+ All Categories
Home > Documents > Chapter 11 Groups of Decision Makers

Chapter 11 Groups of Decision Makers

Date post: 02-Jun-2018
Category:
Upload: thaot
View: 221 times
Download: 0 times
Share this document with a friend
29
8/10/2019 Chapter 11 Groups of Decision Makers http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 1/29 11 Groups of decision makers Why do we get so hung up on what we don’t agree on, when in fact it’s our differences that make life interesting? (Bradley Trevor Grieve) 11.1 Introduction While we have concentrated on decision making by a single DM, most decisions, at least in organisations and society, are the responsibility of groups rather than individuals. How do the ideas that we have been developing generalise to the context of groups of DMs? Why are groups often preferred to individuals? Do groups always make better decisions? These are important questions, and they are addressed in this chapter. There is a generally held belief that groups make better decisions than single individuals working alone. This belief has led many to advocate that important decisions should be deferred to groups – e.g. decisions about innocence or guilt made by juries, decisions about national policy made by Cabinets and decisions about corporate strategy made by boards of dir- ectors. Among the many reasons why groups might outperform single individuals, four of the more important are that they:  have a greater amount of knowledge, intelligence and understanding of the problem than any single individual;  increase members’ motivation to work hard and remain task-centred, since their performance is open to the scrutiny of other members of the group;  facilitate creativity, in that people can often ‘spark’ off each other and develop new ideas better in group contexts; and  increase the watchfulness to error and the resolution of ambiguity, given that ideas and assumptions are stated publicly for other members of the group to evaluate, review and challenge when appropriate. While each of these can lead groups to making better decisions than individuals, there is evidence that these potential advantages are not always realised. For example, Janis (1972) identified a series of high- profile examples in which groups made spectacularly bad decisions – e.g. 320
Transcript
Page 1: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 1/29

11

Groups of decision makers

Why do we get so hung up on what we don’t agree on, when in fact it’s our differences that

make life interesting? (Bradley Trevor Grieve)

11.1 Introduction

While we have concentrated on decision making by a single DM, most

decisions, at least in organisations and society, are the responsibility of 

groups rather than individuals. How do the ideas that we have been

developing generalise to the context of groups of DMs? Why are groups

often preferred to individuals? Do groups always make better decisions?

These are important questions, and they are addressed in this chapter.

There is a generally held belief that groups make better decisions than

single individuals working alone. This belief has led many to advocate that

important decisions should be deferred to groups – e.g. decisions about

innocence or guilt made by juries, decisions about national policy made by 

Cabinets and decisions about corporate strategy made by boards of dir-

ectors. Among the many reasons why groups might outperform single

individuals, four of the more important are that they:

  have a greater amount of knowledge, intelligence and understanding

of the problem than any single individual;

  increase members’ motivation to work hard and remain task-centred,

since their performance is open to the scrutiny of other members of the group;

  facilitate creativity, in that people can often ‘spark’ off each other and

develop new ideas better in group contexts; and

  increase the watchfulness to error and the resolution of ambiguity,

given that ideas and assumptions are stated publicly for other members

of the group to evaluate, review and challenge when appropriate.

While each of these can lead groups to making better decisions than

individuals, there is evidence that these potential advantages are not

always realised. For example, Janis (1972) identified a series of high-

profile examples in which groups made spectacularly bad decisions – e.g.

320

Page 2: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 2/29

t he d ec is io n b y P re si de nt J ohn so n a nd hi s c lo se a dv is er s t o e sc al at e

t he V ie tn am Wa r. T hi s s ug ge st s th at g ro ups a re a s m uc h i n n ee d of                  

d e c i s i o n s u p p o r t a s i n d i v i d u a l s . M a n y o f t h e t h e o r i e s w e h a v e d i s c u s s e d

s o f ar , h ow ev er , a re f oc us ed o n i nd iv id ua l D Ms . T he S EU m od el , i n

p a r t i c u l a r , i s b u i l t u p o n a c o n c e p t i o n o f t h e b e l i e f s a n d p r e f e r e n c e s o f a

‘ ra ti on al ’ i nd iv id ua l. C an i t b e g en er al is ed t o g ro up d ec is io n- ma k in g

c o nt e xt s ? U n fo r tu n at e ly , n o : t h er e h a ve b e en m a ny e x pl o ra t io n s o f t h e

m a th e ma t ic s o f g r ou p c o ns e ns u s p r ob a bi l it y a n d u t il i ty f u nc t io n s, a n d

a ll h av e d is co ve re d t ha t a ny f or ma li sm f or b ui l di ng a s in gl e m od el o f                  

g ro up p re fe re nc e i s s ub je ct t o i r ra ti on al o r u nd em oc ra ti c b eh av i ou rs .

