Document 11070268

advertisement
DEWEY
HD28
ALFRED
P.
WORKING PAPER
SLOAN SCHOOL OF MANAGEMENT
ON-LIlsE GROUP MANAGEMENT PROCESS SUPPORT
Tony E. Gear, Balliol College, Oxford
Martin J. Read, Decision Dynamics, England
SSWP#3561-93
MASSACHUSETTS
INSTITUTE OF TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASSACHUSETTS 02139
on-liinE group management process support
Tony E. Gear, Balliol College, Oxford
Martin J. Read, Decision Dynamics, England
SSWP//3561-93
MASSACKUSEHS
INSTITUTE
OF TECHNOLOGY
NOV
1 4 2000
LIBRARIES
ON-LINE GROUP MANAGEMENT PROCESS SUPPORT
Tony E. Gear, Balliol College, Oxford
Martin J. Read, Decision Dynamics, England
1
INTRODUCTION
.
When alternatives have been debated by a group,
the members accept
and all
that in the circumstances a particular solution represents the best choice,
it is said that a consensus has emerged from the meeting.
participants
which
that
it
may
still
Minkes
1351
proposes
capable
is
decision-process
the
they
preferences,
policies.
in
of
able
be
not
do
agree
to
consensus".
agree
But,
actions
means
here
or
precisely
course,
of
in
personal
on
specific
on
"that good decision
securing
In situations
not
all
consensus decisions are "good" in terms of other criteria such as survival,
for example.
argument
This
terms,
is
most
"the
based
on
important
pragmatism
test
is
decision-making.
in
everyone
that
Adair
though it was their preferred solution".
1
13
In
prepared
is
However,
act
to
communication
the nub of the problem is
thorough debate can be very time-consuming and hard
as
Accepting this view,
.
true consensus may only emerge if sufficient and appropriate
has taken place within the group.
management
to
achieve
in
that
complex
situations.
People
come
into
the
different
group with
goals,
systems which emerge in the context of communication.
emphasised the importance of communication
own viewpoint.
in
perspectives
value
and
Various authors have
groups perceived from
their
This paper supports that view by describing a series of case
applications of on-line technology to support management meetings and groups
engaged on a wide variety of tasks.
The paper is centrally concerned with aiding social processes in groups,
the way to reaching decisions
given
to
group
the
process,
functioning 15,30,33,553
McCelland
The
as
C
defined
313
by
separated
from,
The contention is
group communication process is
decision
We argue that careful attention needs
.
a
Minkes.
but
related
to,
be
individual
that an appropriately organised
necessary condition for achieving a
This
to
on
philosophy
has
been
'good'
discussed
by
The use
of
in terms of a search for consensus.
importance of communication in groups is not a new theme.
on-line technology to aid group processes is quite novel.
In Section
3
we
describe
particular
a
form
group
on-line
of
technology
which
have
we
developed to support group processes, and in Section 4 we describe a series
studies
field
of
applications
of
of
All of the situations relate
settings.
system
this
to meetings
varied
in
managerial
with real consequences,
rather than to transitory laboratory settings with ad hoc groups.
Several
researchers have pointed out differences in conclusions drawn from studies
in contrived as opposed to in situ situations (e.g.
we describe all had important
For
describe
situations
involving full
varying
participants;
anonymity
has
)
The cases which
.
been
maintained.
We
communication between
face-to-face
synchronizations
time
81
implications for the organisations concerned.
reason a varying degree of
this
C
terms
in
the
when participants
of
submit responses during the meeting; variously, both group and/or individual
participant
communication
situations
and
goals;
with
rights,
where
member
one
group
having
members
share
additional
the
equal
role
of
based
on
chairperson or leader.
In Section 5 we
our
fieldwork
guidelines
develop a generalised process framework which
and
existing
the
hardware,
for
literature.
software,
meetings support systems.
and
framework
The
protocol
is
design
provides
room-based
of
Finally, we draw some conclusions in Section 6.
Developments in the area of group decision-aiding can usefully build on from
research
and
developments
in
number
a
multiple criteria decision analysis,
of
associated
particularly
fields,
group processes in social
psychology,
management oriented group techniques, as well as burgeoning capabilities in
information technology.
we briefly review related
In the following section,
work in these extensive areas.
RELATED WORK AND DISCIPLINES
2.
There is growing interest in Group Decision Support Systems
of
this
literature
has
been
concerned
laboratory
principles, and reporting trials in
(
29J
has pointed
implementations
design
and
out:
and
"Much
less
applications
implementation
of
a
establishing
with
discussion
of
a
GDSS
GDSS".
in
a
devoted
We
number
are
of
concepts
and
As Lewandowski
settings.
is
Much
(GDSS)*.
to
practical
the
concerned
actual
with
and
the
varied
International
the many
at
the
for example,
papers presented
Conference on Decision Support Systems (DSS) run annually from 1981 by the
TIMS College on Information Systems, Institute of Management Sciences,
Providence, HI 02093.
See,
decision environments,
methods
for
use
the
as
step
a
GDSS
of
the direction
in
different
to
relating designs
of
decision
contexts.
and
In
this
connection see, for example, 1261 for a study of taxonomies.
Decision-making in a group is
of
alternatives
and
of
attributes with which
commit the group members
relation
to
quantitative
and
framework
theoretic
achieve this it would be
individual
judgements
necessary
into
construct the means
to
collective
a
also
114,371.
Whilst
it
well
as
groups
in
result
There
not
is
as
exists.
aggregation
if
important
decision
to
search
analytic
To
fundamental
group
problems
See in particular Arrow
possible
meet
to
all
the
function,
useful
and
explicit methods
are
be
used
for
methods
reaching agreement in conjunction with GDSS.
a
aggregating
of
requirements of rationality and fairness in one group preference
it is nonetheless
to
represent
to
are
associated with this apparently laudable objective.
and
will
No entirely satisfactory
decision-making
for
preferences in a consistent and fair way.
133,
This
them.
factors,
consideration of imperfect information and risks.
decision
judge
to an analysis of their opinions and judgements in
both qualitative
to
complex process involving the consideration
a
to
an
as
aid
of
to
Some of the more promising of
these have been described and evaluated in (491.
