-1-
TWO-WAY CABLE TELEVISION AND CITIZEN PARTICIPATION:
POSSIBLE MECHANISMS OF MANIPULATION
by
ZALE 11S
B.S., ILLINOIS INSTITUTE OF TECHNOLOGY
(1974)
SUBMITTED IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE
DEGREES OF
MASTER OF SCIENCE IN MECHANICAL ENGINEERING
and
MASTER OF SCIENCE IN POLITICAL SCIENCE
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
(January, 1977)
Signature
Certified
Certified
of Author...
Departments of Mechanical Engineering
and Political Science, Janaury 1977
by ............
Thesis Supervisor, Department of Mechanical
Engneering i
by.......C. - .m. r.,r..C *J.
Thesis Supervisor Department of Political Science
Accepted by........
ARCHIVES
.A6 UG 18 19 77.
-2-
TWO-WAY CABLE TELEVISION AND CITIZEN PARTICIPATION:
POSSIBLE MECHANISMS OF MANIPULATION
by
ZALE ANIS
Submitted to the Departments of Mechanical
Engineering and Polictcal Science on January, 1977
in Partial Fulfillment of the Requirements
for the Degrees of Master of Science
in Mechanical Engineering and Master of Science
in Political Science
ABSTRACT
The purpose of this thesis is to analyze possible consequences
of utilizing two-way cable television technology as a means of increasing
citizen participation in government. Specifically, effects on the
political process are investigated in terms of the problem of audience
manipulation. After an examination of the technical configurations
likely to be employed in large-scale cable participation systems, the
results of a series of laboratory simulations are presented. These
simulations point out several of the characteristics of cable systems
that may eventually serve as a basis for manipulatory effects. Three
specific mechanisms by which such effects may be generated are evaluated
by comparing the results of the simulations to relevant material from
the existing social-psychological literature. It is concluded that the
three mechanisms of 'bandwagon', 'agenda', and 'network' effects, could
all become possiblefactors in any eventual development of manipulation
problems in two-way cable participation systems. Research designed to
specifically investigate the presence of such effects within the context
of actual programming is recommended.
Thesis Supervisors:
Thomas B. Sheridan, Professor of Mechanical Engineering
Ithiel de Sola Pool, Professor of Political Science
-3-
ACKNOWLEDGEMENTS
Many people have contributed their advise, help, and support
to the work of this thesis. Foremost among these are my thesis
advisors, Professor Thomas B. Sheridan of the Department of Mechanical
Engineering and Professor Ithiel de Solb Pool of the Department of
Political Science. In particular, Professor Sheridan deserves special
thanks for his role as supervisor during the course of my work on
the MIT Community Dialog Project. His advice, guidance, and especially
his patience were greatly appreciated.
Among others whose assistance proved invaluable were Edwin
Diamond and Chris Artertin of the Department of Polictical Science,
and Howard Margolis of the Center for International Studies at MIT.
The experimental simulations reported on in this thesis were performed
with the assistance of the Center for Advanced Engineering Study at
MIT. In particular, I greatly appreciate the help given to me by
James Roberts and Norman Lang of CAES. A large part of this experi-
mental work was carried out with the cooperation of four MIT under-
graduates - Jeff Clarke, Marsha Hancock, Michael Lynch, and Glen
Speckert. Their help is also greatly appreciated.
-4-
TABLE OF CONTENTS
CHAPTER ONE: The Problem: Anticipating the Effects
of New Technology.. ......... 6
I. The General Task of Technological
Forecasting................... ... 8
II. Forecasting the Effects of Cable
Technology...................................12
III. Plan of Study........... - - .... 15
CHAPTER TWO: Background: General Knowledge of Relevant
Social Institutions.... ......... 17
I. Citizen Participation in a Democracy........19
II. The Issue of Manipulation.................. 21
CHAPTER THREE: Observation: Simulation of Two-Way Cable
Programming............. 27
I. Focusing on Probable Technology ......... 27
II. Investigating Probable Characteristics
of Cable Systems.... ........... 35
CHAPTER FOUR: Discussion: Implications for the Issue
of Manipulation................... 65
I. Continuous Feedback: Bandwagon Effects... 65
II. Branched Patterns of Discussion:
Agenda Effects..............................74
III. The Structure of Communications:
Network Effects........................... 85
CHAPTER FIVE: Conclusions..........................1......... 101
I. Suggestions for Future Research-...........103
II. Manipulation Effects and the Process of
Technology Assessment......................106
-5-
I. References.................................
IT. Sample Questionnaire Used in Simulations....
APPENDIX:
111
114
-6-
CHAPTER ONE
THE PROBLEM: ANTICIPATING THE EFFECTS
OF NEW TECHNOLOGY
Within recent years a great deal of enthusiasm has been gener-
ated over the prospect of two-way cable television. The technology
for interactive cable is currently available, and pilot systems have
already been introduced. The question of how such systems are to
be used, however, still remains largely unanswered. One possible
application which has received particular attention is the use of
cable technology as a means to increase citizen participation in
governmental decision-making. When we appreciate the fact that cable
technology offers the potential to involve large portions of the
population, it seems apparent that the consequences of any such
citizen participation programs will almost certainly have profound
impacts on society.
This realization has been the source of claims concerning the
great benefits to be had from the implementation of such programs
as well as of ominous predictions of deeply harmful consequences
for society. The problem which this thesis addresses is how to dis-
cover characteristics of the emerging technology which possibly hold
some implication for eventual harmful effects. Specifically, we wish
to focus upon the potential inherent in the technology for covert
manipulation of the audience.
-7-
By manipulation we mean the alteration of group behavior such
that outcomes other than those which would have otherwise emerged
during the course of cable participation programming might occur.
Such alteration of behavior may of course be conscious and intentional.
However, a more likely possibility is that it may arise unitentionally
as a result of systemic factors inherent in the technology of the
communication process. With time, such factors might come to be
exploited intuitively by those skilled in the utilization of cable
technology for public participation.
It is hoped that by identifying possible problem areas, future
empirical reserach might eventually verify the existence or nonexist-
ence of characteristics having manipulatory potential in actual citizen
participation programs conducted over two-way cable. The particular
methodology employed in this analysis has been adapted for the task
of predicting qualitative impacts of technology upon social systems.
For this reason it is necessary to briefly examine the prior problem
of technological forecasting in general.
-8-
I. The General Task of Technological Forecastin
Contemporary society has become aware of the necessity to analyze
new and existing technologies in terms of possible effects upon the
environments in which they are introduced. The complete social setting
within which any technological innovation is implemented consists of
several interelated spheres such as the biological, economic, and
political. The rapid pace of technological advance within the last
few decades has provided innumerable instances of technologies developed
for the purpose of solving a particular problem in one sector, only to
give rise to new problems involving several sectors.
In order to analyze such problems it becomes necessary to make a
distinction between the first- and second-order effects of technologi-
cal change upon the physical and social environment. In the past, the
evaluation of new technological developments in terms of possible
social consequences was often solely concerned with the effectiveness
of the technology as a solution to some immediate problem which had
arisen in society. Assessment of technology on this level, however,
does not extend beyond an analysis of first-order consequences. But
as Raymond Bauer states, the realities of the contemporary world
have forced us to the recognition of the fact that "...even if we did
things that we regarded as inherently desirable, they might produce
second- and third-order consequences that were undesirable, and of
greater ultimate magnitude than the direct consequences of the original
action.f1]
-9-
Bauer's framework for the assessment of technological development
is based upon the assumption that society is sufficiently complex to
insure that any action will inevitably result in some second-order
effects. These effects are not necessarily bad, however; they may
be desirable or undesirable. They can also be intended or unintended,
anticipated or unanticipated. In view of this state of affairs,
Bauer asserts that
...the major task in control over our destiny is
to make as many second-order consequences as possible
intended, anticipated, and desirable; and reduce to a
practical minimum those that are unintended, unantici-
pated, and undesirable."[1]
Methods of Anticipation
The major difficulty encountered in this task is undoubtedly
the anticipation of second-order effects. Within the past few years,
interest in the field of technology assessment has generated an
extensive literature dealing with formal methodology for technological
forecasting. Much of this material, however, deals only with the
future behavior of technology itself. Typical of the predictive
aids to be found isa model of the process of technological innovation
developed by James Bright. In this model the growth of technological
capability is described as a fundamentally exponential process. [2]
Although such theoretical tools are useful for helping us to under-
stand the nature of technological growth, they tell us little about
what the eventual impacts of new technology may actually be.
-10-
This is a point which Bauer makes in his treatment of the problem
qualitative prediction.[ 1 p 3 1  According to his analysis the"...actual
anticipation of a specific future state that may follow from a given
technological change is a work of imagination."{[1p.32 The tools which
we bring to bear on the problem are of two kinds. They either provide
a base of factual information from which the imagination can draw, or
they act as direct stimuli to the imaginative process.
In an essay appearing in 196413], Daniel Bell discusses twelve
modes of prediction commonly encountered in the social sciences. Three
of these techniques often serve as major sources of generalized
knowledge of social institutions. The 'structural certainties' of
a stable social system describe "... an order of events that are
legally preserved and traditionally reinforced". 1[3 ]p. 8 5 4  They enable
us to make behavioral predictions based upon fixed and explicit
patterns of conduct. Patterns of conduct which are implicit, on the
other hand, are referred to as the 'operational code'. This code
sets forth the underlying rules of the game by which political actors
consciously or unconsciously adapt to new situations. The third mode
of analysis focuses on the 'structural requisites' of a social system.
These requirements constitute the minimal set of concerns which the
system must satisfy in order to continue functioning. Predictions
made from this perspective have their greatest utility in the delineation
of the limits of change.
-11-
This sort of knowledge of institutional behavior undoubtedly
serves as a framework upon which predictions of technological impacts
can be built. Yet, as Bauer maintains, qualitative prediction remains
a work of the imagination requiring an imaginative stimulus. Several
of Bell's modes of prediction fall under this category as well. Bell
refers to 'social physics' as the use of explicit models developed
for the purpose of illustrating hypothetical 'social laws'. Another
predictive technique utilizes the formulation of 'alternative futures'.
In the course of this type of speculation, astructured set of con-
ditions is used as a guide for the writing of several possible
scenarios of future behavior. Additional predictive aids cited
include game theory, dectsion theory, and the use of simulation.
Bauer, however, considers the use of analogy to be the most pervasive
stimulus to the imagination.
-12-
II. Forecasting the Effects of Cable Technology
Bauer's analysis of the problem of prediction is based upon the
assumption that second-order effects of technological change must
always exist. In the case of cable technology used to increase
citizen participation, this assumption seems justified based solely
on a realization of the large number of people that would be affected.
However, all mass communication systems are similar in this respect,
and it is for this reason that some insight into our problem may be
gained by examining the social ramifications of communications tech-
nologies already in existence.
An obvious case to single out for consideration would be one
which addresses the specific issue of participant manipulation.
In their study of 1964 presidential elections, Kurt and Gladys Lang
attempt to answer the question of whether or not the broadcasting of
early election returns in areas where the polls were still open had
any real effects upon voter behavior. A major conclusion of the
study was that thebroadcasts had no net effect on the actual outcome
of the election. However, the investigators did discover that
individual attitudinal reactions on the part of voters did occur.
Since the potential for the outcome to have been altered by such
ancillary effects was a function of the historical circumstances
peculiar to that election, it is not clear whether further election
outcomes would remain similarly unaltered. Commenting, therefore, on
-13-
the importance of indirect effects, Lang and Lang state that:
"Direct effects-or the lack of them-do not exhaust
the whole range of effects attributable to commun-
ications. Maintaining stability in the face of in-
fluences toward change involves some kind of adjust-
ment, and communication research has gradually learned
to include these other effects within its range of
interest." [4]p. ix
According to Bauer's classification of technotgical impacts, the
direct effects mentioned above would be considered second-order.
However, the attitudinal changes which may possibly influence the
psychological process by which stability of voting intention is
maintained could be referred to as third-order effects. In fact,
the Lang study makes the conclusion that new election regulations
are necessary to insure against possible harmful effects of early
returns broadcasts. The conclusion is based upon the existence of
the indirect and possibly long-range higher order effects rather
than upon a demonstration of any direct effects.
The Lang study was based upon empirical data gathered on the
attitudinal reactions of voters. Although we may be fairly assured
that citizen participation via bidirectional cable has the potential
for causing similar higher-order effects, we are not able to draw
our conclusions on the basis of actual observation. In our case
the problem is more difficult as the technology in question has not
yet come into actual use. Although pilot two-way systems have
already been developed by private interests such as the Mitre
Corporation and Hughes Aircraft Company[ 5 p 49 5 O, little has been done
to develop non-commercial services such as citizen participation.
In a report prepared for the Rand Corporation in 1971, Walter S. Baer
attributed this situation to a 'chicken-and-egg' effect. That is,
that the capital outlay required for the installation of the necessary
hardware will not be made until a sufficient market for such services
has been demonstrated. However, no demand for the services can be
generated until they actually become available to the public. The
end-result is that:
"Those interested in the community uses of
interactive cable television have two alternative
courses of action. The first is to wait until com-
mercial services have brought two-way terminals into
enough homes to make non-commercial uses attractive
on a marginal cost basis... .The other course of action
is to seek public support today for a few experiments
to develop and test non-commercial use of interactive
television."[6]p.87
In order to meet this need, the National Science Foundation
decided to solicit proposals for experimental programs in the ap-
plication of two-way cable to non-commercial public services in
urban areas. In 1974, grants were made to seven universities and
non-profit research organizations for developing their proposals in
detail by working with local governments and cable communications
operating companies. Several of these proposals, including those
of the Urban Institute, the University of Southern California,[8]
the University of Denver , New York University , and Lehigh
[11]University , dealt explicitly with programs designed to increase
citizen participation in government.
-15-
III. Plan of Study
None of these proposals extend their investigations much beyond
a concern for the first order effects of the technology. That is,
for the basic capability of two-way cable systems to provide such
services. These studies are therefore of only limited utility in an
analysis of higher order effects such as those discussed in the
Lang study. In view of this fact, our analysis must of necessity
remain a work of imagination. The methodology we employ will there-
fore have to conform to that described by Bauer in which two funda-
mental types of tools are used to anticipate secrond-order effects.
Political analysis based upon general knowledge of relevent
social institutions can afford insight into the possible forms that
higher-order effects may assume. This first stage of analysis is the
focus of the second chapter of this thesis. In order to obtain more
detailed knowledgeof possible manipulatory effects, however, new
sources of information are needed to stimulate the imaginative process.
What is needed, therefore, are actual observations Which can be
analyzed for indications of factors likely to lead to manipulatory
effects. For this purpose, the results of simulations carried out
during the course.of the M.I.T. Community Dialog project were utilized.
Chapter Three begins with an investigation of the probable future
configurations of cable participation systems and then proceeds to
compare these characteristics with those of the Community Dialog
system. Relevant results from the simulations are thel reported.