I mp os si bi l it y t he or e ms a bo un d, s ug ge st in g t ha t t he e nd ea vo ur i s i l l-

f o rm u la t ed m a th e ma t ic a ll y . A c l as s i c i m po s si b il i t y t h eo r em d e ri v ed b y    

A r r o w (s e c t i o n   3) s u g g e s t s n o t o n l y t h a t i t i s i m p o s s i b l e t o e x t e n d S EUt o g r o u p d e c i s i o n m a k i n g b u t t h a t t h e v e r y i d e a o f d e m o c r a c y m a y b e a n

i l l- de fin ed c on ce pt . I f t hi s i s s o, a nd i f m os t d ec is io ns h av e t o b e m ad e

b y g ro up s, w hy h av e w e b ee n s tu dy in g t he n or ma ti ve f or ma li sm s o f                  

the   p r e v i ou s c h a p te r s? H ow c an t he y he lp i n t he m aj or it y of r ea l c ir -

c u m s t a n c e s ?

T he t ri ck t o ‘ sq ua ri ng t hi s c ir cl e’ i s t o c ha ng e o ur p er sp ec ti ve a nd t o

r e co g ni s e t h at g r ou p d e ci s io n s a r e u n de r pi n ne d b y i n di v id u al s a n d s o ci a l

p r oc e ss e s t h at t r an s la t e t h e j u dg e me n ts a n d p r ef e re n ce s o f m e mb e rs i n toi m pl e me n te d c o ur s es o f a c ti o n. T h is p e rs p ec t iv e d o es n o t c o nt r ad i ct t h e

a ss ig nm en t o f a cc ou nt ab il i ty , a ut ho ri ty a nd r es po ns ib il i ty t o a g ro up .

G r ou p s c a n h a ve d u ti e s a n d p o we r s. U n de r st a nd i ng a n d f a ci l i ta t in g t h es e

i nd iv id ua l a nd s oc ia l p ro ce ss es p ro vi de s t he k ey t o i mp ro vi ng g ro up

d e c i s i o n m a ki n g , h o we v e r . I n s e c t i o n   4   w e d i s c u s s s o m e t e c h n i q u e s i n t h i s

r e sp e ct , t h en i n s e ct i on   5   w e l oo k a t f ac il i ta te d w or ks ho ps a nd d ec is io n

c o nf e re n ce s t h at b r in g m a ny o f t h es e t e ch n iq u es t o ge t he r i n to a c o mm o n

f ra me wo rk , o ft en s up po rt ed b y d ec is io n a na ly si s t ec hn iq ue s. W e c on -

c en tr at e o n m od er at el y s iz ed g ro up s – e .g ., a b oa rd o f d ir ec to rs , a m an -

agement team or a C abinet. In the   n e xt c ha pt er   we turn to l arg er

o r ga n is a ti o ns , a n d, i n c h ap t er   13, t o t h e l a r g e s t g r o u p o f a l l : s o c i e t y .

O u r o b je c ti v es i n t h is c h ap t er a r e:

  t o i de nt if y a nd e va lu at e t he i nd iv id ua l a nd s oc ia l p ro ce ss es t ha t

u n de r li e g r ou p d e ci s io n m a k in g ;

  t o u n de r st a nd A r ro w ’s i m po s si b il i ty t h eo r em a n d i t s i m pl i c at i on s f o r

g r ou p d e ci s i on m a k in g a n d d e mo c ra c y; a n d

  t o p r es e nt a n d e v a l u at e a p pr o ac h es d e si g ne d t o s u pp o rt ( s ma l l ) g r o u pd e c i s io n m a ki n g.

321 Groups of decision makers

Page 3: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 3/29

11.2 Review of research on decision making in groups

Insanity is the exception in individuals. In groups, parties, peoples, and times, it is the rule.