The case studies described in Section 4 all depend on feedback of individual
judgements so that their differences and agreements are mirrored back to the
group.
raises
This
comparisons
interpersonal
measurement.
impossible
theoretical
of
utilities
1
is
crucial
Arrow's
to
proof
consistent
on
use
of
scales
of
the
interval
issue.
group
It
non-existence
the
functions
psychological
and
should be noted
preference
of
Dalkey
that
can
(141
formulated
be
of
this
shown
that
aspects
has
the
by
anchored individual preference scales with fixed reference objects.
practical viewpoint.
preferences
is
Churchman [121
based on
the
that one
from "peeking into" another's mind and that
inquiring system so that the
that once this position is
such
scalar
design
comparisons
constraint
rejected,
desirable
"The philosophy
argues:
notion
an
of
satisfies a set of defined axioms.
has discussed both formal
401
fundamental
based
with
The assumption that interpersonal comparision of utilities is
aggregation procedure which completely
Rapoport
concerned
issues
is
the
and
is
there
of
From a
subjective
fundamentally constrained
is
removed".
no
He
way
to
design
then points
problem can be defined as:
what
kind
would permit such comparisons?"
of
use
of
an
inquiring
an
out
"Are
system
group or committee based decisions emerge as an outcome
In practice,
Usually,
social process.
the group meets
in
of
a
order to reach decisions as
a
part of, or on behalf of, an organisation or institution of some kind.
personal
personalities
skills,
backgrounds
and
the cultural setting and history of the situation,
participants,
"Social Representations are
influence.
the
of
The
individual
all have an
important because they constitute
a
society's symbolic or mediated reality, which, once common currency, affects
not
communication
only
Himmelweit
of
t
241
"depends
on
available
personal
and
decision-making",
The process by which social representations,
.
interacting
societal
but
concepts
the
and
complexity
communication
evolve
images,
speed
and
media"
of
become
to
or networks
common
currency
communication as well as on
(Moscovice,
For
(361.
a
review
the
of
the
relationship between social representations and GDSS see 1181.
We believe that it is
the
design and support of the social process that is
the key to successful implementation of GDSS for a given group or committee.
The
framework
decision-theoretic
definition of
the
must
selection
short-comings,
that
way
consider
selected
be
process
social
the
the
must
is
decision
group
to
model
in
conjunction
with
be
supported.
This
and
the history of the group and its experience,
its
inevitable
the cultural and
political environment of the meeting, and the nature and timing of desirable
communications within the group.
support
We
the
position of Russo [441:
"When the intelligence-gathering phase for a major decision is complete,
the
evidence on the table should be rich enough that no one
her
views were
ignored".
When
disagreements
fundamental
feels
his
persist,
or
group
the
should seek to separate the "factual" issues and assumptions underlying each
side's position from their value judgements.
Viewed positively a decision model can provide a framework on which to base
each set of responses contributing to the eventual
a protocol of questions,
group
decision.
See,
for
models of decision analysis.
example,
17,22,28,51,531
The vital thing is
for
descriptions
of
that the protocol needs
to
expose adequately key areas of agreement and disagreement at each stage, and
identify
differences
of
understanding
these can be explored by debate,
of
identifying
key
the
problem
definition
and sometimes modified,
habitual domains become clear, 1551.
capable
of
areas
Furthermore,
for
group
as differences
the protocol needs
attention,
that
so
even
in
to
though
be
the
decision model may well be inadequate as the final arbiter of a decision.
It
is
unique
in
support
and
vital
of
the
role
communication process
to
play.
Keeney
1271
in
groups
put
it
that
GDSS
succinctly
has
a
in
a
.
description intended to promote good decision-making:
"...
analysis should illuminate controversy - to find
a good
in values and uncertainties,
out where basic differences exist,
to increase the level of debate and
to facilitate compromise,
undercut rhetoric...".
Once
view
the
adopted
is
that
can
GDSS
a
used
be
as
stimulant
a
to
"meaningful debate, it has the potential for use not only in decision-making
but
also
problem
in
formulation
suitable social process designs,
psychology
that we may
group
team
and
advocate"
it
techniques
"brainstorming"
1453;
to the
is
In particular,
turn.
based
generation.
idea
and
and
there are
"creativity"
well
the
inquiry"
developed
and
"devil's
154];"
"Delphi"
[42,46],
Interpretive Structural Modelling
"nominal group technique" 1161;
for
literature in applied social
"dialectical
of
searching
In
C
521
;
and
cognitive mapping 1171
Another aspect of a GDSS system is its potential to affect group processes
during,
prior,
and after meetings.
can
GDSS
A
support and enhance group
processes of mutual communication and understanding.
group
goal,
between
maintain
and
task-related
relationships,
individual
("instrumental
In order
suitable
a
behaviour
type")
to achieve a
balance
socio-emotional
and
("expressive type") behaviour needs to be maintained 15,101.
Disagreement
can produce tensions in the group which may impede progress on the task, and
these
tensions need
emotional
that
mechanism
behaviours.
types of
demonstrated
a
composite
An
release
their
for
expressive
through
experiment conducted by Gear [201
feedback
of
individual
scores
cognitive perceptions of performance and feelings during
a
or
has
relating
to
meeting can have
strong effects on subsequent behaviour.
Rohrbaugh [431
provided participants in experimental groups with cognitive
feedback of their individual judgements in order to focus their discussions
on possible representations of a uniform policy
for the group overall.
It
was found that groups were thought (by their members) to perform better when
individual cognitive feedback was provided to each member, freely exchanged,
and discussed.
In a series
of experiments with student groups,
[231
found
that "individual learning in tasks of cognitive conflict derives from direct
access
to
explicitly
graphic
representations
communicate
information".
With
the
respect
key
to
of
members'
differences
building
in
group
judgement
their
use
consensus,
policies
that
of
available
full
cognitive
feedback seemed sufficient to focus the groups,
without any imposed method
for structuring the discussions, so that groups were able to reach agreement
on a policy they could mutually endorse.
of this type with a group
Gear [211
of R & D managers
conducted an experiment
the pharmaceutical
in
industry
and received very favourable reactions from the group members.