-16-
These simulations were designed as rough investigations of
the characteristics of cable participation systems in general rather
than of the problem of manipulation in particular. However, three
of the observations made during the course of the Community Dialog
simulations can serve as important inputs to an analysis of mani-
pulat:ory effects. It remains to evaluate the implications of those
observations within the context of the relevant social-psychological
literature. Such a discussion is presented in Chapter Four which
outlines three possible mechanisms of manipulation. The major issue
addressed in this chapter is the question of whether or not it is
valid to grant that such effects might in fact occur. Finally,
Chapter Five restates these conclusions and presents suggestions
for further research.
-17-
CHAPTER TWO
BACKGROUNDP GENERAL KNOWLEDGE OF RELEVANT SOCIAL INSTITUTIONS
Much of the analysis which has already appeared in the political
science literature can provide valuable insights into the nature of
the possible impacts of cable technology on society. Any review of
this literature, however, should begin with a discussion of the
motivation behind citizen participation applications of the cable.
A primary stimulus to the development of cable participation
has been the alarming trend of growing alienation present in American
society. The results of a Harris poll conducted during the late
sixties and early seventies [l2]p.l indicate that in 1972, 53% of
the respondents agreed with the statement: "What you think doesn't
count much." In addition, fifty percent felt that the people running
the country really didn't care about them. Several factors have been
suggested to account for this - among them, the idea that the alien-
ation of citizens is directly related to the increasingly complex
nature of government. [7 ]p.T- 4  Government is no longer only res-
ponsible for providing basic services. It must now administer pro-
grams which interface in complex ways with important social issues;
and it must do this with little knowledge of the correct way to
proce4.
Insufficient communication between policy makers and citizens
often results in decisions which are unresponsive to the needs of
the community. This tends to cause further alienation which in turn
-18-
leads to decreased communication producing a vicious circle.
Another contributing factor which has been cited lies in the
nature of existing communications. In the age of mass-media, the
flow of communication is usuually one-way. This results in a
situation in which
"...the citizen watches T.V., reads newspapers, and
listens to radio, but he has no way of talking back.
He hears, but he is not heard. At least that is the
way he feels."[12]p.1
Since existing channels of communication appear to be inadequate,
the solution may be to provide additional feedback links from the
citizen to decision-makers. In doing so, the individual may gain a
sense of particpation while decision-makers would be provided with
better information on the attitudes of their constituents.
The advocates of two-way cable list several advantages which
seem to favor it as a new means of citizen feedback. The first
advantage is that cable can reach a large potential audience in their
homes (or at least at neighborhood terminal locations). It is also
an economically feasible medium for programming addressed to small
specialized audiences because of its large channel capacity. In
addition, cable is, after all, a visual medium and can therefore
convey more information than most other means of communication.
-19-
TL Citizen Participation In A Democracy
Some of the most important conclusions to be made concerning
the possible consequences of cable technology have been derived from
a general knowledge of the role of citizen partipation in the function-
ing of a democratic society. Some of the benefits which increased
citizen participation might bring have just been stated. However, as
Ithiel de Sola Pool points out in his article on citizen feedback1 1 3 ]
we must pay a price for these benefits. The objective for any new
system for increased citizen participation must therefore be weighed
in terms of a series of trade-offs.
This fact is especially clear with respect to the issue of
referenda. Adherents of the vision of a direct democracy have proposed
the use of two-way cable for instant referenda in which the people
themselves vote on the issues. However, most authorities agree that
increasing citizen participation to this extent would be unfeasible
due to the nature of the legislative process. The typical legislator
is faced with a massive complex of bills and issues which require
his full attention. Even so, most legislators are forced to special-
ize in just one area in order to be able to act effectively. In other
areas they must simply take the advice of trusted colleagues. It
therefore does not seem possible that any citizen participating
during his leisure time would be able to competently decide upon
issues.
-20-
Increasing citizen participation to the point of direct
democracy, however, is an extreme in a wide range of possibilities.
What can be said about the impact of less extreme increases on a
political structure such as our own; that is, one based on a balance
of pluralism and consensus? One possibility which Pool points out
in his article is that high-level decisions may become harder
[13]p.243
rather than easier to make1 . As more and more people are
brought into the decision-making process, discussion becomes more
time consuming and often ends in stalemate,
However, the trade-off in this case involves more than just the
relative efficiency of decision-making processes. Stalemates are
generated when a population becomes more politicized and citizens
become increasingly committed to particular points of view. It soon
becomes apparent that if "...citizens are brought by effective personal
participation to the point of caring very deeply about political
outcomes, then there had better not be too many important political
decisions; for every time one is made there are losers as well as
winners."[ 1 3]p. 24 4  And we know that for any highly politicized
minority
... the more vital the sphere of politics,
the more disillusioned they will be at society's
failure to accept their version of the truth. Thus
the price for having a politically active citizenry
in a free society is a sufficient devaluation of
political decisions so that losing is not intolerable
to the losers."[3]p.245.
-21-
it is for this reason that society tries to reduce the necessity
for consensus in the decision-making process as much as possible.
National government passes the responsibility for many decisions
down to more localized levels such as the city and the state. In
this way different segments of the population may make different
decisions and thereby avoid nationwide conflict. The general
conclusion to be drawn is that the objectives of any system designed
to increase citizen participation must be limited enough in scope
so that the necessary political balance can be maintained.
I1. The Issue of Manipulation
General knowledge of factors which determine the fundamental
role of citizen participation within a democracy must certainly
be a prerequisite for the analysis of cable participation programs.
However, it is unlikely that observing anything less than an actual
large-scale, real-world system would provide any practical information
on how to maintain a proper pluralistic balance. There are other
consequences of cable participation programs, however, which may be
more amenable to analytical treatment on a practical level. The issue
which is of greatest in concern in this paper is that of mani-
pulation. Pool summarizes the problem with respect to two-way
cable when he states that
"...better communication technologies that create
more effecient, more extensive, and more intensive
interaction between public figures and their con-
stituents may reduce the sense of alienation by
-22-
making the public figures better able to
respond to their constituents and to influence
them. There is no electronic difference between
these two prcesses. They are both enhanced by
efficient two-way communication. The spector of
electronic manipulation is simply the other side
of the same coin of the hope of electronic demo-
cracy." [131p.24 2
Knowledge From Past Experience
Past experience with experimental citizen participation programs
which did not utilize cable technology indicate that several possib-
ilities for overt manipulation by government officials do in fact
exist. One study which was completed in 1974 reported on the operation
of the Televote system which was designed to aid rapid two-way communi-
cation between decision-makers and their constituents. [14] The system
was demonstrated during the course of a seven month experiment in
the San Jose Unified School District.
In this experiment, votes on various issues were taken amont a
large number of citizens who voluntarily registered to participate.
For a particular poll, schoold district officials would define the
issues and state the reasons behind the various policy choices. This
information and a questionnaire had a corresponding numerical code
which the participants could dial on their telephones once they had
made their choices. A computer registered and tallied all the votes
entered in this manner, and the results were relayed to school officials
within two days. The polling results and any impacts they may have had
on public planning were later publicized to reinforce participation
in the program.
One opportunity for manipulation to which officials might
possibly ayal themselves is itself related to a question concerning
the first-order effects of the technology, Specifically, did the
system perform the function for which it was intended; that is, to
provide current information on citizen views to public officials
so that their decisions would be more responsive to the needs of the
community. The summary of the report indicates that few conclusions
can be drawn with respect to this question:
.. A long-range aim of a televote system is to make
public decisions more responsive to the desires of the
constituency served. Whether any progress was made in
this direction in this brief demonstration is difficult
to say...but the fact that the televotes often gave
school officials new information at least provided them
better opportunities to be responsive."[14]p.20.
It is apparent that no conclusion on whether decisions actually
were more responsive seems to have been drawn. However, the results
included in the report do appear to indicate a few general trends.
A total of thirteen votes were taken over the seven month period
of the experiment. Seven of these votes were on issues related to
official planning efforts already under way. The officials themselves
took the leading role in defining the questions for these votes, and
it is clear that the voting results had an impact on the final decision
in four of these cases. External circumstances seem to have prevented
the results from having a similar impact in the three remaining cases.
The other six votes taken during the course of the project were not
initiated by officials, but by students, staff, or citizens. Of
these the report states,
-23-
-24-
"...It may be too early to make final judgements
about the eventual effects of such telerote results,
but as yet there seems to be no tangible impact on
district policy or decisions as a reult of those
televotes not initiated in connection with official
planning for a specific decision."144]p.19
Although it appears from this that officials weighed public
opinion only on a selective basis, the citizen response to the
televote system was quite favorable. The program also seems to
have effected a change in attitude among some sections of the population
towards school officials. This is especially apparent in the responses
to a survey question which asked participants how interested they
thought the local shcool district was in their opinions. The
question was asked of suburban residents in a pre-experiment survey
and again a year later after the experiment had been completed. The
number of those who chose either 'very interested' or somewhat
interested' rose from 47% on the pretest to about 62% after the experi-
114]p,31
ment. A control group had a 52% response both yearsi
It seems clear that many of those who had participated found
some degree of satisfaction in doing so even though it is not clear
whether decision-makers were in fact more responsive to citizen needs.
The question then becomes whether the real effect of the technology
is the "...psychic satisfaction that it gives to the participating
"[13]p.237
citizen or the achievement of better governmental performance.
It is therefore conceivable that politicians could actually use the
medium for the reduction of community pressures rather than as a
means of becoming more responsive to them.
-25-
The results of the experiment also seem to indicate a second
potential source for the oyert manipulation.of citizens. We have
already noted that the results of the project indicate that officials
only responded to votes on issues which they had taken the leading
role in defining, In commenting on this fact, the report suggests
that:
t. .giving officials more freedom to define issues
as they see fit would increase use of results. Such
a change would require special attention to whether
all sides of an issue were stated fairly. Planners
might be tempted to propagandize rather than to en-
courage citizens to weight all relevant factors and
make a judgement in the public interest.[l4]p.19
State of Current Knowledge
All these possibilities have been suggested by past experience
and a general knowledge of political systems. But to what extent
are the manipulation problem and similar consequences of the techno-
logy treated in proposed programs of experimentation? Among the
experimental proposals submitted to the National Science Foundation,
[7]that of the Urban Institute is indicative of a growing concern
for possible undesirable effects of cable participation. Unfortunately,
the Urban Institute program does not seem to deal with the problems
of manipulation and privacy beyond the stage of ensuring that they
don't arise during the course of that particular experiment.[7]p.TI-15
It would seem, then, that the base of factual and analytical
knowledge just reviewed can provide us with general guidelines for
determining constraints on future cable participation systems.
-26-
Hpweyer, Lt also seems clear that such knowledge does not provide
us wt th the irore detatled information needed to speculate on the
actual mechanis s of ltgher-order manipulatory effects.
-27-
CHAPTER TRWEE
OBSERVATION SINLATTON OF TWO-WAY CABLE PROGR A IING
In order to discover the characteristics of two-way cable
participation systems that might result in possible manipulatory
effects, it is first necessary to reach some conclusion on the
probable future configurations of such systems. Once the general
technological configuration has been agreed upon, laboratory simu-
lations can be carried out using experimental subjects The results
of these simulations may then serve as part of our basis for
speculation on the higher-order effects of technology.
I. Focusing on Probably Technology
Possible system configurations will obviously be dependent on
the particular functions individual participation programs are
intended to serve, Some of the possible applications which have been
proposed for two-way cable systems include;
1. choosing policy alternatives
2. identification of issues, problems, and areas
of discontent
3. interest group recruitment
4. feedback to decision-makers including stands on
issues and evaluation of government performance
[9]p.84, fl5p.8.
The first possibility amounts to using the technology for
instant referenda, and we may now exclude it from consideration
for reasons previously discussed. As shall be seen shortly, the
-28-
fourth application has the greatest potential for involving large
segments of the population, while the second and third possibilities
seem to be relatively smaller in scale.
Perhaps the most natural way to implement the fouth application
would be to use the cable to extend existing opportunities for feed-
back. The traditional forum for expression of public response on
issues is the hearing. Although public hearings often precede the
making of a decision, they are clearly inadequate as channels of
real feedback. Due to the invonvenience of attending these hearings
and the general lack of knowledge concerning them on the part of the
public, attendence is often limited to a few experienced advocates
of the various contending positions. It may be possible to correct
the inaccessibility of public hearings by televising them and offering
the possibility for instantaneous feedback. A true extension of the
public hearing, however, would require a full video return channel
capability.
Such systems can only be employed on a limited basis, however,
and the potential audience would most likely be restricted to
neighborhood terminal locations. In order to involve a larger
participant audience, the upstream communication link would have to
be limited to carrying audio signals utilizing microphones or ordinary
telephone lines, or simple digital signals produced by terminal
keyboards or some less complex combination of response switches.
-29-
Small-Scale Applications
The possible configurations of two-way cable systems to be
employed in the extension of public hearings encompasses the full
range of technical options available; namely, those which utilize
visual, aural, or digital signals for upsteam transmissions from the
audience to the point of program origination, Configurations which
offer full two-way visual and aural capability are thought to be best
suited for performing the second and third functions listed above,
those of issue identification and interest group involvement. Both
these applications require group-to-group teleconferenceing capability.
One teleconferencing system which has already been implemented is that
of the Metropolitan Regional Commission of New York 7although it
transmits by means of microwave relay rather than cable, However,
several of the experimental proposals submitted to the National Science
Foundation do outline programs which utilize the cable, New York
University's design emplys neighborhood communication centers for
the elderly[10]p. 2 5 , while the University of Southern California's
proposal discusses video conference facilities for the interaction
181p.62
of citizen groups and planners.
In a Masters thesis submitted to the Department of Politcal
Science at M.I.T., Michel Quite' describes attempts to enlarge the
potential audience of such meetings by substituting digital or
purely aural channels for visual ones,[ 5 G4 4  Quite' contends,
however, that these methods are largely unsatisfactory for several
-30-
reasons; the first being that "...button-pusjhing isn't highly sati-
sfying way to participate in group-discussion when the influence of
one button pushed is 4mst iperceptable15p
And yet, button pushing seems to be the only way to involve
large numbers of people, for as Guitet points out, even with the
use of simple aural channels, participation is still restricted to
the number of people whose comments can be broadcast during the
program. [151p. 45 The conclusion to be drawn %s that the face-to-
face requirements for satisfactory group discussion necessarily
restrict cable programming initiated for the purposes of interest
group activity and group identification of issues to relatively
small audiences.
Large-Scale Applications
The primary focus of interest in this paper, however, is the
higher-order effects such as those discovered in the Lang study
which may possibly affect large parts of the population simultaneously.
This is not to say that important media effects are not present in
the case of smaller-scale communication. The Joint Unit for Planning
Research at University College, London has in fact carried out important
studies of teleconferencing networks which include topics relateing
to the decision-making process, interpersonal conflict, and reaction
to strangers.1 1 8 ]
Of the four proposed applications for two-way cable, the one
which, seems most likely to involve large numbers of participants
-31-
is that of feedbask to decjston-makers Qf the already existing
channels of citizen geedbck, the pne which has receiyed the widest
acceptance seems to be the public opinion poib Interactiye cable
seems especially suited for this application as it is a visual medium
with an instantaneous response which lends itself eaily to the polling
process. Participants can receive adequate information before making
their decisions, and the questions canibe modified in light of
feedback already receiyed during the course of the polling, It has
been predicted, howeyer, that public opinion polling by cable will
not come to receive serious consideration because of the fact that
a self-selected sample, such as a television audience, is not
1 l9 1p. 8O
statistically representative.