(Friedrich Wilhelm Nietzsche)

Reviews of research on group effectiveness show that in judgement tasks

in which there is a known solution – e.g. the number of people living

in participants’ home city – groups generally perform better than the

‘average’ individual but worse than the ‘best’ individual (Einhorn  et al .,

1977; Gigone and Hastie, 1997; Hastie, 1986; Hill, 1982; Kerr and Tindale,

2004). Given our reliance on groups for making important decisions, these

findings are rather disappointing. Nonetheless, there is also evidence sug-

gesting that group effectiveness depends in part on the strategy adopted –e.g. groups may adopt an integration strategy that involves combining their

individual judgements or an identification strategy that involves establishing

the person who is likely to have the greatest knowledge and then using his/

her estimate as the primary basis for deriving the group solution (Sniezek 

and Henry, 1989; Steiner, 1972).

Interestingly, the strategy adopted by the groups in the Sniezek and

Henry study depended upon the amount of agreement/disagreement

between individuals at the beginning of the task. When there was relatively 

high initial agreement, the process tended to be based on integration with

approximately equal weight given to each individual’s judgement, and the

overall accuracy was not much higher than the average of the judgements

given by individuals before the group discussion. When there was high

initial disagreement, averaging did not occur. Instead, identification was

more important and the quality of the decision was either much better

or much worse than average individual judgement. Thus, the strategy 

adopted had important implications for the quality of the decision, but

was not actually under the direct control of the group.Overall, these findings suggest that groups do not necessarily make

better decisions than individuals; they may not capitalise on their four

advantages listed in the introduction to this chapter. It seems that the

processes adopted by groups may be critical in determining their effect-

iveness. Some of the factors that are known to inhibit group decision

making are outlined below (for a fuller review, see Baron and Kerr, 2004).

  Group member characteristics . Group decisions are often strongly 

influenced by those individuals who have power, show higher degreesof confidence in the information they present and are more persuasive.

This occurs even though these individuals are not necessarily those

322 Decision behaviour, analysis and support

Page 4: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 4/29

who know the most. Allowing such individuals too much influence and

not listening sufficiently to other group members, who may well be

better informed, can reduce the effectiveness of group decisions and

reduce the motivation of those members who are unable to make their

contribution, thereby further reducing their contribution.

  Group size . Larger groups are more likely to have the knowledge and/

or intelligence necessary to make an effective decision. As group size

increases, however, it becomes increasingly difficult for each individual

to make his/her contribution, given the increased competition for

‘airtime’. Therefore, it is less likely that an individual will be able to

share his/her knowledge and intelligence with other members of the

group, and those who are ignored are also likely to disengage, further

reducing their contribution.   Conformity . Individuals often change what they say in group discussions

in order to be seen to be consistent with the majority. This occurs out of 

a fear of being ‘different’ from everyone else, and may deprive the group

of crucial information that challenges emerging views and assumptions

that are incorrect.

  Social loafing . Individuals are sometimes not sufficiently motivated to

make their full contribution to group activities. As indicated above, this

is particularly likely to occur when they are unable to make theircontribution and this goes unnoticed. They may also disengage if they 

see the marginal benefit of any of their contributions to be low because

it is divided over many contributors. In large groups, social loafing can

involve individuals ‘sitting out’ the meeting by making no contribution

at all. Social loafing is lower when each member’s contribution to the

group output can be identified and evaluated (Comer,  1995).

  Production blocking . Individual contributions are sometimes diminished

because of the need to wait for an opportunity to share knowledge and

experience with other group members. This occurs because group

members compete for limited group ‘airtime’ and so must retain

their contribution until they command group attention. Nevertheless,

participants who cannot verbalise ideas as soon as they occur may forget

them, may suppress them because they seem less relevant later once the

discussion has moved on or may be unable to generate further ideas

because they are rehearsing their ideas while waiting for their turn (see,

for example, Diehl and Stroebe,  1991).

  Evaluation apprehension . When individuals feel that they are being judged or evaluated by others in or outside the group they may self-

censor their contributions, even though these may be constructive and

323 Groups of decision makers

Page 5: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 5/29

relevant to group goals. In groups brought together especially for the

purpose of making one particular decision, the pressure on individuals

to make a positive impression on strangers is likely to increase

evaluation apprehension.

  Biased information pooling . Groups are more likely to discuss and

make use of information that is available to all members – i.e. ‘shared’

information – but often fail to exchange and discuss information that is

available uniquely to a single individual – i.e. ‘unshared’ information.