In
classical
a
series
of
experiments
1
demonstrated
41
high
a
level
of
willingness by people in a group to "go along with" the expressed opinions
of the
are
It appears that the reasons for this at the process level
majority.
either
conform.
lack
a
confidence
of
judgement
personal
in
or
detected an effect called the "risky shift",
1481
desire
a
to
that on average
groups tended to make riskier decisions than the individuals involved would
make on their own.
See,
for a detailed account of social
for example, (501
influence and conformity phenomena.
the existence of majority
Clearly,
and
minority social influence processes in groups needs careful consideration in
designing
protocol.
a
In
"groupthink"
in managerial
organisation'
,
settings,
a
view
Mann 1321
on
removing
proposes
...
effects
"The
of
'learning
where effective group decision-making occurs in an arena of
informed and enlightened
action,
presenting
debate".
Additionally,
terms
in
of
management
it is generally important to bear in mind that full participation in
the decision process is often the key
to
commitment,
and
hence
successful
implemention [11.
The physical design of GDSS to support room-based groups has taken a number
of forms.
For example,
groups
which
in
each
[25,381
have described experiences with managerial
participant
is
with
provided
a
personal
computer
terminal, with all of the terminals linked by hard wiring to a large screen
display.
Experiments
with specially
[471
.
involving automatically
developed
hardware
and
mediated
software
have
dialogue
also
been
in
groups
described
COMMUNICATION TECHNOLOGY
3
DESCRIPTION
We now describe
each
of
the
the
system,
managerial
called TEAMWORKER,
described
situations
which was
in
Section
used
support
to
Some
4.
early
experiments and uses of this system have been described in [15,19,411.
member
Each
comprises a
of
the
group
provided
is
with
personal
a
two-line display and numeric keypad.
handset.
This
The keypad is styled
to
mimic a telephone layout and the overall appearance is that of a simple hand
calculator.
These features of the design are deliberate in
overcome emotional barriers
to
the
help
to
use of technology which a proportion of
intrusive nature of any technology.
minimise the
people do exhibit, and to
order
Numbers entered singly, or in combination up to a total of four digits,
be viewed personally, deleted and re-entered as desired by
the participant,
and finally "sent" as a digitalised message, using low power radio.
radio
of
is
an
advantage
as
it
unobtrusive
is
can
(wire-less)
,
The use
omni-
and
directional, with a 50 metre range.
The messages (digital signals) are received, as sent by each participant, at
a
receiver module which
is
hard-wired
convenient point away from the group.
to
micro-computer
a
located
at
a
The computer is arranged to provide a
large screen display so that there is direct visual feed back to the whole
group.
The design of
the
software
particular meetings or contexts.
the
generalised flow of
shown in figure
2.
The
is
IT
varied
to
suit
the
objectives
system is shown in figure
information between
the
system and
the
1,
group
of
and
is
Figure
1
HANDSETS
SCREEN
FEEDBACK
.^
|
(
CHAIR
~^
MONITOR
PRINTER
KEYBOARD
1
MICRO
RECEIVER
COMPUTER
MODULE
SOFTWARE
Figure 2
THE GROUP
RESPONSES FROM
EACH MEMBER
OF GROUP
VISUAL INFORMATION
FOR GROUP
VISUAL DISPLAY FOR
FEEDBACK OF PROCESSED
DATA, VISUAL PROMPTS
AND QUESTIONS
DISPLAY
APPLICATION MODEL,
PROCESSING AND MEMORY
FUNCTIONS
DATA
PROCESS MODEL
JUDGEMENTAL PROCESSES
Certain processes are especially useful,
depending on the issues
involved.
Some important ones are itemised below:
i)
(
Voting
Uses
include
selecting
an
item
or
option
from
a
set,
multiple
choice.
(
ii)
Scoring Options
Uses
include
evaluating each of
pre-defined scale,
a
set
scoring as a means
of
of
items
options
or
producing
lists
in
on
a
rank
order, scoring as a means of filtering long lists to short-lists.
(iii)
Comparing Options
Uses include weighting the relative importance of a set of criteria
evaluating subjective factors and
by means of pairwise comparisons,
parameters,
assessing
subjective
probabilities
and
probability
distributions, and assessing attitudes to risk.
(
iv)
Direct Assessment
Uses include the estimation of parameters,
probabilities and other
factors by direct and subjectively based inputs.
JUDGEMENTAL MODELLING STAGES
The use of an on-line system supporting the group means that these processes
can be undertaken much more effectively,
the group.
facilitating immediate feedback to
Judgemental modelling involves the following ordered stages:-
(i)
Establish the judgemental issue(s)
(ii)
Present carefully designed questions to the group
(iii)
Obtain responses from the group using individual handsets
(iv)
Display the range of responses back to the group
(v)
disagreement
Stimulate debate of principle points of agreement and
(vi)
Repeat
of
any
there
until
steps
above
the
some
is
form
of
and/or the differences within the group are understood
agreement,
and accepted
JUDGEMENTAL SCALES
requires careful attention to the
Each application of judgemental modelling
It is important that a scale is matched
scale(s) to be used with the group.
are clearly
and that the defined point(s) on the scale
to
task,
the
It
on-line.
use
relating
is
important
See,
for
Belton 191
example
that
logically
are
scales
measurement
to
the
for
understood
be
every
by
interval and ratio scales all
ensure
to
mathematical operations in order to avoid
errors.
must
but
group,
the
of
We have found nominal, ordinal,
participant.
of
help
the
developed with
may sometimes be
The scale and its anchor point(s)
established and agreed.
a
questioning
procedures
consistent
with
any
introduction of systematic
issue
m
number
of
this
of
discussion
connection with pairwise comparisons.
FIELD APPLICATIONS
4.
section
This
is
intended
present
to
in
of
each
brief
Section
applications of the support technology described in
Each
application
model,
and
is
described
experience.
The
under
the
applications
support
generalised process framework presented in Section
The applications
concern groups
in
of
headings
organisations
a
3.
background,
the
process
development
of
a
5.
engaged
in
strategic
as
Another
as "once only" type.
well as routine tasks of a "recurring" as well
suitable process for each
aspect to note is that having designed a
or modified software,
organisation, incorporated into specially developed
left to operate the support
the groups were sometimes largely or totally
any influences and biases
system technology for themselves, so minimising
that a facilitator presence can have (391.
10
Portfolio selection in commercial R and D
A.