Two basic approaches have been taken t response to this problem.
In the televote experiment described in the previous chapter, the
total number of participants was 6000 or about 4% of the eligible
population drawn from the San Jose Unified School District. The
141p.46
average number of peopleparticipating in any single poll was 700
Even though a large number of people voted, the criticism concerning
the representativeness of the sample is certainly still valid. For
the purposes of scientific survey analysis, therefore, the results have
to be viewed critically. However, one response to this is that a
valuable function was still performed in that the amount of citizen
feedback to planners had increased considerably. Therefore, although
a self-selected polling sample is not representative, most existing
alternatives are even less so.
-32-
Anotherresponse to the problem is to simpley employ scientific
polling techniques. This is the approach taken in seyeral of the
NSF proposals that include polling applications of two-way cable as
17]part of their programs. The proposal prepared by the Urban Institute
for instance, contains an extensive polling project to be carried out in
Peoria, Illinois. It describes a series of eleven polling experiments
to be conducted in 150 randomly selected households. These homes
are to be equiped with terminals capable of conveying a one-digit
response to questions presented during special television programs.
As in the televote experiment the questions are designed to supply
information to the local government which would enable it to make
more responsive decisions-,
As several of the NSF proposals point out, opinion polls con-
ducted over two-way systems would not be limited to traditional formats
currently in use,. In the Lehigh University proposal, for instance,
electronic feedback provides the means by which participants can
voice preferences for topics of future polls. 1 1 P 6 6 As Guitel
points out, participant modification of the polling process could
be extended beyond this point through the use of interactive question-
n ir 153p.41naires1. 1With such an arrangement, questions could be modified
within the course of the same poll. The utilization of instantaneous
feedback therefore enables televised polling to become an essentially
dynamic process in nature.
-33-
It is conceivable then, that when combined with yideo back-
ground materials,this sort of dynamic process could result in a
responsive exploration of issues which might approximate that found
in group meetings. The goal in attempting such programming would
be to obtain a feedback channel offering greater flexibility and
potential for participation than the conventtonal opinion poll, but
which would still not be required to handle the face-to-face com-
munications necessary to true group meetings. Thus, while still
unsuitable for the smaller scale citizen participation applications
discussed earlier, two-way cable employing digital upstream trans-
mission does seem optimal as a means for providing large-scale
feedback to decision-makers.
Systems Utilizing Limited-Response Digital Transmission
Once digital upstream transmission has been recognized as the
most likely choice for large-scale cable participation networks,
it becomes a relatively easy matter to predict the technological
features of such systems. Michel Guite', for instance, has proposed
a design for a useful two-way system to beinstalled in a community
115]p.50
with a potential of several thousand subscribers. The con-
figuration which he recommends for the home terminal consists of a
r
ten-button response pad, converte to translate between cable and
home equipment, and a frame-grabber. This recommendation was based
upon an estimated cost for the system ($350) being weighed against
potential benefits of additional equipment,
-34-
The ten-button response pad included in the design provides
enough flexibility to permit an interactive polling process. It is
for this reason that Guite' prefers it to less expensive four-button
response units which don't offer that capability. [15p.51 Increasing
the information capacity of the unit beyond ten buttons, on the other
hand, does not seem justifiable either. Although he feels that a full
alpha-numeric keyboard would permit open-ended response leading to
greater satisfaction in communication, Guite' indicates that the
[l5]p.4l
extra expense involved would be prohibitive.
The conclusion of a most likely terminal configuration consisting
of a limited arrangement of response switches is also reasonable from
perspectives other than the cost-benefit approach taken by Guite'.
This is especially clear in light of the fact that the type of
response capability available for community services will most
probably be determined by the response capability already available for
commercial services. Commercial equipment manufactured for per-program
pay TV systems closely resembles that required for public opinion
polling, and it is therefore not surprising that most of the NSF
projects have obtained their systems from this source.
The Urban Institute's proposal, for instance, gives a detailed
account of its search for equipment suitable for adaptation to com-
[7]VII-17
munity servies. The University of Denver project uses a
modified version of the Hughes SRS system console which consists of
a channel selector and small response button patd The Lehigh Univer-
sity system is similar and is part of the Twin County Trans-Videv
[ll]p.36pay TV system configuration.
II. Investigating Probable Characteristics of Cable Systems
The necessity to limit the response capability of home terminals
used in large-scale polling applications is also clear from another
viewpoint. Guite' excluded full alpha-numeric keyboard terminals
from consideration on the basis of cost. However, in the context
of live interactive programming, they are clearly unsuitable for
reasons he already stated in connection with verbal participation
in group meetings; namely, that participation would be restricted
to the number of people whose comments could be broadcast during the
program. ] Clearly, for any type of polling process involving
large numbers of people, the vast amount of information contained
in the open-ended responses of alpha-numeric keyboards would be
impossible to process adequately, even with the aid of the computer.
The Community Dialog Project
It is evident from the above discussion that our final choice
for a most likely system configuration should include a home terminal
which uses a limited arrangement of response switches for upstream
transmission. Having decided upon this configuration, however, what
can be said about characteristics of the system which may hold some
implication for the problem of manipulation? Speculation on the
possible mechanisms of manipulatory effects must still be based
upon some knowledge of how the system will actually futiction. An
attempt to gain such knowledge has been made in our case through the
use of simulation.
-37-
Figure 1.
Community Dialog Project Equipment
li
2W11
63
7m
6ZL
Controller
e.g ~
Display
Handheld
Response
Switch
-38-
In order to carry out such simulations, use has been made of the
work of the MIT Community Dialog Project. The focus of the Community
Dialog Project is on the use of an electronic polling device as an
aid to discussion inface-to-face group meetings. The hardware em-
ployed in this system is shown in Figure 1. The technology is similar
to that of our hypothetical home terminal in that each member of the
group is supplied with a handheld response unit which can enable him
to vote in one of several possible categories. These individual
switches are wired to a small special-purpose computer which displays
the results of the vote. By using this system, each member of the group
is able to
"...make an annonymous coded response to questions
posed by the moderator or another participant, and
to observe instantaneously a tally of how may people
voted in what category. The purpose is to get a rapid
appraisal of where there is consensus and where there
is controversey, to allow participants to reveal their
ignorance, to deal with controversial questions without
intimidation, and generally to make the discussion more
responsive to real interests and needs of the group'.[20]p.1
The instantanous feedback obtained during the course of such a
discussion enables the moderator to direct the flow of dialog in
response to the desires of the group. In this way a meeting can
very quickly focus upon issues which are of importance to all the
participants. The intended function of electronic feedback in group
meetings is to serve as an aid to discussion rather than as a means
of obtaining finalized votes on specific issues.
Using the Community Dialog voting apparatus to simulate the
-39-
the operation of our hypothetical two-way cable system seems
reasonable simply on the basis of the fact that participants use
the technology to respond in a limited way; they must vote in one
of ten available categories. There are, in fact, obvious differences
between home users of cable technology and the participants in an
actual group meeting who are not limited in their response to the
operation of swithces.
Nevertheless, some of the characteristics of these wired group
meetings may still be of interest. In addition, it is possible to use
the technology under conditions which account for these differences
and therefore simulate the operation of interactive cable systems.
One final motivation for building upon the Community Dialog studies
is the fact that the type of dialog achieved with electronic feedback-
does constitute a loose approximation of the dynamic polling process
which was described earlier with reference to large-scale citizen
participation applications of two-way cable,
Previous Observations
Two of the characteristics of interactive cable technology upon
which we shall base our speculations of possible manipulatory effects
are directly evident in the face-to-face group meetings of the Com-
munity Dialog Project. These two characteristics are the instantan-
eous nature of the feedback and the distinctive branched tree pattern
which dialogs conducted in this manner are constrained to assume.
Bth these features are fundemental to the technique by which the
technology is used.
-40-
The general technique employed has been developed over the course
of the more than two hundred group meeting which have been conducted
using the electronic feedback device. In essence, it consists of a
series of several steps and is continually repeated during the course
of the discussion.[20]p.7 Initially, the moderator poses a question,
typically writing it on a blackboard, and lists a number of alternative
responses, each coded with a number from 0 to 9. In the second step,
each member of the group votes for one of the alternatives by making
the appropriate selection on his voting switch. The vote tally is
displayed, and the group then proceeds to discuss the result. After
this initial reaction the question may be rephrased, or a new question
may be asked and the cycle repeats.
The ability to instantaneously display voting results is essential
to the interactive process of question and response. In addition,
by focusing discussion upon a series of explicit questions linked
by group reaction to preceding votes, a branched discussion pattern
emerges. This pattern has been described in an MIT Doctoral thesis
written by Noam Lemelshtrich. Due to the fact that instantan-
eous feedback and branched discussion appear to be characteristic
of all interactive question-and-response type dialogs, it seems
safe to assume that they would also be part of any dynamic polling
process carried out over two-way cable.
-41-
Although these two characteristics will prove to be of funde-
mental importance in our discussion of possible manipulatory effects,
further information is still needed. Also, it is necessary that
this new information be obtained under experimental conditions
specifically designed to simulate interactive television.
In the face-to-face meeting of the Community Dialog Project,
electronic feedback is used to encourage natural discussion among
the participants. However, in our televised approximation to the
group, digital feedback from the audience completely replaces natural
discussion. A major question therefore arises as to whether such a
limited response capability eliminates the possiblity of ever achieving
a dynamic sortof dialog over a televised medium. Other differences
between face-to-face meetings and interactive television exist as
well, and it is for this reason that other simulations were undertaken
to discover functional characteristics peculiar to the medium of
two-way cable television.
Simulation
Two sets of experimental simulations were carried out in November
and December of 1974 with the cooperation of the Center for Advanced
Engineering Study at MIT. The programming application under study
in each of these simulations was that of interactive polling used
to approximate the type of participation attained with group meetings,
Although the group meeting setting can never really be fully extended
to the case oflarge-scale programming applications, it was hoped that
-42-
a dynamic polling process could prove flexible enough to allow for
more audience-directed participation than could otherwise be obtained.
The underlying purpose of these simulations was to discover the
problems involved in actually implementing such systems so that the
functional characteristics revealed in these problems could be studied
in greater detail. It was for this reason that the simulations were
carried out in two stages; the first concentrated on problem discovery,
while the second explored the characteristics of these problems.
First-Stage Simulations
One cf the most important features of any communication medium
must be the potential quality of interaction it makes available.
Evaluations of the quality of interaction made during the Community
Dialog feedback experiments were based largely upon subjective
criteria. Participants were provided with questionnaires which
enabled them to indicate the degree of satisfaction they found in
the group meetings. Studies of these results and of other observ-
ations have indicated that the feedback technology interacts strongly
.[20]p.14-19
with the behavioral dynamics of the group.
In the Community Dialog meetings, the purpose of the electronic
feedback was to agument natural discussion. All the face-to-face
interactions which are present during normal group meetings were also
present during the Community Dialog meetings, However, this situation
is drastically altered in the case of interactive television. If
the meeting cpnsjsts of interactive program received in individual
-43-
homes, most of the members of the 'group' are physically isolated
from one another. The elimination of all interaction between group
members (other than that made possible by simply observing the
collective feedback) should have a considerable impact on the dynamics
of the group.
An additional interaction which is eliminated in the case of
two-way cable is the one which occurs between the moderator and the
group. In the case of interactive television, the participants can
no longer feel that they are being observed by the group leader.
Of perhaps even greater importarce is the fact that the moderator
no longer receives information about the group's non-verbal behavior.
The only feedback he is able to receive is that which appears on
the electronic display. As a result of these considerations, it was
initially decided that any simulation of interactive programming would
have to reporduce conditions of isolation between group members and
between the group and the moderator.
During the experiment, each subject was seated before a television
monitor on which he viewed the program. The participants were also
supplied with handheld switches that enabled them to feedback their
responses. Because of physical limitations, no more than eight subjects
could be supplied withmonitors at any one time. In addition, the
monitors had to be located quite close to each other so that it became
necessary to separate the subjects with only simple plywood partitions.
The monitors and the partitions rested on two tables with four subjects
-44-
being seated at each table. The television program originated
from the next room so that the moderator would be physically isolated
from the audience. The device which displayed the voting results
was operated by the moderator and was visible to the audience over
the television. A representation of this configuration is shown
below in Figure 1.
w~t
0
K>
00
.7 rVir
00
K>
KI
0
0
A - -- I
0 Vtro 5o3 Switch
Subj ect
v otap-e
Go
tfolerc tor
Studio
V/ct!,
Figure 2. Physical Configuration of First-Stage
Simulations
OC)
-do-Mpg
-LJL
-45-
The programming format used in the first simulation was
similar to the meeting format used in the Community Dialog Project.
As in the Community Dialog meetings, this experiment employed one
moderator who asked questions of the audience and operated the
electronic display. Throughout the entire length of the program,
the camera focused upon the moderator and the display simultaneously.
The topic of the meeting was student residential life at MIT, and
it consisted of a series of prepared questions to which the audience
responded by using their voting switches.
In the non-television feedback sessions, simply drawing up a
list of questions was usually all the preparation a moderator needed
to make before a meeting. The questions served as a framework for
discussion drawn from the audience. As a result, any extensive pre-
paration would often prove tojo inflexible. In fact, the primary reason
for using the feedback technology in these meetings was to stimulate
and direct group discussion rather than to obtain voting statistics
for a list of questions.
The clearest result of the televised experiment, however, was
that such discussion never occurred. Obviously, it had been expected
that the audience of a televised meeting would be incapable of actually
vocalizing their feelings using the electronic devices as their
only means of feedback. But the assumptions prior to performing the
experiment was that by interpreting the responses to his questions,
the moderator could formulate new questions which would reveal the
desires of the audience and thus enable an evolving dialog to take
-46-
place. However, the televised meeting contained little discussion
and was for the most part completely static. The audience responded
to the questions one at a time in a manner similar to the way in
which they would have responded to an ordinary poll.
The Community Dialog experiments indicate that the skill of
the moderator is often an important factor in achieving satisfactory
f20]p. 13dialog, The moderator in this experiment, however, did have
a great deal of experience in running group feedback sessions. The
apparent conclusion is that the methods which had previously been used
in trying to draw discussion from the audience had failed in the case
of the televised meeting.
Most of the experimental subjects for this simulation were
students from an undergraduate course dealing with television and media.
The use of electronic feedback in group meetings had been demonstrated
during a class session prior to performing the experiment. For this
reason, most of the subjects were already familiar with the type of
group discussion carried out in the Community Dialog experiments.
When asked to evaluate the televised meeting, therefore, several of
the replies mentioned the lack of dialog in the simulated cable program.
For instance, in response to the question which appeared on the follow-
up questionnaire.
1) Did you feel a sense of participation in the 'discussion'?
one of the subjects wrote:
-47-
"Yes, but about as much as a questionnaire, even though
there was more immediacy."