Thus the unique contribution that individuals can make is often not

realised (for a review, see Hinsz  et al ., 1997).

An emerging theme from the discussion so far is the importance of the

process adopted when groups make a decision. Accordingly, we now turn

to a brief review of what we know about group processes, beginning withthe manner in which they may reach consensus.

A major challenge for groups is to find ways of bringing together their

different views to agree a single course of action. Hastie   et al . (1983)

investigated how mock juries reached consensus about guilt or innocence,

showing that they followed one of two different strategies. Some groups

used a  verdict-driven  strategy, which involved each individual stating his/

her initial judgement at the outset – i.e. guilty or innocent – followed by 

discussion in which individuals became advocates for their position andtried to persuade others to change. Frequent polling of views was under-

taken until there was sufficient agreement between the members. Other

groups used an   evidence-driven    strategy, which involved members

reviewing evidence together to develop a common understanding of what

they believed happened – i.e. the most likely sequence of events. Guilt or

innocence was derived from this common understanding. Both strategies

allow a consensus to be reached, but in radically different ways. In add-

ition, procedures designed to support group decision making may be

effective only if they complement the strategy adopted by the group – i.e.

we may need different decision support mechanisms for each strategy.

Some groups, such as parliaments, juries and boards of directors, lay 

down formal rules about how to determine a consensus. For example,

 juries in the United Kingdom are often instructed to reach a unanimous

decision, but later, if this is not possible, then a majority verdict is sought –

i.e. agreement between at least ten of the twelve jurors. Thus there is a

switch from a unanimity rule to a majority rule. Other groups do not have

established rules for consensus and may use less formal procedures. Oneexample is what we like to call a ‘mythical majority’ rule, in which single

individuals periodically claim that their position represents the majority 

324 Decision behaviour, analysis and support

Page 6: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 6/29

and a lack of strong disagreement from others is sufficient for this to be

accepted without further discussion or formal evaluation by voting. In a

very insightful article, Miller (1989) suggests that group decision rules can

be differentiated in terms of their strictness (i.e. the extent to which all

group members must express similar preferences in order for a decision

to be reached) and their distribution of power (i.e. whether or not all

members have the same amount of power in determining the outcome).

He argues that the rule adopted has important implications. For example,

the stricter the rule adopted – i.e. consensus rather than majority voting –

the greater the likelihood that

  a decision will not be reached;

  members’ views will change in direction of the consensus;

  members will not dissent from the decision and express greatersatisfaction with it; and

  members will perceive group discussion as uncomfortable, difficult

and conflictual but also more thorough and adequate.

This last point identifies a dilemma for those wishing to develop pro-

cedures for improving group decision making. Procedures designed to

increase the thoroughness of the process by ensuring that group members

confront issues that are difficult and conflictual may lead to negative

emotions, such as anxiety, apprehension or unpleasantness. This can leadto a diminution in members’ desires to be involved with the group in the

future and may also reduce the likelihood that they will reuse such pro-

cedures.

Groups sometimes exhibit unexpectedly extreme attitudes compared

with the views of the individuals making up that group. For example, in his

master’s thesis, Stoner (1961) reports the results of a study in which he

presented individuals with simple everyday scenarios outlining a choice

between a risky option – e.g. start a new business – and a safe option – e.g.

keep working for the current employer in a secure job. Participants had to

report the minimum probability of the small business becoming successful

that they would require before choosing this risky option. Having pro-

vided this judgement, the research participants formed groups to discuss

these same problems in order to agree a group minimum probability 

 judgement. The initial findings were that group judgements were much

riskier than the average of the individual judgements, suggesting that

coming together as a group had made individuals more prepared to take

risks.Subsequent studies have shown that for some kinds of scenarios the

change in risk attitude was reversed in the direction of increased risk 

325 Groups of decision makers

Page 7: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 7/29

aversion in groups (Isenberg,   1986). The direction of the change was

dictated by the individual pre-group discussion positions; thus, if the

majority favoured risk taking then the shift was towards greater risk taking,

and the reverse was true when the majority initially favoured caution. This

shift in group risk attitudes, called   group polarisation , is now very well

supported by laboratory and field studies. Indeed, it is a specific example

of a more general phenomenon, group-induced attitude polarisation:

namely, if individual members of a group hold initial attitudes or opinions

in a particular direction (e.g. positive or negative views about a particular

option), then this view is enhanced following group discussion (Moscovici

and Zavalloni, 1969; Myers and Lamm, 1976). Whether these shifts are in

risk attitudes or opinions, they are an unintended outcome of decision

making in groups; and one that may have important implications fordetermining the effectiveness of the decisions taken.