Background
The
organisation
is
a
pharmaceutical
multinational
with
group,
and
R
D
laboratories collaborating with each other but based at sites in Europe and
North America.
preferred
The
portfolio
decision area
of
projects
concerned
is
which
to
with
the
allocate
to
selection
scarce
of
a
resources,
especially manpower.
A
number of factors are taken into account in this process, including market
potential, competition, stage in development,
chances of technical success,
resource constraints
therapeutic area balance, long and short term returns,
and key resource availabilities.
A
series of meetings
held each year to
is
decide the portfolio, and other review meetings are called at other times to
decide changes.
Process Model
The
was
GDSS
used
by
committees
within
therapeutic
each
area,
and
the
outcomes of these meetings were used as inputs to subsequent meetings of the
development committee.
under headings
of
Rank ordered lists
"early
stage"
were produced within
"near market
and
considered in terms of various commercial and
other factors and information.
projects".
each
area
These
were
technical criteria as well as
The development committee produced
overall
rank-ordered lists for the two project categories, and from these an overall
prioritized list.
Resource constraints and issues of portfolio balance were
considered in order to decide the preferred overall portfolio.
was
designed
facilitate
a
voting on issues,
criteria,
software
to
enabled
displays
variety
of
processes,
differences
of
scoring on
especially
and pairwise comparisons.
of
The software
opinion
At
every
to
be
stage
debated
the
and
changes to be made so that the group was able to progress flexibly with many
interactions.
Experience
The whole process was spread
linked meetings.
at certain points
inputting.
over several
weeks
and
involved
a
number of
Most of the time there was a high level of agreement,
there was great disagreement,
but
and much debate and repeat
During the meetings there was a small number of occasions when
11
fundamental and important disagreements persisted and there was not even a
The leader then focussed debate on isolating
simple majority position.
whether the differences related to "factual"
to fundamental values.
and
in only
the
chair.
there
resolve
this
at
the
most cases a consensus emerged at these points,
In
stage
difference
a
of
carried
person
this
that
consensus
this
that
that
given
opinion
responsibility,
ultimate
the
taken by
finally
one instance a finely balanced judgement was
Given
was
information or assumptions or
preferred
was
the
the
issues
way
been
had
to
well
explored and understood in debate.
possible to take a
The Director of Project Management stated that "it is
the meeting on
vote at the beginning of a discussion to gauge the feeling of
an
That
issue.
whether
pinpoints
need
you
to
discuss
it
more.
It
clear-cut
facilitates the process of debate and rapidly ensures you get to a
for facilitating appropriate
tool
very useful
a
is
it
decision ...
need to
discussion and ensuring that you spend time dealing with things that
He
conclusions" 161.
be discussed rather than on things that are foregone
decreased and
also noted that the time taken in meetings was substantially
These
taken".
yet stated: "I have greater confidence in the decisions
comments
generally
some
support
of
the
conclusions
of
12,34],
that
the
from R &
stimulation of communication in R & D is a pre-requisite to output
essential for an efficiently
D, and also that face-to-face communication is
operating information network.
B
Research Proposal Selection in a Public Body
Background
Committee-based decisions
The organisation is a British Research Council.
submitted by
concerning the selection for funding of research proposals
peer review.
researchers in Universities and Polytechnics are considered by
in each of a
Committees of 15 to 25 members consider long lists of proposals
number of scientific
precise
apportionment
areas.
of
The
this
overall
figure
budget
between
is
the
pre-defined,
scientific
but
areas
the
is
deliberately left open when the committees meet.
Process Model
The complete list of proposals was held in the software,
each be called up for display in list or other order.
1
p
so
that they could
Each
proposal
was
presented
committee
defined
by
referees'
members,
comments
short discussion of views and points of clarification held.
the chairperson summarised the position and
of
a
type
fundable
(as
opposed
When fundable,
resubmission).
to
and
a
At this point
stated whether this project was
reserve
a
noted,
list,
rejection
a
the committee members each scored
or
a
the given
proposal on each of the two criteria: "scientific excellence" and "relevance
to policy".
The scale used was based on
familiar
the
to
(low)
1
membership having been
use
in
to 5
over
(high),
and was very
several
years.
The
extremes were defined in a "global" rather than a "local" sense.
Mean score and standard error of the mean was displayed after each
These were infrequently debated further at this stage.
inputs.
proposals
fundable
treated
been
had
projects
excellence"
was
this
displayed,
way,
an
based
on
displaying
while
only,
in
the
overall
mean
rank
scores
"relevance
to
set
of
After all
ordered
list
of
"scientific
for
policy"
mean
and
standard error as well.
running
The
total
of
aggregate
first
year
costs
was
also
displayed.
Definition of an approximate range from minimum to maximum budget allocation
to the given committee allowed identification of the "grey" projects falling
within this range.
the
then carefully reviewed again by
Further discussion took place and recommendations were made
the committee.
to
The grey projects were
officers
concerning
budget constraints.
the
preferred portfolio under
Exceptionally,
some
various
overall
proposals were discussed
further
and then re-scored at this stage.
Experience
All committees of the Research Council have been using the system for three
years on a twice yearly basis.
After initial reservations by some members
(111,) both the Council Officers and the peer review groups appear to find
the system a useful
aid.
A
senior officer concerned
stated:
"The
Council
gets more information than before, with significantly more precision".
It may be argued on statistical grounds that the scale with five points only
is too coarse,
this
is
the
and that a scale with more divisions is preferable.
fact
that
the
scale
has
been
used
for
several
Against
years
in
a
"global" sense by committees in which the membership is fairly stable from
year to year.
A change would mean the
understood process.
re-establishment of a presently well
Further details of this application have been described
13
.
by Cherfas (1990)
Perceptions in Retail Marketing
C.
Background
The
organisation
audio,
video
is
and
medium-sized
a
equipment
TV
image.
In particular,
certain
up-market
with
a
significant
wide
regional
thought
was
It
some
by
range
presence
achieved a pre-eminent position
it had
brands.
chain selling a
retail
managers
store
and
terms
in
of
of
that
prospective customers viewed the company as expensive and up-market because
the expensive brands were stocked and
The
management
priced,
were
group
associated strongly with
disagreement
in
as
how
to
other
their
brands were perceived by their customers,
mainly Japanese,
chain.
the
lower
and how
closely the chain's image related to these other brands.