And when asked to compare the televised meeting to the face-to-face
electronic feedback session, another_ respondent replied that in the
face-to-face meeting..."you could reclasification of a question.1
But in the televised meeting, however..."I felt a good deal of
frustration about clarification, and also because I couldn't explain
my answers."
Several of the questionnaire responses described the problem as
being one of 'frustration'. The audience's limited ability to respond
apparently led to a breakdown in communication. This is indeed the
problem which had been anticipated prior to performing the experiment.
Although the feedback devices used by the audience were the same as
those used in the Community Dialog meetings, the ability to respond
was severely restricted in the case of the televised meetings in
which the moderator was unable to call upon members of the audience
to explain their vote or to pose new questions, In addition, televised
meetings eliminate all the non-verbal cues which often enable a moder-
ator to remain touch with his audience. It is therefore apparent
that it is the limited response capability of electronic feedback
which poses the fundamental problem for its application to inter-
active television.
-48-
Second-Stage Simulations
a. Preliminary Sessions:
Although the results of the first simulation reveal the existence
of frustration in communication, they do not give us much information
on the specific characterisitcs of the group meeting which actually
play a role in the communication breakdown. Only general conclusions
concerning the effect of the limited respone capability of electronic
feedback could be drawn from the questionnaire responses. However, if
we wish to achieve satisfactory dialog without resorting to additional
channels of communication, we must discover features of moderation
technique and program software which might prove useful in solving
the problem.
In order to obtain results which would be more concrete than
those of the first simulation, some type of controlled experimental
method had to be employed. This required a familiarity with specific
factors Qf interest prior to the formulation of the experiment design.
For this reason, a series of weekly preliminary sessions were held
during which televised group meetings were simulated and then analyzed.
During the preliminary sessions the experimenters themselves acted as
moderator and audience. By using closed-circuit television equipment
and the electronic voting apparatus, group members had the opportunity
to test a variety of programing and moderation techniques before
deciding upon the yariables to be investigated in the final experiment.
-49-
During the first of these weekly sessions, the group discussed
the subjective criteria by which any particular technique was to be
judged. It was decided to assess the quality of any interactive process
by evaluating the three general characteristics of participation,
nature of feedback, and efficiency. A subject's sense of participation
should reflect the extent to which he/she is actively involved as
well as the impression he/she develops of the importance of their
own feedback. The latter factor provides a measure of the perceived
responsiveness of the moderator, The second major characteristic,
the nature of the feedback, denotes the extent to which the qualities
of subtlety and richness of information may be found in communication.
Finally, the third characteristic, efficiency, signifies the amount
of time and effort which must be spent on any particular programming
technique, This factor has implication for the continuity of the
program and for its consequent ability to maintain audience interest,
During the preliminary session, the techniques employed in each
trial were analyzed in the context of the above general criteria. Most
Qf these trtals consisted of televised group meetings moderated by a
single person in a manner similar to that of the first stage experi-
ments, By occasionally interrupting the proceedings, the experimenters
were able to analyze the ways in which the moderator interpreted or mis-
interpreted the feedback. The' conclusion immediately drawn was that
the results of these first nitial trials were identical to those
of the first stage simulation, That is, the inflexibility of the
-50-
response system had led to a breakdown in communication.
This breakdown most often occurred in relation to the
moderatorls inability to correctly interpret audience responses.
For example, members of the audience would sometimes vote for
category numbers which had not been associated with any of the
alternative choices for that vote. The moderator would often take
these votes to mean that a portion of the audience simply couldn't
decide on an answer to his question. In actuality, however, voting
for an undefined category was usually the only means available for
signalling any one of a number of possible messages. Aside from
signifiying 'don't know', these votes for undefined numbers were
also seen to have meant 'question unclear', 'other', 'insufficient
information', 'object to question', and 'combination of these'.
Providing an extra category such as 'other often does little
to end the confusion as the moderator is still unable to discover
what the audience is trying to say. This is illustrated in the record
of one of these sessions:
Moderator: Do you think that grading at MIT should be based on:
1. letter grades (2 votes)
2. pass-fail (1 vote)
3. other (2 votes)
Moderator: What do those who voted for 'other' want?
1. combination of grades and pass-fail (3 votes)
2. no- grades at all (0 votes)
3. other (2 votes)
Moderator: Should we have a combination consisting of:
1. only freshman year pass-fail (0 votes)
2. courses outside major pass-fail (1 vote)
3. certain number of courses pass-fail (2 votes)
4. other (2 votes)
-51-
This exchange is typical as the moderator never does find out
what the two individuals who voted 'other' in answer to the first
question had in mind. It appears that none of his quesses may have
been correct as two votes still appear in the !otherl category.
However, the answers to the second question do indicate that those
who voted for the choices of letter grades and pass-fail in the
first question may actually have wanted a combination of the two.
Being unable to discover what the two 'other' votes signified, the
moderator decided to concentrate of the three ndividuals who wanted
a combination of letter grades and pass-fail, From the results of
this third yote, it is clear that he had indeed lost the other two
members of the audience as two votes continue to appear in the 'other'
category.
Typical of the comments made by several of the individuals who
acted as moderator was the statement; "The information is there, but
I just don't know how to get at it.' One common opinion as to the
reason for this was the idea of the moderator being too close to his
own questions. It was felt that having framed the questions from
his own point of view, the moderator would often misinterpret feedback
from others. This led to the idea of using two moderators, one of
whom should not have been involved in the formulation of the questions.
It was hoped that by doing this, the discussion could be directed in
the context of a wider perspective enabling better interpretation of
feedback. In several cases, such as in the following discussion of
-52-
public transportation, this procedure seemed to work:
First Moderator: How often do you use the MTA?
1. often
2, sometimes (majority vote)
3. never
Second Moderator: How much difference would it make if the
Red Line stopped at 77 Massachusetts Ave.?
1, big difference
2. some difference
3. no difference (majority vote)
(Discussion between first and second moderators and a new line of
questioning follow).
In this exchange, the second moderator wanted to find out why
most of the audience only used public transportation occasionally.
The responses to the second and third questions indicate that improved
service and lower cost would not increase use appreciably. During
the discussion which followed; the two moderators concluded that the
reason for the low usage of public transportation was the geographic
characteristica of this particular audience. That is the audience
was composed of students who lived on or very near to the campus.
They therefore had little need for public transportation. At that
point the moderator decided to discard several additional prepared
questions concerning individual use of public transportation. If only
the first moderator had been present, it is likely that he would have
continued down his list of prepared questions without realizing that
they were not suited to this particular audience.
-53-
Cowments made by the moderators and by members of the audience
during the follow-up discussion confirmed this conclusion. There
was general agreement that having one unprepared moderator increased
the ability of the moderators to correctly interpret the feedback.
The two moderator technique was also judged to be yaluable on the
basis of other considerations as well One important conclusion
was that it resulted in an increased sense of participation on the
part of the audience. A major element of this effect seemed to be
the fact that having two moderators in front of the camera enabled
the audience to see a discussion which they could feel a part of.
With a 4single moderator, the discussion was usually reduced to little
more than a poll.
However, it was also observed that too much discussion between
moderators had the effect of shutting out the audiene. It often
appeAred that prolonged discussion between moderators resulted in
their reaching a conon conclusion which caused them to be as un-
receptive to the feedback as the single moderator had been. The
tendency for the moderators to become overaly involved in their own
discussion led the group of experimenters to hypothesize that
having a full panel discussion before the camera could result in
even greater isolation of the audience.
But for the most part, the group agreed that the technique of
using two moderators was an extremely valuable one. By providing
more objectivity and increased ability to formulate questions, it
-54-
enabled more responsiveness to the audience. By introducing actual
discussion into the televised program, it added an element of
entertainment so that the audience was able to maintain greater interest.
The number of moderators was not the only variable investigated
during these preliminary sessions. Tn addition, several new program-
ming techniques which consisted of variations in the question format
were introduced. The complaint which was most often made during the
first stage experiments was that the multiple-choice question format
was too restrictive. The basic problem of frustrtation arises when
members of the audience cannot agree with any of the avaiable response
choices. They then tend to vote for an undefined category number or
for an 'other' category if one is provided. If a significant number of
participants vote in this way, the moderator must try to discover the
audience's problem.
When trying to formulate a strategy capable of accomplishing this,
the first idea which. came to mind was that of a simple game of 'twenty
questipns'. It was felt that by developing some optimum technique for
progressing through a binary tree of yes-no type questions, the moder-
ator would have a procedure which could eventually tell him what his
audience was thinking. However, this idea was ruled out during the
first of the weekly preliminary sessions. It was felt that any
procedure which required a long series of questions in order to
clarify one particular point of discussion would simply be too time-
consuming.
-55-
As an alternative, a frustration question' format was developed.
It consisted of a prepared list of alternative responses which the
audience was confronted with as soon as communication began to break
down over some particular point of discussion. This list included
such choices as nsufficient information' object to question',
'don't know', etc. Although such a format cannpt completely solve
the problem, it was hoped that its use would introduce a greater
degree of flexibility. Another attempt to increase flexibility was
made by developing a gradated chqice scale. In using such a scale,
the audience would be instructed to rank their opinions on a scale
of 1 to 5 with 3 being the null position. Thus, if the extremes
of the scale were defined as 'agree' and 'disagree, for instance,
subjects would haye a range of five possible choices: 'agree',
lagree somewhat', 'no opinion', 'disagree somewhat' and 'disagree'.
Experiment Design
In order to discover the relative merits of various programming
and moderation techniques, the experiment had to be designed so that
the effects of any particular technique could be isolated. For this
reason, the experiment was divided into three separate sections, each
consisting of a program segment which was approximately twenty minutes
in length. The first segment consisted of a televised group meeting
which was moderated by a single person. The second segment had two
moderators, and the third was a panel discussion in which there were
four persons appearing before the camera. During each of the program
-56-
segments, three types of question formats were employed. The re-
suiting arrangement of variables is summarized in Figure 2. One
audience was to be shown each of the three program segments in the
above order. The, three types of question-format were incorporated
into each session.
Program 1 2 3
Format Single Double Panel
Moderator Moderator Discussion
Questi:on (20 min) (20 min)
Format
Standard
Multiple-Choice
Scale
Frustration Technique
Figure 3: Experiment Design Scheme
The degree of control which can actually be obtained by using
this scheme is not as great as it might appear. It must be remembered
that the nine specified variables were purposely incorporated into
the experiment design as a result of the deliberations which took
place following the preliminary sessions, Therefore, the design does
not take into account the effects of any variables which may not have
been uncovered during those sessions. But in view of the fact that the
experimental goal is still one of general problem exploration, this
deficiency seems acceptable. If there were any other major factors
at work dur~Lng the experiment, general observations and questionnaires
-57-
responses could hopefully reveal their existence in a manner similar
to the way in which the original variables were discovered.
Aside from an analysis of program content and participant be-
havior made on the basis of videotape recordings, additional data was
obtained in the form of questionnaires. It has already been stated
that during the preliminary sessions, the various techniques were
evaluated on the basis of three different subjective criteria. A
questionnaire was designed so that the opinions of the audience could
be indicated in the context of these criteria, The same set of questions
was Asked of the subjects immediately after the conclusion of each
program segment so that the relative merits of each technique could be
evaluated by comparing responses. In addition, a set of more general
questions was asked after the entire experiment had been completed.
A copy of the questionnaire appears in the appendix.
An additional factor which had to be considered in ;the experiment
design was that of program content. The acceptability of the program
material to the audience is a variable of obvious importance. If
subjects were to become bored with a particular program segment, cross
comparisons between the results of that segment and the results of
other segments would become difficult if not impossible to make.
Attention was therefore given to the proper choice of topics for each
of the three program segments. The topics chose were grades at MIT'
for the single moderator segment, 'students and the right to privacy'
for the double moderator segment, and !world hunger' for the panel
segment. All three segments were limited to twenty minutes in length,
-58-
Another considerati~on for experiment design was the question of
whether to isolate the suhjects physically from one another. During
the first stage experiment subjects had .been separated by plywood
partitions in order to simulate the actual conditions of interactive
prograxming received in the home, But when partitions were used each
subject had to be aupplied with his own television monitor, and the
limitaion on the number of monitors available for the experiment held
the maximum number of subjects down to eight. When this was considered
in light of the fact that the small number of subjects was a major
deficiency in the first simulation, it was decided that the partitions
should be dispensed with. Instead, a larger audience was gathered
in front of a single television monitor. As before, each member of
the audience was supplied with his own hafid-held response switch.
Subjects were instructed not to talk with each other in and attempt
to minimize the influence of one subject on the behavior of another.
c. Results
The second stage television feedback simulation was performed
early in December 1974. An audience of thirteen students was seated
before a single television monitor to watch a program which originated
from another room. In each of the three program segments, the modera-
tor or moderators appeared before the camera with the electronic
display of voting results. A videotape recording of all three program
segments was made so that general experimental observations could
be made at a later time, Observations made from the tape as well as
-59-
additional observations of the audience were combined with the
questionnaire responses to form the body of experimental data.
Upon analysis, this data was found to exhibit several clear trends.
The most distinct of these trends occurred in relation to
the number of moderators. As expected, the single moderator case
drew favorable comment with regard to being the best prepared and
best organized segment. However, the questionnaire responses also
show that it was considered to be the most responsive of the three
segments. This is surprising in light of the fact that in the
opinion of the experimenters, the videotaped record of the segment
tends to bear out te prior conception of single moderator performance
which developed during the preliminary sessions. The tape shows that
the moderator often had to gnore votes which appeared in undefined
categories, confirming an impression of the single moderator being
less responsive,
In the case of the double moderator segment, favorable comments
were received concerning the smooth flow of discussion during the
program. The audience also seems to have felt that the two moderators
were better able to frame questions and alternative responses. The
same conclusion was reached by the experimenters on the basis of
viewing the videotape. No other distinct trends were observed in the
double moderator case.
The audience's characterization of the panel discussion, however,
is quite explicit. This segment ranked last in almost all the response
categories, which appeared on the questionnaire,
-60-
Of all these results, the unfavorable impression of the panel
discussion is perhaps the most easily interpreted. One of the most
important variables to be considered in the interpretation of any
[201P.13result appears to be moderator performance.JHowever, this
factor does not seem to explain the poor rating of the panel segment.
In the opinion of those who viewed the videotape, the quality of
moderator performance appeared to bi -approximately the same for all
three program segments. No obvious failure on the part of any
individual moderator was indicated.
Two external factors, though, do seem to have played a role in
the audience's evaluation of the panel segment. Firstly, the program
itself was not performed as originally planned. The intention was
to have a panel consisting of one moderator and at least two 'experts'
of opposing viewpoints. It was hoped that the audience could then
participate in the debate so that a high level of interest could
be maintained during the program. However, the experts never
materialized, and their places were taken by several students who
had no particular stand on the issues. As a result, much of the
potential interest in the discussion was lost. A factor of perhaps
even greater significance was that the panel discussion was the last
of the three segments to be televised. Observers who were present in
the viewing room have indicated that the audience was extremely restless
by the time the panel discussion began. It is clear that the audience
had become impatient as a result of the length of the experiment.