Unsurprisingly, work has shown that the majority view has a stronger

influence on the outcomes of group activities than minority views,

regardless of whether these activities are agreed by formal rules such as

majority voting or by other less informal ways, such as the ‘mythical

majority’ rule described earlier. Minorities can be highly influential,

however, when there is a clear set of initial beliefs shared by group

members and when the minority position is clearly a better fit to thesebeliefs (Laughlin and Ellis,   1986). In addition, Kameda   et al . (1997)

introduced the notion of   cognitive centrality , defined in terms of the

amount of knowledge shared with other members. Those with greater

centrality have greater influence whether they hold the minority or

majority position. Apart from providing insights into some of the factors

that allow minorities to influence group decisions, this finding suggests

how those holding minority positions might seek to influence majorities.

Perhaps the most influential and in many ways provocative body of 

research on group failure relates to   groupthink   (Janis,   1972). Having

undertaken a detailed analysis of a series of major fiascos, Janis developed

the groupthink model in terms of five antecedent conditions, eight symp-

toms of groupthink and eight symptoms of defective decision making. The

antecedent conditions thought necessary for groupthink to occur are

  high levels of cohesiveness between group members, derived from

working together often;

  insularity – i.e. little or no inputs or influences from outside the group;

  directed leadership – i.e. a leader who dominates the group;   a lack of established procedures for searching for and the evaluation of 

information; and

326 Decision behaviour, analysis and support

Page 8: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 8/29

  a low perceived confidence of finding an alternative solution to that

advocated by the leader.

Janis suggested that these antecedent conditions give rise to a distinctive

style of decision making that has up to eight negative characteristics or‘symptoms’:

  an illusion of invulnerability that induce overly optimistic beliefs that

positive outcomes will occur, which can lead to extreme risk taking;

  collective rationalisations that lead the group to ignore warnings that

would require a reappraisal of their basic assumptions;

  a belief in the inherent morality of their group that leads members to

overlook crucial ethical and moral issues concerning their intended

actions;   a negative stereotyping of opponents that leads them to underestimate

the power of opponents and to judge them as too immoral to negotiate

with;

  direct pressure on group members to conform that reduces the likeli-

hood that alternative views are presented or considered;

  self-censorship by group members that stops them from expressing

doubts or counter-arguments;

  self-appointed ‘mind guards’ who ensure that contradictory infor-

mation, doubts and counter-arguments from group members or from

outside the group are not considered; and

  an ‘illusion of unanimity’ derived from the previous points above

that leads the group to take what it believes to be the correct decision

with confidence given, it seems, that all group members support it;

this occurs even if many group members have doubts and concerns –

the fact that others have not voiced these leads each individual to

suppress voicing them him-/herself.

The eight symptoms of defective decision making may be summarisedas deficiencies due to poor information search and biased information

processing that lead to an incomplete survey of alternatives and objectives,

a failure to examine any risks associated with the preferred choice, a failure

to work out contingency plans about how to manage negative outcomes

and a failure to reappraise previously rejected alternatives.

The concept of groupthink has had a huge impact on academic research

on group decision making (see, for example, Turner and Pratkanis, 1998),

but has been criticised by some for being grossly overused as a conceptgiven the lack of strong empirical support for the model (Fuller and Aldag,

1998). The model has also been used extensively by individuals and

327 Groups of decision makers

Page 9: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 9/29

organisations outside the academic community to describe and explain

poor group decision making – for example, in trying to understand how 

false premises about weapons of mass destruction (WMDs) led to the

decision to invade Iraq without any apparent consideration of the risks

involved and the failure to develop contingency plans to deal with these

risks. The US Senate Intelligence Committee reported:1

Conclusion 3: The Intelligence Community suffered from a collective

presumption that Iraq had an active and growing WMD program. This

‘group think’ dynamic led Intelligence Community analysts, collectors and

managers to both interpret ambiguous evidence as collectively indicative of a

WMD program as well as ignore or minimize evidence that Iraq did not have

active and expanding weapons of mass destruction programs. This presumption

was so strong and formalized that Intelligence Community mechanismsestablished to challenge assumptions and group think were not utilized.