Process Model
A number of exercises were run through with a group
of seven store managers
at a one-day meeting in order to measure perceptions and debate differences.
The work involved is briefly described in what follows,
The managers considered the set of brand names
have been described in tlSl.
that they stocked in
each
and further details
of
a
number of product areas.
Each
brand
considered on each of a set of attributes using "local" scoring scales.
attributes
themselves
pairwise comparisions.
and also from the
were
assessed
terms
in
of
relative
importance
(perceived)
point of view of their
displayed
and
by
they
perceived
Holistic
customers.
their
stores
with competing store chains on each of a number of attributes.
discussion
The
The managers responded from their own point of view
analyses were also undertaken of how
stage,
was
debate
was
agreements/disagreements
related
of
to
feedback
the
opinion,
rather
compared
At
every
screens
which
than
results
of
mathematical analysis.
Experience
The meeting continued for several hours and most of this time was spent in
discussions of the reasons for differences in views represented on feedback
screen
displays.
differences
was
At
times
generated
in
considerable
this
way,
even
surprise
though
regarding
this
was
a
revealed
group
of
managers who had worked together for a considerable time and knew each other
14
well.
the conclusion
At
of
participants had decided on
meeting,
the
brands which offered the most scope for their chain,
the
and were formulating a
marketing strategy including some well defined attributes for subsequent use
in market research exercises.
Option Selection in Defence
D.
Background
The
organisation
A set
Defence.
to meet a well
is
research
a
establishment within
of seven options had been
defined threat.
the
Ministry
U.K.
identified for a military system
considerable amount of experimental
A
simulation based data was available.
of
criteria
The
which
by
judge
to
and
the
options were pre-defined.
Process Model
As a means of exploring further the relative merits
of the options,
of twelve military experts was assembled for a one-day meeting.
were considered on each of the criteria in turn.
to
define and anchor the end points
(and at
of the given
Following
criteria.
The options
with
the
this,
group
entered
they were
each set
As
immediately displayed back to the group
During these debates, re-definition of some of the criteria
for discussion.
Sometimes
and/or the number of criteria changed.
took place,
personal
their
ratings on each option for each of the criteria taken in turn.
of inputs was completed,
personal
their
highest and lowest rated options on each
confidential)
this stage,
group
Each participant was asked
rating scales
of
a
ratings
the
were repeated as a result of this or other debates.
The
relative
importance
of
using a constant sum method.
allocated
points
discussion
between
(see [131
allocate
these
relative
importance
completed,
tables
the
across
for
points
(e.g.
set of
discussion and a re-run.
The
criteria were
pairs
each
by
"constant sum"
according
importance
the
the
themselves
criteria
the
weights
were
criteria and
Finally,
was displayed by calculating a
weights"
an
pairwise
displayed
the
as
15
100
prior
asked
to
definition
of
comparisons were
rankings
and
This
generated
"overall" result for each individual
summation of the scores
for each option over the criteria set.
were
values
participants.
and
without
participant
"swing
all
pairs
in
They
When
see [531).
taken
considered
then
method).
the
to
were
times
the
weights
Differences displayed at this stage
An exploration of
generated further debate.
sensitivity
the
change
to
of
first the scores and then
these results was presented on-line by changing
by the group overall
In a final debate the option preferred^
the weights.
was selected by consensus.
Experience
with the process, especially the frequent
The participants seemed very happy
The meeting was considered
opportunities to debate well focused issues.
meetings concerned with other
effective by all participants, and further
As seen by the establishment facilitator
issues are now frequently held.
our sponsors to examine potentially
"the use of TEAMWORKER enao.ci us and
producing
and transparent manner,
contentious issues in a methodical
and "policy" factors for use in
acceptable quantifications of "quality"
decision making".
E.
Performance
Analysis of Key Competencies and T ask
Background
brewing
This application is located in the
statement
had formulated a clear vision
established a set of thirteen associated key
expected
were the set of task abilities he
managers in order to achieve the vision.
assess
the
strengths
and
weaknesses
of
industry.
for
the
The
Chief Executive
business,
and
also
task performance factors. These
in his
It
was
the
key
key team
of
considered
team
of
operational
essential
managers
to
both
individually and as a whole.
Process Model
A meeting was held of
the group composed of board directors.
Each of
the
task factors using a scoring scale.
managers was assessed on each of the key
differences of opinion and this
At each stage, summary screens displayed
to
led
also
It
re-scoring.
resulted in mini-debates and occasional
definitions of the task factors
improvements in understanding of the
the competencies was also
The relative importance of each of
themselves.
Summary screens enabled the
evaluated by means of pairwise comparisons.
level
at the individual and group
board to evaluate gaps in key task areas
Discussion then led to conclusions
and to assess their importance.
This involved a consideration
concerning how certain gaps should be closed.
16
centred
options
of
recruitment.
around
development
internal
staff
of
opposed
as
to
competency data was matched to the task data in order
Later,
to predict future task performance.
Experience
The meeting resulted in management action with the help of a team-building
consultant working within
carried
ratings
multiple
the
organisation.
out
in
this
consultant
The
were
way
sometimes
stated:
"the
odds
with
at
performance appraisals and pinpointed weak areas quickly so that they could
be given attention".
Long Range Planning
F.
Background
The organisation is a large multi-national manufacturing company in a high
Concern was expressed to ensure that the company was
technology business.
making
good
decisions
technologies
regarding
to
be
in
commercial
for
introduction and exploitation on 10-30 year future time frames.
Process Model
A
long
list
session with
"business
of
group
a
areas of the company.
the
relative
social,
twelve
of
developed
executives
from
business
future
of
political,
during
scientific
These are the factors likely
importance
technological,
was
drivers"
to
options.
cultural,
economic,
brainstorming
a
commercial
and
strongly
influence
involve
can
They
environmental
and
military scenarios.
A
larger group of 45 executives representing all areas of
well as scientific and commercial expertise,
was assembled.
the
company,
as
The long-list
of drivers was then scored on a high-low interval scale in order to discuss
differences,
average inputs,
and filter to
a
much shortened high priority
list.