-61-
The audience characterization of the single moderator segment
as the most responsive of the three is a puzzling result. Because
of the single and double moderator trials performed during the pre-
liminary sessions, it was exptected that the double moderator
segment would be the more responsive of the two. In the experimenter's
opinion, an evaluation of the videotape tends to confirm this expect-
ation. However, a majority of the questionnaire responses contradict
it.
One possible way to interpret these two conflicting sets of evi-
dence might be that several good characteristics of the single moderator
segment such as its high level of organization may have fostered a
favorable overall impression of the program. This impression may
have in turn had an effect upon the way in which the audience perceived
other characteristics of the segment such as its degree of responsiveness.
Another factor which may explain the conflicting results is that
the topic of the single moderator segment was grading policy. Although
the topic of the double moderator segment was equally interesting, the
subject of grades may have had a more immediate appeal to the student
audience.
It is also possible that an external factor may have played a
role in the audience's preference. A great deal of difficulty had
been encountered in trying to obtain suitable subjects for both the
first and second stage experiments. Most of the subjects for second
stage experiments were students who had been contacted by members of
-62-
the group that conducted the experiment. As it turned out, a large
number of these students were later found to be unsuitable as experi-
mental subjects. Unfortunately, they were found to have been biased
in favor of the single moderator segment even prior to their having
viewed the program. It therefore comes as no surprise that their
responses favored it.
The second set of variables to be studied consisted of several
types of question formats. These included a gradated choice scale
and a special 'frustration question' as well as the standard multiple-
choice format used in previous experiments. On the basis of the
questionnaire responses, it appears that of the three, the gradated
scale seemed the most satisfactory to the audience. However, the
significance of this result is limited due to the fact that although
the questionnaire is comparative in context, the techniques themselves
differ in purpose. For instance, the gradated scale is of no use in
trying to eliminate a particular source of frustration. It is also un-
suitable for manysituations in which a multiple-choice format must
be used. Yet it is clear that when used in conjunction with the
multiple-choice format, the gradated scale provides for a good deal
more flexibility than multiple-choice alone.
It was also found that the multiple-choice format rated highest
in frustration during the single moderator and panel segements. This
conforms to the expectation developed during the preliminary sessions.
During the double moderator segment, however, the multiple-choice
-63-
format was exceeded in frustration by the 'frustration question'
format. There seem to be several reasons for this result. First
is the fact that the three techniques were not employed equally in all
program segments. In particular, the frustration category technique
was mainly used in" the double moderator segment.
Another factor may be that the set of alternative responses
set aside for use in a frustration situation did not include a choice
for those who had not been frustrated originally. This resulted in
some confusion. The inevitable consequences of prolonging such
confusion over a relatively long portion of the program were irritation
and even more frustration on the part of the audience.
In order to be of any real use, the frustration question format
must receive extensive revision. But even allowing for eventual
imp rovement of the technique, it does appear to exhibit one defficiency
which seem uncorrectable. As the process of probing one particular
point of discussion begins to occupy more and more time, portions
of the audience become increasingly disinterested. And since the
frustration technique must of necessity take up a disproportionate
amount of time, it seems to be an inherently inefficient procedure.
In order to justify spending relatively large amounts of time
in trying to discover a particular problem bothering one segment of
the audience, the technique being used must have some value in itself.
That is, the procedure should be of interest to the entire audience.
The conclusion was reached on the basis of comments and audience
-64-
observation. It therefore seems that actual live discussion, of
interest to the audience in itself, should be utilized rather than
a list of more unanswered questions.
Thuswe find that in order to supply the additional information
about the audience's feeling of need to deal with problems of frustration,
at least one channel of natural language communication must be included
in any overall system configuration. Such supplementary channels
are common to many communication systems already in use. The in-studio
audience and the telephone call-in are examples used in contemporary
broadcasting. Regardless of the form of such supplementary channels,
we are forced to conclude that they Ui1ll most probably have to become
an integral part of any future large-scale, limited-response communication
System such as digital-return cable.
-65-
CHAPTER FOUR
DISCUSSION: IMPLICATIONS FOR THE ISSUE OF MANIPULATION
As a result of the investigations carried out in the preceding
chapter, we are aware of three characteristics of large-scale cable
participation systems. From the work of the Community Dialog Project,
we know that the first two characteristics, instantaneous feedback
and branched discussion patterns, are likely to be fundamental
features of such systems. In addition, special simulations have
indicated that supplementary communication channels must be available
if these systems are to be utilized to their full potential. The
purpose of this chapter is to speculate on the possibility of each
of these characteristics serving as a source of manipulatory effects.
I. Continuous Feedback: The 'Bandwagon1 Problem
Ideally, dynamic polling conducted within the context of inter-
active cable programming should create an opportunity for responsive
exploration of citizen views on a larger scale than would otherwise
be possible. Although cable participation sessions can only approximate
a group meeting type of format, the exploratory nature of their content
requires an iterative pattern of question and response. Thus a
fundamental distinction between cable programming and conventional
polling is the fact that in a televised group session, key questions
are bound to be repeated or voted on again later in modified form,
while in a static poll the respondent casts his voted on an issue
only once.
-66-
A fundamental requirement for the development of a dynamic
polling process is that the audience be provided with instantaneous
feedback of its response to questions. This feedback is obviously
essential if the moderator is to elicit audience reaction to the
results of previous votes. However, because the outcome of a session
is usually the result of an entire series of questions rather than of
one single vote, this feedback also has the effect of providing members
of the audience with prior knowledge of group preferences. We can
therefore draw a parallel between this situation and that of announcing
predictions of early returns during elections. This parallel could
serve as the basis for a fear of manipulatory effects. Herbert Simon
describes the nature of such effects and states that they are based
on the supposition that
... the voting behavior of at least some persons is a
function of their expectations of the election outcome;
published poll data are assumed to influence these ex-
pectations, hence to affect the voting behavior of these
persons. If persons are more likely to vote for a
candidate when they expect him to win than when they
expect hi7 to lose, we have a 'baniwagon effect'.[22]p.245
In view of the traditionally held belief that bandwagon effects
stem from a desire to be on the winning side,[ 4 1 p. 144 the parallel
between elections for public office and group votes on issues seems
a reasonable one to make. However, the suspicion that bandwagon
effects may play a role in televised citizen participation sessions
of the Community Dialog [20OP.35is also supported by the experience fthCounyDilgProject.
-67-
Within the context of demonstrations performed over the course
of the Community Dialog Project, the moderator would ask his audience
to choose among alternative answersto some factual question. (The
question typically asked was "How long is the Nile River?".) After
displaying the voting results, the moderator would procede to elicit
comments from individual members of the audience. The vote would then
be repeated and the results displayed a second time. After several
iterations of this process, it was often found that the group's votes
would gradually converge upon one particular choice (usually the
wrong one). The explanation for this phenomenon appears to be that
the instantaneous display of group preferences provided the basis
for a bandwagon effect.
The Fleitas Experiment
Before concluding that bandwagon effects do indeed play a signi-
ficant role in group feedback sessions, we must first refer to attempts
already undertaken to verify the existence of bandwagon effects in
elections. The experimental study which appears to approximate
our situation most closely is that done by Danial Fleitas.
In Fleitas' experiment, several large groups of college students voted
for one of two imaginary candidates under conditions of minimum in-
formation. A minimal-information election is one characterized by
... a relative or total lack or relevant information with which
voters can evaluate the candidates."[23]p.434 A condition of
minimal inforamation also characterized the Community Dialog sessions
-68-
in which participants had to vote for factual alternatives of
which most had to real prior knowledge. For this reason the two
cases do in factiappear analogous.
As it turns our, however, Fleitas was not able to demonstrate any
effect on voting behavior similar to that occuring in the Community
Dialog meetings. Three groups of voters participated in a series of
four successive ballots in a simulated mayoralty election. The
experiment was designed to isolate the effects of facked balloting
results and qualitative bandwagon/underdog messages on votes. In his
conclusion, Fleitas states that the experiment
"clearly demonstrated that mere poll results are
insufficient to impel would-be bandwagon or underdog
identifiers to switch their votes. Rather, the voting
behavior that results from these identifications does
not occur until sensitized or cued by a strong quali-
tative stimulus."[23]p.438
In light of this conclusion, we are now forced to question the value
of the Community Dialog demonstration. If bandwagon effects cannot
be attributed to poll results in the analogous case of minimal-
information elections, what are we to conclude in the case of issue
votes in which voters may have considerable prior knowledge?
Although the situation is unclear, though, there still appears
to be a possible explanation for the differing results of the Fleitas
experiment and the Community Dialog demonstrations. Fleitas found
manipulatory effects to be linked to the qualitative messages he
determined to his experimental subjects. This fact makes it reasonable
to suspect that the bandwagon effect observed in Community Dialog
meeting did not stem so much from the instantaneous display of
vote tallies alone as from the discussion which followed each vote.
-69-
The fact that the occurence of bandwagon effects seemed highly
dependent upon moderator technique appears to bear out this theory.
The Lang Study
Such explanations remain conjecture, however, and it would
therefore do well to examine other attempts to verify the existence
of bandwagon effects. One report which has already been described is
that of the Lang study on the effects of broadcasting early returns.
This study refers to bandwagon effects as part of the 'political
folklore' often cited by individuals as reason for their opposition
to broadcasts of early returns. The study concludes, however, that
"...the impact of early returns on behavior was minimal.
Very few examples of bandwagon and underdog effects, of
either the direct or indirect variety, were encountered.
The few we were able to document had little to do with
what respondents had heard about the presidential race
prior to voting."[4]p.109
The report discusses possible reasons for this result and
attributes the relationship between early returns and voter reaction
1 [4]p.4
to what is referred to as the 'law of minimal consequence
According to this theory, voter intentions crystallize gradually
over a period of time preceding the election. Because the effects
of mass communication are essentially cumulative in nature, the voter
can only absorb new knowledge, such as early returns broadcasts,
in terms of the overall information structure already built up in
his mind. Thus
-70-
"t. ..on election day, when the returns begin to come in,
few members of the electorate are still open to influence...
Had there been any significant bandwagon effect...it would
already have had its impact on vote intentions before
election day."[4]p.4,5
At this point, a distinction between the dynamic polling procedure
employed in cable participation sessions and the election process
investigated in the Lang study should be made. Votes taken during
Community Dialog sessions are not completely analogous to election
votes for two reasons. The first is that because questions are often
repeated, issue votes are not irrevocable as election votes are.
Secondly, in issue exploration sessions, final outcomes are the
result of an extended series of votes rather than of one single vote.
It is this feature which may in fact account for the bandwagon
effect encountered in the Community Dialog meetings in a way consistent
with the Lang theory of minimal consequences. Throughout the Community
Dialog sessions, no really dramatic shift in group preferences was
observed as the result of any individual vote. Instead, the effect
manifested itself as a gradual shift occurring over the course of the
entire series of votes. This suggests that sudden manipulatory effects
similar to those discounted in the Lang study were not needed for the
outcome of the sessions to change. Instead, the bandwagon effect
had made its impact felt during the gradual process of attitude
formation. The conclusions on vote stability made by the Lang report
for the case of election outcomes may have to be altered considerably
-71-
in the case of interactive polling due to the introduction of
the dynamic element characteristic of such processes.
The Asch Experiments
In addition to theories such as the 'law of minimal consequences',
researchers have posed other explanations for the inability to observe
bandwagon effects. Both Lang and Fleitas[231P.436 cite the
results of past social-psychological studies of conformity as a
reason for discounting the existence of such effects. In this, they
[24] [25]
are both referring to the Asch experiments of the late 1950's
These experiments focused upon disagreements generated between
individuals and groups over matters of simple fact. In an introduction
to one of these studies, Asch states that he was concerned with
... public independence and lack of independence in the
face of arbitrary group opposition. The aim was to
observe the impact of these conditions when the question
at issue was that of resisting or bowing to a prevailing
group direction!'[24]p.2
To] the individual subject, the experiment was ostensibly designed
to measure visual perception. A group of college students would each
he asked in turn to judge the lengths of several lines displayed
before them. However, only one of these students was actually a
subject, the others having been told by the experimenters which
answers to give. In one experiment, all of these students were
instructed to give the same wrong answer so that the reaction of
the subject to unanimous opposition could be observed. Asch found
-72-
that in these cases, fully one third of the minority answers were
distorted toward the majority.[ 2 4 1p. 6 9  Moreover, the effects of group
pressure appeared to be greatly reduced in other experiments in which
[25]
the subject was opposed by less than a unanimous majority. Due
to the fact that unanimous majorities are not characteristic of most
elections, Land and Fleitas maintain that group influence would probably
not have a significant effect in those situations.
Although it certainly seems reasonable to draw such a conclusion
from the results of the Asch investigations, it should be kept in mind
that those experiments were performed under idealized conditions which
don't necessarily conform to those characterizing most elections. Asch
himself .explicitly listed these conditions and discussed the limitations
[24]p.66
which they impose on any interpretation of his results. 4 One
factor relates to the object of controversey itself. In the Asch
experiments, the object of judgement was clear and unambiguous; this is
hardly case in most disagreements that concern issues or candidates.
As ambiguity increases, voting takes on the character of minimal-
information elections so that we would expect the effects of group
pressure to increase accordingly.
The experiments can be distinguished from voting situations on
the basis of other conditions as well. Unlike most participants in
group controversy, the experimental subjects of the Asch studies had
independent access to the facts in question. In addition, there was
no possibility of restructuring facts as in most social disagreements,
-73-
so that the conflict situation occuring in the experiment was ir-
reconcilable. In fact, the character of the controversy itself was
unique due to the fact that it was not understandable.
All of these conditions point out the difficulty encountered in
trying to apply the results of social experiments to actual situations.
This is not to say that the lessons of the Asch experiments cannot be
applied to the issue of bandwagon effects. However, the significance
of differences in the conditions governing the two situations is not
yet clear. For example, a key factor in the Asch study was that
subjects were forced to declare their choices publicly before the
group. In the Community Dialog meetings, on the other hand, all votes
were taken anonymously. If anything, we would expect the condition
of anonymity to all but wipe out any effects of group pressure. And
yet, it was during these very meetings that such effects were observed.
It therefore seems that few if any real conclusions can be drawn
regarding bandwagon effects in interactive polling situations. Although
the existence of such effects have not been proved, they cannot be
entirely discounted either. The role of instantaneous feedback in
supplying audiences with prior knowledge of group preferences there-
fore remains suspect as a possible source of manipulatory effects.
-74-
II. Branched Patterns of Discussion:.Agenda Effects
The second characteristic of interactive polling processes that
was mentioned in the preceding chapter was that of branched discussion
patterns. Due to the fact that the dialog in cable participation
sessions must follow a series of questions with explicit audience
responses, a formal order is imposed on the discussion. This order
can often be represented as a branched discussion tree of the type
shown in Figure 4. Although audiences are usually able to determine
the direction of such discussions to some extent, it is the moderator
who maintains the greatest degree of control over the evolving
structUre. It is this fact which leads us to investigate the possible
role of these patterns in the manipulation of group outcomes.