Overall, the model does provide some useful insights about the errors that

can occur in group decision making and when and why these come about.

Over-reliance on the model may be inappropriate, however, in that it

highlights only a subset of factors that need to be considered when identi-

fying what causes poor group decision making and how this may be rectified.

We began this chapter by identifying the potential advantages that

groups have over individuals when making decisions, and have followed

this by a very brief review of theory and research explaining how and why 

these advantages are not always realised. This body of work not only 

demonstrates the need for procedures to facilitate group decision making,

but also provides important insights about the form these should take

given our knowledge of the pitfalls and limitations involved. We now turn

to the other side of the coin: formal normative models for group decision

making – or the lack of them.

11.3 Arrow’s impossibility theorem

Democracy substitutes election by the incompetent many for appointment by the corrupt few.

(George Bernard Shaw)

Real group decision-making behaviour is far from perfect, then; but what

is ‘perfect’? What is the ideal form of group decision making? Sadly, it is

non-existent. While we have presented normative models of ideal

behaviour for individual decision makers, there is no parallel theory for

1 See http://intelligence.senate.gov/conclusions.pdf  (see also Newell and Lagnado, 2003).

328 Decision behaviour, analysis and support

Page 10: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 10/29

groups of decision makers. Early studies of voting theory (such as that by 

Marquis de Condorcet,   1785) sounded many warnings that it would be

difficult to set up a fair, democratic and rational voting system. For two

centuries efforts were directed at finding better ways of voting. In the

middle of the last century, however, Arrow showed that several key con-

cepts of democracy were incompatible. To give a flavour of some of the

problems, we begin with some examples of the difficulties that can occur

with the simplest and most popular of voting procedures: the   simple 

majority rule . This rule suggests that a group should, as a whole, strictly 

prefer  a   to  b  if a majority of its members strictly prefers  a   to  b . If equal

numbers prefer a  to b  as prefer b  to a , then the group should be indifferent

between  a  and  b . Members who are indifferent between  a  and  b  are not

counted, and so do not affect the group preference.Consider a problem with three individuals and three alternatives,  a, b 

and c . Suppose that the individuals hold the following strict preferences:

individual 1:   a  1  b  1  c ;

individual 2:   b  2  c  2  a ;

individual 3:   c  3  a  3  b .

Note that the strict preference relation has been subscripted to indicate

the individual. Using g  to indicate strict preference for group, the simplemajority rule leads to the following:

a g   b ;  since two out of three prefer a  to b ;

b g   c ;  since two out of three prefer b  to c;

c  g   a ;  since two out of three prefer c   to b :

The simple majority rule can lead, therefore, to intransitive group pref-

erence. Since the money pump argument (section 3.2) is just as persuasive

when money is being pumped from a group as when it is from an indi-

vidual, this is surely a worrying result.

The example misses a very important point, however. The simple

majority rule is seldom, if ever, used to compare all the alternatives

simultaneously. Instead, alternatives are considered a pair at a time:  a 

might be compared with   b ; the less preferred is discarded; the more

preferred is then compared with next alternative,   c ; and so on. For

instance, in the passage of a parliamentary bill in the United Kingdom,

amendments are voted on one by one; adopted, if passed, and dis-carded, if not; and, ultimately, the bill itself is voted upon. Thus, in the

example, the three individuals might first compare   a   with   b  and then

329 Groups of decision makers

Page 11: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 11/29

compare the ‘winner’ with   c . The result would be that the group would

choose  c , since two out of three prefer  a  to  b  and then two out of three

prefer   c   to   a :

a b g ! a 

c g!c 

Suppose, however, that they had begun by considering the choice

between  b  and  c . Then  a  would have been chosen:

b c g ! b 

a g!a 

Similarly, had they begun with the comparison of  a  and   c, b  would have

been their ultimate choice:

a c g ! c 

b g!b 

Thus, the alternative selected by this procedure of pairwise comparison

would be determined simply by the order in which the alternatives were

considered, and not primarily by the group members’ preferences. In this

example this prospect is not, perhaps, very worrying. The individuals’

preferences are clearly substantially and symmetrically opposed, and there

is a strong case for arguing that the group as a whole should be indifferent

between the three alternatives. The ultimate choice is therefore a matter

of little concern. In other cases, however, we might be more concerned.

