Professional Norm also determines
system behavior, but we are purposely omitting these systemic properties.) is to be
explained by Homans' description of the subgroup formation. The process is a function of
social interaction and the social values are the clues to this interaction. There is no mention
of the professional norms playing any role in the subgroup formation. On the contrary,
there is much evidence in Homans work that the informal subgroups to a large extent
precede the professional norms.
The Professional Norm includes a dynamic interaction between individuals on professional
matters. The professional Communication, here portrayed as time-independent, is also
formulated omitting its systemic properties described under its social counterpart. The goal
of the model being to describe the dynamics of the formation process and its impact on
productivity makes it possible to deviate from certain real-life facts.
The Perceived Professional Norm.
This variable is dimensionless and normalized to 0. It then describes any down- or upward
pressure to attain the goals that are put onto the work group from the organization outside.
With a static communication, the variable serves the purpose of smoothing out eventual
discrepancies between the Communication and the Norm.
The Change in Perceived Professional Norm.
Measured in yearly change of the discrepancy, this variable measures the speed of the
smoothing process.
The Professional Communication.
This dimensionless parameter purports to maintain the link with a non-portrayed dynamic
professional communication environment.
The Production
The dimension of the Production is units per week. This parameter enables the modelling
of different individuals disposing different Technologies and Ambitions.
Technology
26 THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA
As the Technology varies, so does production. Hence a Technology constant is included at
the individual level.
Production Demand
The Production Demand is the standard production norm used in the work group. Homans
mentions it is 700 units per day. In a systemic sense, it would have been possible to model
this variable as a'sliding goal, slipping towards actual average production depending on the
organizational culture and the time constants inherent in that culture.
The Ambition
The Ambition is dimensionless and converts the Social Values into attitudes towards work.
It is assumed that individuals espousing conservative goals are more ambitious on
achieving the goals than are the more liberal individuals. The ambition for an ultra liberal is
then to produce 630 units per day (90 % of Demand), whereas the ultra conservative aims
for a daily production of 770(110 % of Demand). This rather arbitrary range is based on
the verbal description in the case, where the quiet and older workers produce more than the
noisy young ones. In addition to Social Values, the skill level or Ambition also influences
Ambition.
The Ability
Skills and intelligence is also determines production. Although not specifically mentioned
in the case, we still know these factors matters. The whole theory behind learning curves is
based upon the fact that skills matter and that they can be improved. Likewise the rationale
behind IQ tests is that IQ matters for performance. +
THE INTERACTION SECTOR
The interaction sector describes the process taking place when individuals are exposed to
each others’ conversation. When there is social interaction; there is a seemingly strong
correlation between closeness in social value content and the time spent and subsequent
impact of that communication. The more people share values, the more time they spend
together and the more impact differences in opinion have on each individuals’ Perceived
Social Norm. This hypothesis is confirmed in a lot of recent social psychology literature,
where communication is set to be a result by either shared interests or forced physical
closeness. But on the other hand, "opposites attract”. So there must be a discontinuity in
the function of time spent together as a function of shared interests. We propose a
formulation where time spent together increases monotonically when difference in Social
Values decreases from 10 to 2. There is however a decrease in time spent together when the
Social Value difference drops below 0. This graph function is shown in figure 8.
(2900,
Time TO
SOCIAL
COMMUNICATION
OIFFERENCE
BETWEEN SOCIAL
VALUES
THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY, CHINA 27
Time to Social Communication
This variable is shown above, and its measurement unit is dimensionless.
The Difference in Social Values
This variable is measured in terms of conservatism, and it calculates the absolute difference
between the Social Values between between 2 individuals.
The Impact of Social Communication.
The Impact of Social Communication is determined by the degree of conservatism revealed
by a Social-Communication boosted by the impact of this communication. Both variables
are described above. Below is shown a graphical representation of the impact formation in
the model.
SoVa2_2 SoVa2_3
PARTIAL MODEL TESTING
The model is a description of a social control theory phenomenon, also often referred to as
homeostasis in biology. In that sense, any disturbances in the system will be smoothed out
28 THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY, CHINA
over time. Since there are no growth processes, we are assured no disequilibrium caused
by positive feedback loops. To test the formulations we first looked into what will happen
if an individual has an initial discrepancy between his Value System and his Social Values.
Given a situation where the Value System and Perceived Social Norms are set to 5 (and
these variables are not dynamically connected to the system) and the Social Values for some
reason start at -5, the behavior resulting from this discrepancy is shown in figure 10. After
210 weeks (4 years) practically all the initial discrepancy between the Social Value and the
perceived Social Social Norms has disappeared.
Figure 11 shows another graph of same run where Ambition grows as the person adopts
more conservative Values. Production follows the same pattern.
ines twa, a vo sin
— er
= 1s
ss cy ete
: cr)
+ a epepenene
mend x 4 RG
oo 4adoo 128000 var $0, 730 900 22 a 0 12800 ox" 70 one
S oi
The interface with the interaction module was then included in the model by letting a neutral
person interact with a conservative one. The neutral person slowly adapts the conservative.
The reason this happens is to be found in the lack of feedback from the initially neutral
individual back to the interaction process. Figure 12 shows this process, where the Social
Value is a smooth of the Perception of Social Norm.
THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA 29
MODEL RUNS
The whole model includes, as we have seen, the five individuals, each with two separate
Social Values and one Value System as well as a Professional Norm. To each individual is
also assigned a five-person interaction sector. The initial setting allows for all Values and
norms to be consistent for each individual. If the Value System is 4, then the initial value of
Norms and Values are also set to 4. In the first run, figure 13, persons 1, 2, 3, 4 and 5 are
characterized by initial levels of 5, 4, 0, -4 and -5 respectively. The figure shows that two
subgroups are formed, but also that they converge into one by the fourth year.