One feature of the discussion tree over which the moderator
retains complete control is the sequence in which topics are taken
up for voting. Although the sequence in figure 4 appears to be the
most logical one, it is not hard to conceive of situations in which
any number of alternative agendas would be acceptable. Common
experience with various types of group meetings has shown that the
form of the agenda can often be crucial to the outcome of a discussion.
In a book entitled How Nations Negotiate, F.C. Ikle analyzes the
importance of agendas in international diplomacy.
) 4 4 ) I 4 I ) I I
Figure 4. Example of Branched Discussion Tree
Taken from (20) Fig. 4, this diagram represents the structure of a discussion on education.
The discussion took place during a 90 minute television program conducted as part of the
MIT Community Dialog Project.
GOALS OBJECTIVES OBSTACLES ACTIONS
Lack of Readiness Work with parents to
Enrich Pre-schoolers
Lack of Alternatives
Communications
Prepare for Jobs Home School-Home Communications
Student as Actor Nr rmMotivation More Drama
Well Roundedness
Inquiry Process Lack of Perceptsion In-Service Teacher Training
Self-Fulf illment Lc
Master Variety of .. Improve Student-Teacher Ratio
Training
Learning Skills
Learning to Learn LerigSls/Learingto Lar Flexibility of Group $ize
Student Set Attain- Inability to Read
Cope with Change able Goal/
C e t .aa oAccess to Resources Self-nstruction
Self-Reliance Know How to Use
Resources Inflexibility of
Cfjrriculumas Instructors
Understanding Thought Teacher Outside Involvement
Process Consider Teacher Tenure
Teachers and Evaluation
Altering Environment
for Teacher
pl%
.3
-76-
..Agreement on the agenda has two implications. First,
it means that prior to, or at the beginning of a conference
the parties agree on the subjects they will discuss. Second,
the agreed agenda constitutes a procedural agreement as to
the sequence in which the subjects will be negotiated. Both
of these implications affect the outcome of negotiations:
the subjects that are discussed obviously determine the
issues that may be settled, and the particular sequence in
which separable issues are negotiated may favor one side
and handicap the other."[26]
The issue of the agenda determining the topics to be discussed
appears to be relatively straightforward. In fact, this problem
has already been mentioned with regard to audience manipulation
in the discussion of the televote project appearing in Chapter Two.
Analyzing how the sequence of topics affects outcomes, however, seems
to be a more complex problem. Although the influence of agenda seems
to be a fact intuitively, to what extent has there been empirical
validation or theoretical understanding of this intiuition?
Empirical Research: The Work of Charles Plott
Charles Plott of Cal Tech has addressed this problem in the con-
text of group meeting having the task of choosing one alternative
from amon# many. There are several conclusions made in Plott's
paper; among them, the claims that,
"...first, the agenda or groupings in which alternatives
are considered for adoption or elimination can be a
major parameter in determining what the group will actually
choose. Second, the nature of the influence is sufficiently
systematic to yield to an analytical model."[27]p.2
-77-
The paper reports on several stages of an effort to develop
a formal theory of how the agenda of a meeting can affect its outcome.
The work began when Plott was offered an opportunity to draw up the
agenda for a special meeting of a flying club to which he belonged.
The purpose of the meeting was to decide on the composition of a new
fleet of airplanes which the club was to buy for the use of its
members. On the basis of a crude theory, it appeared that several
equally acceptable agendas would each favor a different outcome.
In the way of an experiment, Plott decided to devise an agenda which
would favor the outcome he desired personnally. The experiment turned
out to be a success with the result that work soon began on a formal
mathematical model. Finally, laboratory experiments were undertaken
for the purpose of validating the model. This attempt met with only
partial success, however, with the result that Plott comes to conclude
that "... even though our general theory may be right, the specific
means of expressing or modelling it that we have developed is
probably wrong5 [27.p.4O
A brief explanation of the flying club example would be highly
instructive for the purposes of understanding both the general nature
of the theory and its implications. The club was faced with a choice
of several types of aircraft from which to compose its new fleet
of six or seven planes. The four models being considered were:
1. Beechcraft Bonanza A36 (abbreviated: A)
2. Beechcraft Bonanza F33A (abbreviated: F)
3. Refurbished Beechcraft BonanzA E33A (abbreviated: E)
4. Cessna 210 (abbreviated: C)
-78-
From among these alternatives, Plott drew up his own list of prefer-
ences:
1st choice: 7 planes, EEEEECC or EEEFFCC
2nd choice: 6 planes, EEEEEC or EEEFFC
3rd choice: 7 planes, EEEEEEA or EEEFFFA
In order to design an agenda which would bring about the outcome he
desired, Plott first needed information on the preferences of the
group as a whole. These preferences were estimated as follows: [27]p.6
35% 35% 10% 3%
1st choice 6 or more E's 6 or more E's mostly A's mostly C's
and F's and F's
2nd choice several E's and several E's
F's with a few and F's with
C's (*) a few A's
3rd choice several E's and several E's and
F's with a few F's with a few
A's C's (*)
where (*) indicates Plott's own preference.
This breakdown was based on the assumption that most of the group
would prefer to continue flying Beechcraft (the type of aircraft which
made up the old fleet), but would also want less expensive alternatives
such as Cessnas (C) and Beechcraft E's. A small but influential
portion of the group prefered the more expensive A's, while the
governing board of the club wanted Cessnas, apparently because of
thier relationship to the local Cessna dealership. As can be seen
-79-
from the above chart of preferences, it seemed highly unlikely that
in the absence of any manipulation Plott's choices could be voted in.
Essentially, Plott's theory states that the agenda can influence
the way in which the initial set of alternative choices is partitioned
by a sequence of votes into successively smaller subsets. This
influence can be felt in two ways. First, because of the fact that
group preferences are revealed by voting and the contents of each
vote are determined by the agenda, it is possible to limit the
infromation available to individuals on the patterns of group pre-
ference. Thus there is little chance for the preferences of others
to influence individual voting behavior adversely. The second way
in which the agenda determins outcomes is be determining the set of
[2p. 14strategies available to the individual at any point in time.
A final element needed to implement this theory is a knowledge of
the decision rules by which individuals choose between sets of
alternatives. One hypothesis is that people vote for the set
containing their most preferred alternative, while another holds that
they tend to vote against the set containing their least preferrred
alternative. In devising the agenda for the flying club meeting,
Plott assumed that club members would follow only the first rule.
The fundamental task encountered in the designing of the agenda
was to order the five questions correctly. The sequence finally chosen
as well as the resultant pattern imposed on the discussion are
presented in Figure 5. The first question voted on concerned the
-80-
makeup of the primary fleet. The reason for putting this question
first was because if given the opportunity, most of the group would
have voted for a fleet made up entirely of inexpensive planes. The
next question concerned the desired number of planes to be bought.
Many members wanted six E's and F's and no A's or C's. By pitting these
members against those who wanted seven plane fleets, Plott insured that
there would still be an opportunity for choosing some A's or C's later
in the agenda.
The next question asked the group to decide whether they wanter
any planes of a type other than that already chosen for the primary
fleet. Since more members preferred a secondary fleet, the meeting
moved on to the decision of whether that secondar fleet should have
one or two planes. Due to the fact that the successful vote on a mixed
fleet had already ensured a margin large enough to vote in at least
two A's or C's, this question was also decided in Plott's favor.
Finally, the type of aircraft in the secondary fleet was also choosen
according to plan because the already high cost of a seven-plane fleet
had compelled the group to select C's over the more expensive A's. In
the end, therefore, the agenda has achieved its intended purpose.
One of the most interesting things to be learned from the meeting,
however, centered around the behavior exhibited by the chairman.
The meeting chairman was himself determined to see to it that the
final choice included at least some A's. He therefore made several
attempts to change the agenda during the meeting in order to achieve
) )
)
4A 4 11 ) ) I)
STRUCTURE OF FLYING CLUB MEETING
Question 1:
Type of Pri-
mary Fleet.
(original set
of alternatives)
Question 2:
How many
planes total?
Question 3:
Planes of
Type other
than in Pri-
mary Fleet?
Question 4:
How many
planes in
secondary
Fleet?
Question 5:
Type of planes
in secondary
Fleet?
EEEFFF no EEEFFF -
EEEFFC
6 EEEFFA CS E
planes 
EEEFFCs
yes As
EEEFFF EEEFFA EY_
EEEFFC EEEFFFC C'S....... EEEF FFC
EEEFFA 
EEEFFA A
EEEFFFF JEEEFFFFEEFA
77 EEEFFFF no
EEEEFFC planes EEEFFFC
EEE CEEEFFCC* <EEEFFFC Zone IEEEFFCC*
EEEFFAA IEEEFFAA yes EEEFFFA EEEFFCC* C
EEEFFCC* two EEEFFAA
EEEFFAA 
EFA
*indicates Plott's prefernce
- indicates path actually taken in meeting (Diagram adapted from (27) p.10, Fig, 3)
Figure 5.
00
H
)4 ) ) 4 ) 4 ) 
) )
Figure 6. Chairman's 2nd Attempt to
Change Agenda
Figure 7: Chairman's 3rd Attempt to Change Agenda
QUESTION 1: QUESTION CHANGE:
Type of Primary Fleet A's or C's in
Secondary Fleet?
QUESTION 3:
Planes of Type
other than that
in Primary Fleet?
QUESTION CHANGE:
Do we want at least
one A?
EEEFFF
EEEFFC
EEEFFA
EEEFFFF
EEEFFFC
EEEFFFA+
EEEFFCC*
EEEFFAA+
EEEFFA+
A's a EEEFFFA+
35% + 10%
EEEFFA+
C's
35% + 3%
EEEFFC7
EEEFFCC*
EEEFFFCJ
EEEFFFFJ
no
EEEFFFA+
ys EEEFFAA+l
35% + 10%
ys EEEFFC
EEEFFFA+
EEEFFCC*no
EEEFFAA+ n0 EEEFFCC
EEEFFC
* indicates Plott's preference
+ indicates Chairman's preference
-> indicates path of discussion if change were to be made in agenda
N3
-83-
this goal. Plott, however, managed to get each attempt ruled out
of order. The first such attempt came as a call for a vote at the
beginning of the meeting on whether or not an all Beechcraft fleet
should be chosen. Had this vote actually been taken, it would
almost certainly have passed ensuring the inclusion of at least
one A in the final outcome.
The other attempts to change the agenda were equally skillful.
Before the size of the fleet was to be decided upon, the chairman
called for a vote on whether A's or C's should make up a secondary
fleet. At this point the group would have chosen A's with the result
that the ultimate choice would have been a seven-plane fleet that
included one A. This situation is illustrated in Figure 6. Before
a vote was to be taken on the number of planes in the secondary fleet,
a final attempt was made to change the agenda. In this case the chairman
asked whether there should he at least one A. At this point the vote
would again have been affirmative as shown in Figure 7. Plott finally
observes that
"... the remarkable thing about the chairman s behavior is
that agenda rigging is not easy business-for us at least.
There are thousands of competing agendas. Yet he seems to
know which ones were to his advantage. Not once did he
make a mistake, at least according to our theory. Could
he have instinctively understood the thing that we had so
laboriously arrived at?"[271p.13
While it is true that future cable participation sessions may
not always be direct real-world analogs to meetins such as the one
-84-
described above, the implications for cable meetings are nonetheless
clear. If it is indeed true that the sequence in which questions are
asked can affect outcomes to the extent that intuitive knowledge of
the process is available,then the agenda of a meeting must certainly
be considered as a possible source of significant manipulatory effects.
-85-
III. The Structure of Communications: Network Effects
A major result of the simulations reported earlier in Chapter
Three is the finding that additional channels of information must
be utilized if large scale dynamic polling Processesare to be feasible.
The simulations clearly indicate that if cable participation programs
come to rely solely on the limited response capability of pushbutton
home terminals, breakdowns in communication become inevitable. There
are several ways to circumvent this problem, however. One method
with which we are already quite familiar is the use of telephone
call-ins during the broadcast. In addition, live interaction between
the moderator and a studio audience may also serve to clarify the
desires and frustrations of the audience. Finally, several of the
proposed NSF experiments suggest the possibility of rIaighborhood
communication centers where full video upstream transmission capability
would be available on a limited scale.
If we consider combinations of these supplementary channels, the
number of possible physical configuations for future cable participation
pystems seems quite large. It therefore seems reasonable to ask
whether variations in the communication structure of these configurations
might possibS affect audience behavior. One area of research contained
within the social-psychological literature does in fact suggest the
existence of a relationship between networks of communication and group
behavior. [28] These investigations have taken the form of laboratory
-86-
studies in which the number and pattern of communication channels
linking the members of small groups have been taken as independent
variables. The hypothesized relationships have included a number
of dependent variables such as task performance, morale, satisfaction,
and influence, any of which may have some possible implication for
the problem of manipulation.
The relevance of small group studies to large-scale participation
programs becomes clearer upon examining several probable configurations
for such systems. Several of the communication structures which
characterize alternative cable participation systems are illustrated
in Figure 8. Figure 8-a represents the situation in which the system
consists of only a moderator transmitting from the point of program
origination at the cable's head-end and the members of the home audience.
Full audio and visual communication channels and face-to-face communi-
cations are depicted in the diagram as solid lines, while digital
transmission is indicated by dashed lines. Figures 8-b and 8-c
illustrate networks resulting from the introduction of studio audiences
and neighborhood terminal locations; obviously, any number of other
arrangements are possible as well.
The three patterns shown in Figure 8 are examples of what will
be referred to as 'full-communication' networks. That is, they imply
that all participants have unlimited access to broadcast time, and
that any communication initiated by a participant is transmitted
directly over the cable. In other words, the program moderator has
-87-
Figure 8. Several Alternative 'Full-Communication' Networks
8-a 8-b
Moderator
Home Audience
Moderator
Studio Home
Audience Audience
8-c
Studio Moderator
Audience
Home Audience
Audio-Visual Channel
Digital Channel
Neighborhood
Terminal
Location
Single Moderator
- --Audience Segment in whichci-. all participantsO separated
Audience in which members
not separated
-88-
little or no control over the communication of other participants;
he can neither restrict the use of channels nor intercept messages
in order to relay them himself.
Such a characterization of communication structure is undoubt-
edly unrealistic. However, by describing alternative physical con-
figurations in terms of 'full-communication' networks, the differences
in potential patterns of communication are emphasized. If the nodes
of these networks are thought of as individual members of small groups,
Figure 8 can then be considered to represent different patterns of
small group communication.
The differences in patterns of communication that possibly hold
some implication for cable systems go beyond differences of physical
configuration alone. If the assumption of 'full-communication' is
altered, a large number of variations in communication structure
become possible for any one physical configuration. In such 'restricted-
communication' variations, the flow of communication in any of the
original channels is controlled by the moderator who acts as a gate-
keeper. Such control is usually considered necessary to the success
of group participation meetings. However, if exerted beyond the
level required for such success, manipulation problems inevitably
arise.
In order to illustrate this, Figure 9 depicts several 'restricted-
communication' variations of the full network appearing in Figure 8-c.