Should the group’s choice have any dependence upon the order in which

alternatives are considered?

There is another worrying aspect to this problem. Suppose that the first

individual knows his/her companion’s preferences. If the alternatives are

to be compared in the order a  and b  and then the winner with  c , he/she will

be able to predict that the ultimate choice will be c . Suppose that he/she liesabout his/her true preferences, however, while the others honestly reveal

theirs. If he/she says that he/she holds  b 1  a 1  c , the ultimate choice will

be determined as

a b g ! b 

c g!b 

Thus, by lying, he/she will ensure that the group selects   b , which he/she

prefers to the ‘true’ group choice   c . In these particular circumstances the

simple majority rule encourages him/her to lie, or, less pejoratively, to vote

tactically. That surely is a little worrying. Arrow sought to characterise the

330 Decision behaviour, analysis and support

Page 12: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 12/29

properties that we would expect of a democratic voting system. He framed

these as axioms and then showed them to be mutually contradictory.

Essentially, no voting system – or  constitution , as he called it – exists such

that in every possible circumstance it satisfies some basic principles of 

rationality. This argument is summarised in table  11.1.

Since Arrow proved his uncomfortable result many have tried to reframe

his axioms slightly to avoid the impossibility, or to argue that one or more of 

them is irrelevant to democracy and can therefore be dropped. All have,

essentially, failed. Worse, it transpires that versions of the impossibility still

hold if any one of the properties is abandoned (Kelly,  1978). Arrow has

identified a real inconsistency in our hopes and demands for rational

democratic decision making. Others have enlarged the debate and shown

that any constitution is susceptible to manipulation through strategic vot-ing, the dishonest revelation of preferences or agenda rigging. Perhaps the

most hopeful way forward seems to be to allow group members to express

not just their preference rankings but also their strength of preference.

While there are a number of mathematical results that offer some

hope, all fail because they need interpersonal comparisons of preferences

(Bacharach, 1975; French, 1985, 1986; French and Rıos Insua, 2000; Hodge

and Klima,   2005; Raiffa   et al .,   2003; Taylor,   2005). Such comparisons

require, for instance, an unambiguous interpretation of ‘I prefer coffee to teamore than you prefer a sports car to an mp4 player’. Despite many attempts,

no sensible operational meaning has been offered for such statements.

However one formulates the issues, there seems to be no way to develop

mechanistic algorithms and prescriptions for voting and group decision

making that are fair, just, democratic, honest, open . . . : choose your

adjective embodying a good moral imperative!

French et al . (2007c) note that over the years several distinct approaches

to group decision analysis have been proposed, as follows.

GDMGSEU    This assumes that the subjective expected utility model

applies at the group level and, moreover, that group proba-

bilities and utilities are constructed from those of its

members. Thus, the process is to elicit each group member’s

subjective probabilities and utilities, and combine these into

group probabilities and utilities, respectively. Finally, the

corresponding group expected utilities are calculated and the

resulting ranking used to recommend choice (Bacharach,1975).

331 Groups of decision makers

Page 13: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 13/29

Table 11.1  Arrow’s impossibility theorem

Suppose that there are n  individuals who are jointly responsible for choosing an action from a given finite set of 

possible actions. Let <i  be the weak preference ordering of the i th individual. Similarly, i  will be his/her strict

preference. We also refer to the entire preference ordering of the i th individual by <i  (i.e. to the set of alternative

actions as ordered by his/her preferences). The preference orders, <i , may be selfish  and simply reflect what the

individual wants him-/herself irrespective of what that may mean for the other group members; or they may be

altruistic  and reflect what the individual wants for the greater good of the group. Given the  n  individuals andtheir preferences  <i ,  i  ¼  1, 2,   . . . n  , the problem is simply stated. We must prescribe how to combine them

into a preference order for the group as a whole. We write <g  for this order and  a <g  b , a g  b  for particular

statements of group preference. The voting system or mechanism whereby <1,  <2,   . . .  ,  <n  are combined to

give  <g   is called the  constitution  of the group. Arrow (1963) suggests that the following axioms encode the

minimum requirements of justice, fairness and rationality that we might ask of a constitution.