A
long
list
of
"business options"
important business drivers.
by
scoring.
high-low
The
interval
then generated
in
relation
to
the
These were similarly filtered to a shorter list
short-list
scales
was
was
then
established
scored
for
17
on
two
criteria
each participant.
The
on
local
criteria
were defined in terms of "likelihood" and "zone of influence".
viewed, discussed, and sometimes re-scored.
were displayed on a two dimensional grid,
Eventually, the averaged scores
along with ellipses based on the
Differences between scientific and commercial
standard errors.
were explored for significance and discussed.
business options was reached
top priority
Results were
sub-groups
Eventually agreement on the
along with
the
implications
for
appropriate technologies of greatest concern to the company in the future.
Experience
The final
three
of
days
runs
were undertaken without
facilitation
in
the
conference room in order that no "external influences" were present to bias
On completion,
or distort results.
the process in
terms
of
the participants were very
opportunities
to
both
input
pleased with
judgements
their
handsets and to express their feelings during focused debates.
via
The managing
team were provided with clear steers in terms of the areas requiring further
Following
attention.
the
meetings,
the
organising
couldn't have had that event without your system.
other,
conventional,
more
meeting
but
I
don't
officer
We
think
stated:
might have
"We
had
some
have
been
it
would
a
university-based
nearly as productive".
G.
University Based Education and Training
Background
The
situation
Department
involved
Pharmacy.
of
teaching
undergraduate
The
was
need
in
ensure
to
professional skills when dealing with patients,
with
the
patient
in
order
to
of
and in accurate prescribing
of appropriate drugs or medication for a given condition.
relationship
development
the
This involved the
a
proper
screen.
This
make
diagnosis
and
prescription decision.
Process Model
Video
training
film
was
shown
commercially available material,
on
large
a
and
sometimes
vignettes
was
sometimes
produced
by
the
The video film was paused at appropriate points
in a
series of sequences of dialogue between the patient and the pharmacist.
The
students themselves.
same
large
screen
alternative answers.
then
displayed
an
appropriate
This multiple choice
18
question
and
set
of
format was sometimes varied into
"yes-no" and other binary options.
The students responded with handsets,
up
Histograms of
four.
to
either individually,
received were
responses
the
or
in groups of
displayed on
the
screen so that the lecturer concerned could easily run a discussion based on
the range of answers.
The next video clip was then presented, and so on.
Experience
Lecturers working with the system have noted a greatly
increased
level
of
attention, interest and preparedness to participate in classroom debate from
When the students formed small teams to a single
all the students involved.
handset,
meant
spirit
a
"within
that
competition between
of
debate
team"
enhancing mutual learning.
The
the
well
as
as
teams
was
generated.
debate
plenary
system is being regularly
department concerned which we believe is most significant.
evaluate
whether significantly
enhanced
learning
and
This
took
utilised
place,
the
by
It is planned to
retention
taking
is
place by comparison with other student groups not using this approach.
Panel Testing and Tasting
H.
Background
The organisation is a major brewer with a large range of beers and lagers at
a
Production is tasted and tested by panels of
number of dispersed sites.
The objective is
experts on a daily basis.
to
identify,
for
each
branded
product, deviations in either direction from a defined product profile.
The
profile is based on 35 dimensions related to identifiable aspects of taste,
appearance and smell.
Process Model
It was
important
process
in
to
his/her
allow
own
each
time,
participant
and
by
to
respond
exception.
A
during
scoring
the
tasting
with
scale
allowance for positive or negative deviations from a correct (zero) profile
for
each
beer
was
already
in
use
very
and
This
familiar.
was
scale
incorporated into the software.
The screen was sectored for responses from
particular
On
typically
histogram
numbered
handsets.
eight beers,
form
for
the
each
completion
values were
beer
on
displayed
each
19
of
the
back
dimension.
inputs
to
the
Where
on
each
panel
in
of
a
significant
differences of opinion were observed they
were debated.
The
participants
sometimes re-tasted and re-scored at these points.
Experience
The
process
was
delivered
and
fast
timely
The system is now being adopted by
tastings.
outputs
for
debate
during
the
organisation with a view
to maintaining standards within and between sites.
The system is also being
used to facilitate the training of new panel members,
to set standards,
and
also to identify trends in production output quickly.
A
5.
PROCESS FRAMEWORK
In Section 2 we have overviewed the literature connected with social choice,
multiple criteria decision making, and social processes in groups engaged on
related
decision
summarised
Section
in
From
tasks.
we
4,
work,
this
postulate
our
and
own
following
the
field
experience
general
guidelines
within which to focus the design of GDSS in varied settings:
1.
Each participant should be fully involved in those stages of the process
which
are
not
pre-ordained
by
context.
the
particular,
In
setting
objectives, generating alternatives,
establishing and defining criteria
with which
These
judge
to
alternatives.
should be
openly
discussed,
changed and filtered as the meeting progresses in a dynamic way.
2.
As a means of identifying useful points of debate,
discussions,
group
display
alternatives
participants should have
all
judgements
their
against
each
importance of the criteria,
stages.
At each stage,
of
the
one
opportunities
terms
in
and aiding meaningful
evaluations
of
criteria,
with another,
input
to
and
at
of
the
relative
series of defined
a
the complete set of inputs
the
of
for
should be
received
prior to display to minimise social biasing effects.
3.
Feedback
of evaluations
at
each
stage
should
be
simple
displaying the set of values received from the group
way,
with
as
little
processing as possible,
in
and
a
graphic,
transparent
enabling a discussion
of
points of agreement and of disagreement (especially of polarised views).
4.
At any
stage,
indicate
the
display of differences of opinion followed by debate may
need
to
revisit
that
20
stage
or
any
earlier stage
of
the
process, and this should be taken without hesitation.
5.
At some stages,
participants
it may
be useful
some
mathematical
in
On each occasion,
averaging.
to aggregate results
process of,
for
from each of the
adding
example,
or
this process should be made very clear to
the group as well as any inherent shortcomings it may contain.
6.
As
a
result of the
group should reach
inputs,
feedbacks,
discussions
and
analyses,
the
collective choice, which is an agreed and preferred
a
result, achieved by means of debate.
7.
The technology of decision support should seek to aid and enhance group
processes
inhibit
not
and
dominate
or
which
them,
requires
careful
attention to design aspects concerned with both hardware and software.