In 9-b for instance, the moderator relays all communication himself
with the result that the separate segments of the audience can no
-89-
'Restricted-Communication' Variations For A 'Full' Network
Studio Modetator
Audience
Home Audience
Neighborhood
Terminal Location
9-a. Full Network
9-b. Moderator acts
as gatekeeper
9-c. Segement of Audience
isolated
Figure 9.
-90-
longer contact each other directly. The moderator is even able to
selectively isolate particular sections of the audience as in
Figure 9-c by shutting off out-going communication. This situation
is especially likely when audience segments are not representative
of a cross-section of opinion, but instead possess distinctive
viewpoints. The major concerns of the group psychology studies,
however, are with the more subtle effects of network structure on
group behavior.
Early Network Studies
The earliest discussion of such effects appears in the work of
Alex Bavelas.[ 2 9 1 Although most patterns of communication emerge as
the result of social processes, Bavelas chose to focus upon fixed
patterns. Such patterns may be imposed by a larger organization (as
Bavelas himself had in mind) or by the actual technology of communication as
in the case of cable participation systems. In a paper appearing in 1950
Bavelas raises the question of
"...how a fixed communication pattern may affect the
work and life of a group... It may be that among several
communication patterns, all logically adequate for the
successful completion of a specified task, one gives
significantly better performance than another. What
effects can pattern, as such, have upon the emergence
of leadership, the development of organization, and
the degree of resistance to group disruption?"[29]p.726
Bavelas hypothesized that the mechanisms of such effects might
be grounded in the geometric properties of patterns of communication.
Specifically, he felt that differences between patterns might account
-91-
for differences in the behavior and performance of groups, while
geometric differences within patterns might account for differences
between individual members of the group.
The geometric concept which serves as the basis for these
differences is the notion of 'distance'. If the internal distance
between two members of the group is defined as the number of links
that separate them,then the sum of these distances for all positions
in the net can be used as a measure of the pattern's dispersion. This
measure of dispersion could then serve as a basis for comparison
between patterns. Bavelas also states that comparisons between
positions in the same pattern can be made on the basis of the
'realtive centrality' of each position. This quantity is defined
as the ratio of the sum of internal distances for all positions to
the sum of distances from any particular position.[29]p.72
Among the first of the experimental studies to investigate
Bavelas' geometric hypotheses was that conducted in 1951 by Harold
[30]
H. Leavitt. Using an experimental procedure which was to later
serve as the model for many future investigations, Leavitt employed
a characteristic type of group task. Each member of the group
was supplied with a list of symbols, only one of which appeared on
the other members' lists. The group was then faced with the problem
of discovering the common symbol. The subjects communicated by
passing written messages to each other through slots in the partitions
that separated them. By selectively closing some of these slots,
-92-
Leavitt was able to form four communication nets, diagrams of which
appear in Figure 10. These nets, as well as the individual positions
within the nets served as the independent variables of the experiment,
while task performance measures, questionnaires, and content analysis
were used to study the dependent variables of interest,
Based on the results of these experiments, Leavitt concludes that
differences in behavior can in fact be attributed to factors
relating to the structure of communication in groups. In addition,
the chief correlate of these differences does appear to be an operational
notion of centrality. In conclusion, he suggests that centrality
affects behavior by imposing limitations on the independent action
of individuals within the net.This leads to the prediction that
"...Where one position is low in centrality relative to
other members of the group, that position will be a
follower position, dependent on the leader, accepting
his dictates, falling into a role that allows little
opportunity for presitge, activity, or self-expression."[30]p.50
Investigating Influence Relationships
Although of obvious interest, Leavitt's conclusions are essentially
very general in nature. As a result, his study was eventually followed
by almost twenty years of additional study by other experimenters.
During this period, interest has shifted from variables intended as
measures of performance to psychological indicators of influence,
leadership, and group morale. The work of Goldberg[ 3 1 1  and of Shaw,
Rothschild, and Strickland{3 1 1 are of particular interest as they
-93-
Figure 10. Communication Nets Used in Leavitt Study
Chain Circle
IV Wheel
-94-
investigate the patterns of influence relationships within communication
networks.
Citing Leavitt's conclusions on the role of centrality in networks,
Goldberg hypothesized that the more central the position held by a
group member, the less that individual would be influenced as the
group reached its decision. In addition, Goldberg also believed
that the greater the centrality of a position, the more likely
it would be that the person occupying it would be perceived as a
leader by other members of the group. The networks employed in the
experiment were the 'chain', 'wheel', and 'Y' patterns of the Leavitt
study. It had been Goldberg's intention to show that his hypothesized
relationships could be demonstrated for all three networks.
Designed to yield an operational measure of group influence,
the experimental task required that each individual judge the
number of dots appearing on a card held up before him. By passing
messages to each other, group members had to reach a common conclusion
on the number of dots that had appeared. The initial guess of each
individual minus his final guess served as a measure of the extent
of group influence. In addition, information on leadership attributes
was obtained by polling each of the subjects individually.
Goldberg's results confirm the relation reported by Leavitt between
position centrality and leadership perception. This confirmation seems
particularly strong in light of the fact that the group task used by
Goldberg differed considerably from the one used by Leavitt. However,
-95-
the experiment does not appear to confirm the hypothesis concerning
centrality and influence. Goldberg himself, though, sums up his
results in the following way:
...Our results can be considered an indication of
the importance of situational factors operating in
influence and leadership phenomena. We have shown
that one's position in a communication network has
a bearing on the extent to which'one is influenced
by one's group in decision making. However, the
hypothesis we proposed was confirmed only in the
case of one type of network, the fork.[31Sp.121
Rather than reject the hypothesis for the reason that only
partially confirming evidence had been obtained, Goldberg assumes
that some additional variables must have intervened in the case of
the 'chain' and 'wheel' patterns. He therefore concludes that more
experimentation is necessary in order to explore such a possibility
in a greater variety of networks.[111p. 2 1
The conceptual perspective of Shaw, Rothschild, and Strickland
differs slightly from that of Goldberg in that they focus upon the
relation between net position and the ability to maintain a deviant
opinion. Employing a group task and influence measure that resemble
those used by Goldberg, the Shaw study attempts to verify the existence
of one of two theoretical possibilities. If it aan be observed that
those in central positions change their opinions in the face of opposi-
tion more readily than those in peripheral positions, then it would
be reasonable to assume that those in central positions are more
vulnerable to group pressure. If on the other hand, the center were
-96-
to change less readily, then it would be likely that those individuals
use their positions in the net to convince others.
Shaw, Rothschild and Strickland divided their investigation into
two separate experiments. In each experiment, only one network was
used; a four-man 'Y' pattern. In the first experiment, all four
positions were filled by subjects who were permitted to change their
opinions. The experimental results show that when all other positions
disagreed, the center position changed opinions as often as any of
the peripheral positions. However, when the group was evenly divided,
the half made up of the center and one peripheral position changed
opinions less frequently than the other two positions did.
Shaw considers this second result to correspond to Goldberg's
finding that more central members of the group are influenced less
as a decision is approached.[3 2 Ip.329 However, he also raises the
question of whether this lack of change on the part of the center
position is due to the fact that:
1) only two opponents remain,
2) a supporter is present, or
3) there is non-unanimous opposition.
An answer to this question in the context of communication nets is
certinaly relevant to the previous discussion of the Asch experiments.
This fact seems to suggest the possibility of interaction between the
influence characteristics of communication nets and the phenomenon
of bandwagon effects.
-97-
Shaw's second experiment was intended to shed light on the
reason behind the center position's relative lack of opinion change.
In this experiment, only one member of the group was actually an
experimental subject; the other three were plants instructed by
the experimenters as in the Asch experiments. Aside from position
in the net, the degree of group support was also used as an independent
variable. Among the conclusions which the investigators draw are
that:
"... (a) The central S more than the peripheral S tries
to change the opinion of those who disagree, but if he
fails, he himself changes more, presumably because of his
greater vulnerability to direct pressure, (b) the presence
of one supporter strengthens the resistance of an S
relatively more than the mere reduction of size of opposition
and more than the simple fact that the opposition is not
unanimous ."[32]p.330
Net Studies in Persepctive
The confirmation of Goldberg's conclusions and the finding that
the presence of a supporter (which in the case of cable participation
systems might be represented by one of the several audience segments)
strengthens the influence of the central position lends credance to
the idea that communication structure can serve as a possible source
of manipulatory effects. However, the implications of both these
studies are not this clear-cut. The fact of the matter is that
Goldberg found confirming evidence only for the case of the 'Y' network.
As a result, he was forced to postulate the existence of some intervening
variable to explain the negative result obtained in the cases of other
-98-
patterns. While Shaw does claim to have confirmed Goldberg's
positive result, the confirmation was also made only for the case
of the 'Y' network. Other networks in which additional intervening
variables might play a role were simple not investigated.
When viewed as a body, all the communication network studies
seem to display a similar sort of ambiguity. In an article reviewing
the first decade of research in this area, Glanzer and Glaser[28 ]
state that the two fundamental questions originally posed by Bavelas
concern the effects of communication structure on group efficiency
and morale respectively. After examining the results of the
experimental studies, however, the occurence of ambiguous and even
contradictory conclusions lead them to state that:
"..There is no simple answer to the first question.
The effect of structure depends in part on the
requirements of the task. Contrary to Leavitt's
original generalizaton, in a number of studies the
highly centralized structures are less efficient
that other structures."[28]p.18
in 1968, Robert L. Burgess conducted a series of experiments
with the intention of discovering variables capable of resolving
the contradictions of earlier studies.[3 3 1 Hypothesizing that
motivation and learning behavior play important roles in group
performance, Burgess increased the number of consecutive trials
in his experiment to approximately 1000. (This is in comparison
to the 25-60 trials typical of previous experiments.) In addition,
he also introduced reinforcement as another independent variable.
-99-
The results of the experiments reveal a learning curve in which
differnees in efficiency between nets disappear in a progression
with time. Burgess therefore concludes that
"...previously asserted 'differences' in solution
rates between communication structures, in which
there were no physical limitations favoring one
network over the other, were a function of experi-
mental artifacts. Had previous experimenters in-
cluded reinforcement contingent upon performance,
and had they observed their experimental groups over
sufficient time periods, the collection of a vast
array of contradictory findings could have been
avoided."[33]p.335
In light of this result, are we not to conclude that communi-
cation structure cannot really serve as a source of manipulatory
effects after all? At this point in time, the answer must remain
unknown. Burgess's results don't force us to reject such a re-
lationship outright for the reason that his conclusions only con-
cern the issue of efficiency and not that of influence. In addition,
it appears that any reluctance to draw general conclusions from the
earlier studies would be well applied in this case as well. This
is because important differnces exist between the experimental
situations described in these studies and the probable reality
of future cable participation systems.
The greatest of these differences lies in the nature of the
group task itself. It would certainly be difficult to generalize
results based on the performance of 'Leavitt-type' tasks to the
case of complex decision processes. Task complexity has in fact
been made a variable of interest in several studies. Aside from
-100-
being examined in the Shjaw, Rothschild, and Strickland article,
this variable was also the focus of an experiment conducted by
[34]Shaw in 1954. In addition, it is also an issued of concern
in an article written by Heise and Miller.[35] The Heise and
Miller study also pointed out other differences which happen to
exist between the common experimental context and our proposed
reality of two-way cable systems. Specifically, they investigate
the effects of channel noise and one-way channels on the performance
of different networks. Finally, Macy, Christie and Luce have
studied the role of semantic noise on group performance as well. 3 6 ]
In order to draw any solid conclusions on the role of communication
structure in the generation of manipulatory effects, it is therefore
necessary to conduct experiments within the actual context of
cable participation systems. Previous studies may suggest the
possibility of manipulatory effects, but because of the differences
and ambiguities contained in them, they can in no way serve as a
guide for the prediction of those effects.
-101-
CHAPTER FIVE
CONCLUSIONS
The mechanisms by which manipulatory effects may arise in
future citizen participation programs are almost certainly dependent
on the cable technology used to implement them. The preceding
discussion has focused upon several of these characteristics as
potential sources of manipulatory effects.
Observations made during the course of the MIT Community Dialog
Project have served as a basis of conjecture for some of the relevant
aspects of cable technology. In particular, these judgements are
based upon an investigation of the likely configurations of future
large-scale cable participation systems. This investigation indicated
that such systems would most probably utilize digital signals for up-
stream transmission in a manner similar to the operation of the
Community Dialog apparatus.
Although not designed for the purpose of investigating the
manipulation problem specifically, these simulations were nonetheless
able to serve as useful inputs to this study as they pointed out the
existence of three characteristics of probable large-scale cable
participation systems. These three characteristics, continuous
feedback, formalized patterns of discussion, and the necessity
of supplementary channels of communication, were in turn analyzed with
respect to their potential as sources of manipulatory effects.
-102-
After examining literature pertinent to the task of establishing
whether or not the suggested manipulatory effects may in fact exist,
it was concluded that each of the above three characteristics remains
suspect as the basis for several possible mechanisms of manipulation.
Within the context of dynamic polling processes used to approximate
group discussion, it has been seen that continuous feedback of group
preferences may give rise to 'bandwagon effects'. In addition,
formalized patterns of discussion can result in a situation in which
the structure of the agenda used in such meetings may have a
significant influence on group outcomes. Finally, it has also been
suggested that the structures of communication imposed by cable
systems as a result of the necessity to include various arrangements
of supplementary channels might, generate characteristic sets of
influence relationships.
But just how likely are these possibilities? Even after the
brief examination of the literature likely to pertain to each of
these three cases, the question still remains unanswered. With regard to
the possiblity of bandwagon effects and to the implications of
different communication networks, the evidence appears to be contra-
dictory. And, although the influence of agendas on the determination
of group outcomes has been demonstrated, it is not clear that cable
participation sessions would in fact be real-world analogs of the
experimental situation.
-103-
Thus the most that can be done in a paper such as this is to
speculate on the possibilities of manipulatory effects rather than on
the probabilities of their actual occurence. The phenomena of
interest are sufficiently complex to prohibit widespread generalization
and extension of existing experimental results. It should therefore
be apparent that only experimentation conducted within the specific
context of two-way cable participation systems can be of assistance
in assessing the actual likelihood of various manipulatory effects.
I. Suggestions for Further Research
In carrying out such research, there are several issues of
experimental concern which should be kept in mind. Various factors
that might contribute to the existence of bandwagon effects have
already been mentioned in the preceding chapter. Among these are
qualitative audience stimuli, the dynamic qualities of the discussion,
the structure of communication, and the degree to which audiences
possess prior knowledge of the issues. Ideally, therefore, experi-
mental designs should control for the inclusion of these variables.
The design of Fleitas' experiment in which a series of successive
votes was taken among several different groups of similar participants
seems particularly suited for this purpose.
Due to the fact that no study has as yet disputed Plott's results,
the problems involved in investigating the role of agenda effects in
determining outcomes are not as numerous. However, the fundamental
-104Z
question that we do have in this case concerns the correspondence of
cable participation sessions to the experimental situation of success-
ively partitioning an initial set of choices into alternative subsets.
Future experiments should therefore concentrate on the problem of
determining the range of applications for which Plott's general
conclusions seem to hold true.