Interestingly, the "loner" becomes part of the final group.
Compared with the reference mode in figure 1, we see that the model run allows a group
norm to be established. The difference can be explained in two ways. Either that the
empirical base did not enable us to see long term (5 years) behavior, so that the description
in Homans book was inaccurate. A more plausible explanation, however is to be found in
the lack of bifurcationary elements in the model structure. Such elements would be able to
assure the development of antagonism, and hence also of "negative communication"
between the separate sub-cultures.
Figure 13 should be interpreted as the development of a Corporate Culture. The concepts of
shared norm and culture are very related. Bur whereas our definition of a norm deals with
one dimension of interest (Leisure or Family Activities), the culture deals with a multitude
of dimensions. The culture construct then becomes more ill-defined than the norm, but its
impact on productivity is probably more important than the Norms mentioned in our model.
A more refined model would have to take the culture construct into consideration in a
structural way. Another construct of interest, which is omitted is the intergroup conflicts
and its impact on productivity.
Due to the negative feedback structures, and personal symmetries, the equilibrium degree
of conservatism would be equal to the arithmetic mean of the Values held at the individual
levels. However, since conservative individuals are more stubborn and have longer
adjustment times than liberals, the final Values are more conservative than that.
Sco
Eee SOGLAL-VALUE-
oa Ae
13
ee Slee,
SOC: Ale MALI 2
2.500
oo
{seceai VALUE... 4s
am tf ge
Pe SOCIAL VALUE &
ge
-soc0 Lee a —_
shown,
Figure 14 portrays a situation where people are pleclustered in only 2 subgroups.
Interestingly, the fact that person 3's individual Values are changed from 0 to 4.5, makes
30 THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA
no perceptible difference on the group formation pattern. The culture development pattern
remains, : 5
5.000
2.500
0.0
5.000
Figure 15 shows the slow change of the Value System with the latter initial levels. Not
shown is individual 5, starting with initial Value of -4. It should be noted that the Other
Factors are equal to the levels initial degree of conservatism. So there are two opposing.
forces influencing the Value System. Other Factors push the Value System towards its
initial Value, whereas the Social Values push the Values towards the perception of the
environment. .
8.000
2.500
oo
+2.500
250.000
One approach to improve the fit between figure ! -the reference mode- and figure 13 would
be to strengthen the influence of the initial Value System level. This would assure a slower
development of the corporate culture, but at the sarne time unfortunately also prohibit the
subgroups to form. In figure 16, the Cognitive adjustment time changing the Value System
is lengthened to 20 years and it is made uninfluenced by degree of conservatism. The
Cognitive adjustment time influencing Perceived Social Norms is redyecd to 2 years on the
THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA 31
other hand. In figure 17 we have rerun 16 with an adjustment time regulating Perceived
Social Norms reduced to 1/2 year. Especially figure 16 shows a closer fit to the reference
behavior. Note that the initial Values are set to 5, 4, 4.5, -4 and 5 respectively and that
individual 3 is not shown.
5.000
00
5.000
25200 137 00 25000
Ei A is itl i from Val
32 THE 1987 INTERNATIONAL CONFERENCE OF THE SYSTEM DYNAMICS SOCITY. CHINA
IMPLICATIONS AND FURTHER WORK NEEDED
Group formation is a complex issue. We have explained and formulated one case of group
formation with a System Dynamics model. In doing so, we have intergrated cognitive
concepts and a specific description. It seems, however, that our findings have general
validity also outside our specific case. The group formation is the core of human socializing
and society emergence, and if we substitue individuals for groups or organizations, we
would be able to explain higher order social aggregations as well. With added antagonism,
our model should also be able to explain coalition formation.
The lack of covergence between Warsaw-pact and NATO countries might also be explained
with an improved model. In such a model the antagonism formulation could build upon the
notion of “diabolic enemy images" (White 1984).
Further work should, in addition to a polarization/bifurcation structure, also include a more
specific understanding of the relationships between specific values and productivity. Also
lacking in our paper is a good statement of communication patterns between individuals
with different Values; Who is submissive and who is dominant. We nevertheless hope that
our work can be helpful in that future research as well as we hope that our model can be
used in practical settings to make policy analysis of how individuals should be put together
in groups to ensure high productivity.
REFERENCES
Alderfer, C D “Boundary Relations and Organizational Diagnosis” in
Meizer & Winkler (ed) Humanizing Organizational
Behavior, Charles Thores, NY 197
Coleman & Bexton (eds.) "Group Relations Reader”, Sausalito, Grex 1975
(Chapters 2,3 & 4)
Festinger, L "A Theory of Cognitive Dissonance", Row Peterson,
Evanston, Ill 1957
Forrester, J W “Industrial Dynamics", MIT press, Cambridge, Ma 1961
Goodman, P S "Asessing Organizational Change", John Riley and Sons,
NY 1986
High Performance Systems "STELLA users manual", Hanover, NH 1985
Kolodny, H F "Towards the Development of a Sociotechnical Systems
& Kiggundy, MN Model in Woodlands Harvesting", Human Relations, Vol33
Number 9, pp 623-645 1984
Morecroft, JDW "A critical view of Diagramming tools for conceptualizing
Feedback Systems Models”, Omega1982
Richmond, B “Bringing System. Dynamics to the Other 99 %" in
Proceedings from the 1935 System Dynamics Conference,
University of Nebraska, 1985
Simon, H “A Formal Theory of Interaction in Social Grou
American Sociological Revue 17, pp 202 - 214, 1952
White, R K "Fearful Warriors: The Psychology of Superpowers", 1984