8.
Procedures and evaluation methods should be constructed to build on from
the cultural
settings and prior
experience of
joint learning and development experience.
the
group as part of a
These aspects may well
need
attention in advance.
9.
Questions requiring input of judgement from
participants should always
be formulated in clear and unambiguous terms.
There is never a case for
introducing fuzziness over and above that which is inherently
present.
definitions for scoring or comparing
This is particularly true of scale
items.
CONCLUSIONS
6.
Many
important decisions in organisations are made in group contexts,
though
one
actually
member
taking
extensions
of
the
the
group
A
is
decision.
new
The dynamics
sometimes
entity
is
involve
ultimately
decision
Group
appropriated
processes
the
decision-making.
dynamics.
of
to
created with
complex
social
software,
and
processes
social
incorporating
decision-making,
dialogue.
these
other
managerial
in
means
processes
group
A
of
21
when
not
merely
engaged
decision
in
related
decision
appropriately
from
multiple
developed
which this can be done,
particular context.
are
for
processes of communication
features
selected
There are many ways
by
own
its
and position modification between participants.
system can aid
responsible
processes
individuals
even
to
support
designed
criteria
aid
group
depending on the
This
choice
needs
to
take
account
aspects
of
of
considered to be deleterious in their effects.
are
the
management
tools
of
Nominal
various forms of Brain-Storming,
as
social
influence
often
Of value in this connection
Group
Technique,
Group
Delphi
and
these seek to encourage dialogue after
sets of inputs from group members, so reducing conformity, risky shifts, and
other related phenomena.
The choice also needs to take account of the degree of abstraction implicit
in
the use of a particular
decision analytic
framework,
for aiding individual rather than group decision-making.
"provide a common language
for
disagreements"
(
primary
quantification
as
opinions
can
mathematical
51J
.
a
The
tool
for
explored
be
operations
in
of
communications
use
of
structuring
debate,
developed
The framework can
and pinpoint agreements
the
framework
thought
rather
aggregation,
usually
so
than
which
do
to
is
that
merely
have
the
to
and
provide
range
of
perform
fundamental
shortcomings.
Finally,
and
perhaps paradoxically,
the
technology
group process by enhancing individual processing.
[33]
of
GDSS
may
"The individual .... becomes an object to himself only by
taking the attitude of other individuals towards himself
a
social environment or context of experience and
behaviour in which both he and they are involved."
22
a
Or as George Herbert Mead
elegantly stated:
within
assist
.
REFERENCES
1.
Adair, John (1986)
2.
Allen,
'Effective Teambuilding'
"Managing
(1977)
T.J.
.
Flow
the
Gower Press
Technology"
of
MIT
Press,
Cambridge, Massachusetts
3.
Arrow, K.J.
4.
Asch, S.E.
of
'Studies of independence and conformity.
(1956)
against
one
a
unanimous
Psychological
majority'
A minority
1.
Monographs
,
1-70
70(a),
5.
(1951) Social choice and individual values Wiley, New York
Bales, R.F.
Interaction process analysis - a method for the study
(1950)
of small groups Cambridge, Mass: Addison-Wesley
6.
Bartrum,
Peter
Strategy
7.
Helton, V.
1
,
Talking
the
Had
Executive
Stop'.
to
p 20
,
(1986)
decision
'When
(1991)
'A
comparative study of methods for multiple criteria
Unpublished
aiding'.
Ph.D
Emmanuel
thesis,
College,
Cambridge, England
8.
Belton, V.
and Gear,
Proceedings
comparisons'
'Assessing weights by means of pairwise
(1986)
T.
of
Multiple Criteria Decision Making
9.
Belton, V. and Gear, A.E.
(1983)
Analytic Hierarchies'. Omega
Brown, Robert (1988) Group processes
11.
Cherfas,
Jeremy
Science
12.
,
October 1990,
Churchman, C.W.
(1966).
p.
Software
Hard
for
Choices'.
"On the Interccmparison of Utilities".
Essays
Methodology ".
in
In "The
Edited
by
pp 243-256. Prentice-Hall
S.R. Krupp.
Cleland, David
Oxford: Blackwell
367
Structure of Economic Science:
13.
228-230
Review:
'Peer
(1990)
.
on
Shortcoming of Saaty's Method of
11(3),
10.
Conference
Kyoto. Springer-Verlag
,
'On a
,
International
Vlth
the
I.
and Kocaoglu, Dundar F.
(1981) Engineering Management
,
McGraw-Hill Inc.
14.
Dalkey,
An
(1975)
N.
functions
.
Davies, Barry (1989)
16.
Delberq,
C,
probability
group
for
'The decision in your hands'
Van de Van,
Jones,
S.
&
and Gustofson,
A.
for programme planning
17. Eden,
theorem
Rand Corporation, P. 4862
15.
A.,
impossibility
.
Glenview,
Sims,
A.
D.
OR Insight
(1975)
,
Group
2(4),
3-5
techniques
111.: Scott Foresman
(1983)
"
Messing
about
in
Problems ",
Pergamon Press
18. Gear,
Andrew
E
(1992)
decision-making
Behaviourist"
in
"Social
groups
Representations
from
the
in
standpoint
In preparation for publication
relation
of
a
to
Social
19. Gear,
Read,
and
T.
Proceedings
of
M.J.
Decision
Criteria
Making
Conference
Manchester,
,
decision
group
International
Vlllth
the
'On-line
(1988)
England.
support'
Multiple
on
Published
by
Springe r-Verlag
20. Gear,
Tony, Marsh, Nicholas R. and Sargent,
Peter (1985)
feedback and team behaviour' Human Relations
21. Gear,
'Semi-automated
38(8) ,707-721
,
Tony E., Lockett, Alan Geoffrey and Muhlemann, Alan Paul (1982) "A
Unified Approach to
Acquisition
the
IEEE Trans, on Eng. Management
22. Hammond,
K.R.
McLelland,
,
judgement
Gary
decision
and
in
R
D"
&
Vol EM-29, pp 11-19
,
H.
Subjective Data
of
,
making:
Murnpower
and
theories,
methods
Jeryl
Human
(1980)
procedures
and
Praegar Press
23.