If anything, the third potential source of manipultory effects,
namely, the network characteristics of the communication system,
poses even greater problems for the implementation of practical
experimentation. The role of communication networks in the develop-
ment of influence relationships remains unclear despite almost two
decades of research. Factors such as the nature of the group task
and the characteristics of the communication channels must obviously
become part of any experimental configuration.
However, in addition to these factors there is one general issue
which, although relevant to any investigation of manipulatory effects,
is of special concern in the case of communication net experiments.
Simply stated, we are faced with the question of how to measure re-
lationships of influence. On the most fundamental level, one encounters
the theoretical problem of trying to define power and influence re-
lationships. It is generally argued that power relationships are
causal; for A to have power over B, A must cause B's behavior.
However, in accepting this argument, we must also accept the difficulties
involved in detecting causal relationships. [37]p.4l0,[38]
-105-
Besides basic theoretical issues, we are also faced with
practical problems of measurement. James G. March discusses several
advantages and disadvantages of the different techniques used to
measure influence. [391 The first method which he discusses is that
of attributed influence; that is, simply asking subjects to rank
each other in terms of perceived influence. While directly asking
subjects about how they were influenced does distinguish real from
psuedo-influence, it does have a number of other shortcomings.
Specifically, this method necessarily involves some degree of dis-
tortion, and in addition, provides no basis for comparison between
ranking systems.
The second method which March discusses involves the use of
before-and-after measures of opinion change. This is the technique
employed in the Goldberg and Shaw studies. While this method, unlike
measures of attributed influence, is external to the individual, it
does require that answers be given on a continuous scale. A problem
then arises as to the meaning of such units. In addition, overt
measures often fail to distinguish between real and apparent influence.
Finally, one other approach would be to count attempts at influence
using the techniques of interaction analysis. Interaction analysis
breaks down all observable behavior and communication into a number
of well-defined categories. During experimental sessions, trained
observers record the occurences of each type of event. One of the
problems with this approach is the equality assumption on the impact
-106-
of message units. Another is the fact that when analysis is
directed towards interaction, it often tends to ignore unobservable
events.
In light .of the advantages and disadvantages associated with
each of the methods used to measure influence, it seems reasonable
to conclude that a combination of techniques should be incorporated
within the framework of future experimental studies.
II. Manipulation Effects and the Process of Technology Assessment
Undoubtedly, the results of experiments and simulations conducted
within the specific context of citizen participation applications of
two-way cable can be of great benefit in the prediction of higher-
order technological effects. Even so, however, we should realize
that the boundless complexity of social and political realities
makes absolute confidence in the extrapolation of such results
an impossibility. The only method by which unequivocal knowledge
of these effects might ever be obtained is the actual observation
of functioning systems.
This fact does not lessen the significance of speculative studies,
however, for they are an essential part of the overall process of
technology assessment. By anticipating possible problem areas,
they permit us to fotus attention directly on the aspects of
technological systems most likely to be of importance. In his article
on prediction, Bell states that:
-107-
the correct identification of relevant problems
is obviously the first step in the conjecture about
the future; it is easier to make because it tends to
be an extrapolation of the present."[31p.869
Because such extrapolations are dependent upon present reality,
it seems reasonable to conclude that the process of problem identi-
fication does not remain constant over time. Different problems
can obviously be identified before and after implementation of any
particular item of technology. However, the process of technological
innovation does not consist of these two stages alone. James Bright
describes the innovation process in terms of eight stages occurring
within the technological environment, each of which may be dependent
on factors deriving from non-technical environments.[ 2 ]p 7
Present reality changes with each successive step of the innovation
process, and for this reason, problems can continually be perceived
in new ways. Thus, prediction of the higher-order effects of technology
must of necessity be a gradual process of speculation, simulation,
and experimentation in which complete knowledge crystallizes only
in the final stage of the innovation process.
With respect to the introduction of cable participation systems,
we now find ourselves between stages four and five in Bright's out
line of the process of technological innovation. .7That is, we
are at the stage of technological development in which laboratory
demonstrations have already been completed, but full-scale field
trials have only just begun. It seems clear that it will be
-108-
some time before the technology comes into actual widespread use
within society. It is therefore also true unfortunately that we
still have some way to go before any practical knowledge of the
problems of manipulation can be obtained. But this fact should not
make us particularly uneasy, however; for as the Lang study notes:
"...Any innovation in communication-the admisstion of the
press to legislative debates, the broadcasting of political
party conventions, the televising of presidential press
conferences, etc. -has invariably invoked alarm. Some
change always accompanies such innovation, but the potential
danger to political institutions is, in retrospect, almost
always found to have been exagerated. The process by which
individuals react and the remifications of their reactions
for the workings of institutions is always more complex, and
hence the effects far more subtle thanthe alarmist can
anticipate. Indeed, the high pitched sense of alarm may
itself be on of the firmest guarantees against sudden disjunctive
change."[4]p.163.
Although knowledge of the changes brought about by technology can
only be obtained gradually, such changes are usually too complex to
result in far reaching impacts on society in only very short periods
of time. Thus, we have the possibility of anticipating problems
in time to direct observation to the more critical aspects of tech-
nological innovations. In this way, it may be possible to design
around problem areas as they are encountered.
The conclusion of this thesis is that three specific mechanisms
by which-manipulatory effects may be generated are in fact distinct
possibilities for future large-scale two-way cable participation
systems. Based on currently available information, there does not
-109-
appear to be any way of determining the likelihood of these event-
ualities. However, we do have the ability to design experiments
which can determine the existence of these mechanisms within the
context of the actual systems as they come into use. Hopefully,
the results of such investigations will enable us to deal with
what might have otherwise remained important but unforseen higher-
order effects of, the technology.
-110-
APPENDIX
-111-
I. REFERENCES
1. Bauer, R.A., Second-Order Consequences: A Methodological Essay
on the Impact of Technology; MIT Press, Cambridge, Ma. 1969.
2. Bright, J.R., The Process of Technological Innovation-An Aid
to Understanding Technological Forecasting: in: Bright, J.
and Schoeman, E (eds) "A Guide to Practical Technological
Forecasting", Prentice Hall, Englewood Cliffs, N.J. (1973).
3. Bell, Daniel, Twelve Modes of Prediction - A Preliminary Sorting
of Approaches in the Social Sciences; Daedalus, V. 93, Sept. 1964,
p. 845-80.
4. Lang, Gladys and Lang, Kurt, Voting and Non-voting: Implications of
Broadcasting Returns Before Polls Are Closed, Blaisdell Publ. Co.
Waltham, Ma., 1968.
5. Jurgen, R.K., Two-Way Applications for Cable Television Systems in
the '70's, IEEE Spectrum, Nov. 1971, p. 39-55.
6. Baer, W.S., Interactive Television: Prospects for Two-Way Services
on Cable, The RAND Corp., R-888-mf, 1971
7. The Urban Institute; Design Study for Urban Telecommunications
Experiments, proposal to the N.S.F. , 1975.
8. University of Southern California; Implementing Two Urban Tele-
communications Experiments Designed to Deliver Municipal Services,
proposal to the N.S.F., 1975.
9. Denver Research Institute, University of Denver, Design Study for
Urban Telecommunications Experiments - Summary Report, proposal to
the N.S.F., January 1975.
10. New York University, Experiment in Urban Telecommunications; proposal
to the N.S.F., 1974.
11. Lehigh University, Implementation of Urban Telecommunications
Experiments, proposal to the N.S.F., 1974.
12. Pool, I. de Sola (ed), Talking Back: Citizen Feedback and Cable
Technology, MIT Press, Cambridge, Ma. 1973.
-112-
13. Pool, I. de Sola, Citizen Feedback and Political Philosophy, (in
Talking Back); MIT Press, Cambridge, Ma. 1974.
14. Campell, V., The Televote System for Civic Communication: First
Demonstration and Evaluation, American Institutes for Research,
Palo Alto, Ca., 1974.
15. Guite', J.C.M., CATV Technology for Citizen Feedback to Government,
M.S. Thesis, Dept. of Politeal Science, MIT, 1974.
16. Jones, W.B., Riter, S. and Hambrick, R., Telecommunications in Urban
Public Services, IEEE Transactions on Communications, V23, N.l, Jan. 1975.
17. Bretz, R. and Dougherty, L.A., Two-Way TV Teleconferencing for
Government: The MRC-TV System, RAND Corp, Santa Monica, Ca,
R-1489-MRC, April 1974.
18, Reid, A., Communications Study Group: Interim Report, Joint Unit
for Planning Research, University College, London, England, July 1971.
19. Pool, I. de Sola, Politics in a Wired Nation, in Talking Back, MIT
Press, Cambridge, Ma. 1973.
20. Sheridan, T.B., MIT Community Dialog Project-Progress Report, July 1974.
21. Lemelshtrich, N., Feedback Technology and Small Group Interaction,
Interdepartmental Ph.D. Thesis, MIT, October 1973.
22. Simon, H., Bandwagon and Underdog Effects and the Possibility of
Election Predictions, Public Opinion Quarterly, V28, p. 245, Fall 1954.
23. Fleitas, D.W., Bandwagon and Underdog Effects in Minimal-Information
Elections., American Political Science Review, V65, p. 434, 1971.
24. Asch, S.E., Studies of Independence and Conformity: I. A Minority
of One Against a Unanimous Majority, Psychological Monographs, V70,
N9, p. 1, 1956.
25. Asch, S.E., Effects of Group Pressure Upon the Modification and
Distortion of Judgements, in EE. Maccoby et al, Readings in Social
Psychology, 3rd ed. Holt, Rinehard, and Winston, New York, 1958.
26. Ikle', F.C. How Nations Negotiate, Harper and Row, New York, 1964.
27. Plott, C.R. and Levine, M.E., On Using the Agenda to Influence Group
Decisions: Theory, Experiment and Application, Social Science Working
Paper, N. 66, California Institute of Technology, April 1975.
-113-
28. Glanzer, M. and Glaser, R., Techniques for the Study of Group Structure
and Behavior: II, Emperical Studies ofthe Effects of Structure in
Small Groupl, Psychological Bulletin, V58, Nl, p. 1-27, 1961.
29. Bavelas, A., Communication Patterns in Task-Oriented Groups, Journal
of the Acoustical Society of America, V22, N6, p. 725-30, Nov. 1950.
30. Leavitt, H.J., Some Effects of Certain Communication Patterns on
Group Performance, Journal of Abnormal and Social Psychology, V46,
p. 38-50, 1951.
31. Goldberg, S.C., Influence and Leadership as a Function of Group
Structure, Journal of Abnormal and Social Psychology, V51, p.119,
1955.
32. Shaw, M.E., Rothschild, and Strickland, J.R., Decision Processes in
Communication Nets, Journal of Abnormal and Social Psychology, V54,
p. 323-30, 1957.
33. Burgess, R.L., Communication Networks: An Experimental Reevaluation,
Journal of Experimental Social Psychology, V4, p. 324-37, 1968.
34. Shaw, M.E. Some Effects of Problem Complexity Upon Problem Solution
Efficiency in Different Comudnication Nets, Journal of Experimental
Psychology, V48, N3, p. 211-17, 1954.
35. Heisse, G.A. and Miller, G.A., Problem Solving by Small Groups Using
Various Communication Nets, Journal of Abnormal and Social Psychology,
v46, p. 327-36, 1951.
36. Macy, J,, Christie, L.S. and Luce, R.D., Coding Noise in Task-Oriented
Groups, Journal of Abnormal and Social Psychology, V48, p. 401-09, 1953.
37. Dahl, R.A., Power, article in International Encyclopedia of the Social
Sciences, V.2, Macmillan Co. and The Free Press, 1968.
38. Simon, H., On the Definition of the Causal Relation and Notes on the
Observation and Measurement of Political Power, in "Models of Man'
Social and Rational", Wiley, N.Y. (1957), p. 50-70.
39. March, J.G. , An Inttoduction to the Theory and Measurement of Influence,
American Political Science Review, V49, p. 431-51, 1955.
-114-
II. TELEVISION FEEDBACK EXPERIMENT QUESTIONNAIRE
(Second Stage)
SEGMENT(l): Single Moderator
1. Indicate the degree of participation which you felt during this
segment:
circle: no participation 1 2 3 4 5 much participation
2. Did you often feel a lack of communication during the segment?
circle: never 1 2 3 4 5 often
3. Was the pace of the discussion too fast or too slow?
circle: too slow 1 2 3 4 5 too fast
4. Did the discussion hold your interest?
circle: no 1 2 3 4 5 yes
5. Do you feel that the direction of the discussion was responsive
to the audience's interests?
circle: not responsive 1 2 3 4 5 responsive
6. Check off any question format techniques which you found particularly
satisfying:
yes-no , multiple choice , scale of choice
frustration category___.
7. Check off any question format technique which you found particularly
frustrating:
yes-no , multiple choice , scale of choice
frustration category_ .
SEGMENT(2): Two Moderators
1. Indicate the degree of participation which you felt during this
segment:
circle: no participation 1 2 3 4 5 much participation
2. Did you often feel a lack of communication during the segment?
circle: never 1 2 3 4 5 often
-115-
3. Was the pace of the discussion too fast or too slow?
circle: too slow 1 2 3 4 5 too fast
4. Did the discussion hold your interest?
circle: no 1 2 3 4 5 yes-
5. Do you feel that the direction of the discussion was responsive
to the audience's interests?
circle: not responsive 1 2 3 4 5 responsive
6. Check off any question format techniques which you found particularly
satisfying:
yes-no , multiple choice , scale of choice_
frustration category
7. Check off any question format techniques which you found particularly
frustrating;
yes-no , multiple choice scale of choice
frustration category
SEGMENT (3) Panel Discussion
1. Indicate the degree of participation which you felt during this
segment;
circle; no participation 1 2 3 4 5 much participation
2. Did you Qften feel a lack of communication during the segment?
circle; neVer 1 2 3 4 5 often
3. Was the pace of the discussion too fast or too slow?
circle: too slow 1 2 3 4 5 too fast
4. Did the discussion hold your interest?
circle: no 1 2 3 4 5 yes
5. Do you feel that the direction of the discussion was responsive
to the audience's interests?
circle; not responsive 1 2 3 4 5 responsive
-116-
6. Check of f any question format technique which you found particularly
satisfying:
yes-no , multiple choice , scale of choice
frustration category_
7. Check off any question format techniques which you found particularly
frustrating:
yes-no , multiple choice , scale of choice
frustration category_.
-117-
TELEVISION FEEDBACK QUESTIONNAIRE - SUNARY
1. Rank the single moderator, double moderator, and panel segments in
terms of your sense of participation, Explain.
2. Rank the single moderator, double moderator, and panel segments in
terms of the quality of communication you felt. Explain.
3. Rank the single moderator, double moderator, and panel segments in
terms of their ability to hold your interest. Explain.
4. Did group feedback make the discussions more satisfying for you?
Explain.
5. Did the group's feedback influence your response in any way? Explain.
6. Do you think your responses have differed had this been a large
face-to-face meeting? Explain.
7. Did you like this technique? Please make any comments or suggestions
which you might have. (Use reverse side).