Joel, Rohrbaugh, J.,
Harmon,
(1990)
'Social judgement analysis and small
group decision making: cognitive feedback effects on individual and
Organisational Behaviour and Human Decision
collective performance'
Processes
24. Himmelweit,
Hilde
Scope'
T.
25.
34-54
46,
,
p.
30.
'Societal
(1990)
T.
Implications
Psychology:
In Societal Psychology
and
Edited by Himmelweit, Hilde
,
and Gaskell, George. Sage Publications
Islei, Gerd.
Lockett, Geoff., Cox, Barry and Stratford, Mike (1991)
,
"A
Decision Support System Using Judgemental Modelling: A Case Study of
R
&
the
in
D
Pharmaceutical
Industry".
IEEE
Trans.
on
Eng.
Management Vol 38(3), 202-209
26.
Jelassi, M.T., Baueclair, R.A.
decision support system design'
for
Research
on
'An integrated framework for group
(1987)
IRMIS Working Paper W703
Institute
School
of
analysis
in
and Morse,
P.
Information Systems,
Management of
the
,
Business, Indiana University
27. Keeney,
and Raiffa,
R.L.,
(1972)
H.
public decison-making'
In Drake,
(eds) Analysis of public systems
Ralph
28. Keeney,
and
N.W.,
Keeney,
(1976)
Lewandowski,
Andrzej
decision making:
Systems
30.
Lewin, K.
,
(1989)
'
multiple
with
New York: John Wiley
SLDAS - Decision support system for group
decision
theoretic
Decision
framework'.
Support
5(5) 403-423
(1948) Resolving social conflicts
31. McClelland,
R.L.
Decisions
objectives, preferences and value trade-offs
29.
formal
of
MIT Press
.
Howard
Raiffa,
critique
'A
W.
Grigor
(1976)
"And
a
New
.
New York: Harper & Row
Earth".
Lecture.
Swathmore
Published by Friends Home Service Committee, London
32.
Mann,
Carl
P.
(1986)
Decision-Making".
"How
to
Remove
Groupthink
Public Relations Quarterly
.
from
Executive
31(3), 28-31
33.
Mead, George Herbert (1934)
W.
34.
in Mind self and
society
edited by Charles
,
Morris, University of Chicago Press
Meyer, Arnoud De (1991)
R&D
35. Minkes,
How Managers are Stimulating Global
"Tech Talk:
Communication" Sloan Management Review
A.L.,
'Notes
(1983)
Lindblom Revisited'.
32(3), 49-58
.
Decision-Making
on
Hong Kong Journal
Organisations:
in
of Business
Management
1,
27-32
35.
Serge
Moscovice,
European
Representations".
Towards
"Notes
(1988)
Description
a
Journal
Psychology
Social
of
Social
of
18,
.
211-250
37.
Herve
Moulin,
(1988)
Axioms
of
cooperative
decision
making
Cambridge
University Press
38.
Nunamaker,
Vogel
Jay.,
"Experiences
,
IBM
at
Decision Support Systems
39.
Phillips,
L.D.
(1988)
Daukidis,
G.
5,
.
management support systems
40. Rapoport,
Anatol
.
group
Read, Martin and Gear, Tony
International
Meeting
support'
in
based
Chichester: Ellis Horwood
Comparison
,
of
Utilities".
Kyoto, Japan
'Interactive group decision support'
(1989)
on
(1989)
Study".
Knowledge
(eds)
Presented at XXII International Meeting of TIMS
41.
Field
A
decision
G.
"Interpersonal
(1975).
Ben
Martz,
Systems:
Miller,
and
F.,
and
183-196
'People-centred
Land,
,
Support
Group
with
Alan
Heminger,
Doug.,
Multiple
Criteria
Making
Decision
,
organised by IIASA, Helsinki. Springer-Verlag
42. Rickards,
T.
(1974)
"Problem-Solving Through
Creative Analysis"
Gower
Press, UK
43. Rohrbaugh,
J.
(1981)
'Improving the quality of group judgement:
judgement analysis and the nominal
Behaviour and Human Performance
44. Russo,
Edward &
J.
Making",
45. Schwenk,
p.
161.
Schoemaker,
,
Paul
28,
group
272-288
J.H.,
(1989)
"Confident
Howard
(1983)
'Formulating
role of decision aids in problem formulation' Omega
46. Souder, William E. and Ziegler Robert W.
(1977)
'A
Decision
,
the
mess:
the
11(3), 239-252
review of creativity
and problem solving techniques' Research Management
,
20(4), 34-42
David (1981) "Automatic mediation in group problem solving".
Behavior Research Methods and Instrumentation
48. Stoner
Organisational
technique'
Published by Judy Piatkus Ltd., London
Charles and Thomas,
47. Stodolsky,
social
J.A.F.
(2961)
Involving Risk".
"A
Comparison
of
,
Individual
Unpublished Master's Thesis,
Management, Massachussetts Inst, of Technology.
13(2), 235-242
and
Group
Decisions
School of Industrial
.
49. Swap, Walter
C.
et
(eds)
al
Group
(1984)
decision
making
Sage
Publications
50.
Turner, John C.
(1991)
"
Social Influence ", Open University Press, Milton
Keynes, UK.
51.
Von Winterfeldt,
and
Edwards,
Ward
(1986)
Decision
analysis
and
behavioural research Cambridge University Press
52.
Warfield,
J.N.
Complexity
(1990)
through
Science
'A
System
Generic
of
Design'.
Design
Intersystem
-
Managing
Publication,
Salinas, Ca.
53.
Watson, Stephen, R. and Buede, Dennis M.
(1987) Decision synthesis.
principles and practice of decision analysis
.
The
Cambridge University
Press
54.
W.C,
Wedley,
Jung, R.H. and Marchant,
G.S.
(1978)
Delphi way' Journal of General Management
55.
Yu Po
L
(1988)
"Effective
Decision
,
Making
5,
'Problem solving the
23-26
Using
Habitual
Domain
Analysis". ORSA/TIMS Joint National Meeting, Denver, Oct 23-26.
Address for correspondence: Dr
T.
Gear, Balliol College, OXFORD, 0X1 3BJ
MAR
20Q1
Date Due
MIT LIBRARIES
DUPL
3 9080 02238 6079
Download