Axiom A1:  weak ordering 

<1,  <2,   . . .   ,  <n  and  <g  are all weak orders – i.e. complete and transitive (chapter 3 section 2)

Axiom A2:  non-triviality 

(i) There are at least two members of the group:  n  

 2.(ii) There are at least three alternatives.

Axiom A3:  universal domain 

<g  is defined whatever  <1,  <2,  . . .  ,  <n  may be.

Axiom A4:  binary relevance 

Let <1, <2,  . . .  , <n  be a set of individual preference orders over a set alternatives,  A. Let <01, <0

2,  . . .  <0n  be

another set of individual preference orders over a set of alternatives, A0. Suppose that alternatives  a  and  b  lie

in both A  and  A0: {a , b } A\A0. Suppose further that <1, <2,  . . .  , <n  and <01, <0

2,   . . .   , <0n  are identical on

{a ,  b }:  8i,

a  <i  b  a  <0

i  b  and  b  <

i  a  b  <0

i  a 

Then the constitution should lead to the same group preference between  a  and  b :

a  <g  b  a  <0g  b  and  b  <g  a  b  <0

g  a 

Axiom A5:   Pareto’s principle for strict preferences 

If every individual holds  a  i  b , then the group holds  a  g b.

Axiom A6:  no dictatorship 

There is no individual whose preferences automatically become the preferences of the group independently of 

the preferences of the other members.

Axiom A1 is a common requirement for rational preferences, and, while we might well acknowledge that

democratic group decision making should encompass the preferences of all, whether ‘rational’ or ‘irrational’,

we would also surely agree that it should not be permitted to fail just because the group members were

rational. Axioms A2 and A3 simply require that the constitution applies to all situations however large

the group and however many alternatives. Axiom A4 is a version of the axiom of the irrelevant alternative

(French, 1986; French and Rıos Insua, 2000), which requires that the group preference between two

alternatives depends only on the members’ preferences between those two alternatives and not on their

preferences between any others. Axiom A5 is a simple requirement of unanimity. Axiom A6 prohibits the

existence of a dictator – even a hidden one, who neither knows that he or she is, nor does any other group

member. Unfortunately, Arrow has shown that these requirements are mutually contradictory.

Arrow’s impossibility theorem

There is no constitution that allows <g  to be defined from <1, <2, . . . , <n in a manner that is consistent with 

axioms A1–A6 .

Proof  

See, for example, Arrow (1963), French (1986), French and Rıos Insua (2000) and Kelly (1978).

Page 14: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 14/29

GDMvote    An analyst works with each individual and develops a

personal decision analysis to guide his/her choice. In the

light of this understanding, each individual votes within

the group and a group choice is made according to the

result. In variants of this, the numerical values of the

individuals’ expected utilities are used to indicate their

strength of preference and this information is incorpor-

ated into the voting (Rıos and Rıos Insua, 2008).

GDMSupraDM    A  supra decision maker   is imagined to exist. She observes

the entire elicitation and decision analysis process for each

individual and altruistically uses this knowledge to construct

a single decision analysis for the group. The choice is made

according to the supra decision maker’s analysis (Keeney and Raiffa, 1976).

GDMFac    The group gathers together in a facilitated discussion of 

issues (we discuss facilitation in section 5). Through

discussion between participants, consensual agreement

is reached on group probabilities and utilities without

formally eliciting individual ones. A group analysis is

developed and areas of disagreement explored via sensiti-

vity and robustness analysis. Ultimately, a decision isobtained by consensus without formal voting (Eden and

Radford, 1990; French, 2003b).

The many paradoxes and impossibilities stemming from Arrow’s and

related results have led most decision analysts to doubt the efficacy of 

GDMGSEU   and GDMvote . Note that the generality of these impossibility 

results means that the difficulties apply not just to SEU approaches

but, essentially, to all decision analytic methodologies. Sadly, such a rec-

ognition is not common among the designers of some GDSSs. Some

systems require individual inputs, manipulate them and ultimately pro-

duce a group ranking, inevitably risking an inconsistency, irrationality or

injustice or laying themselves open to manipulation.

Approach GDMSupraDM   seems more promising, because all interper-

sonal comparisons are made within the mind of the supra decision

maker, and it is the issue of defining valid interpersonal comparisons

that tends to cause the paradoxes and inconsistencies. In some cases,

the supra decision maker actually exists. There might be an arbiter,formally responsible for recommending a decision that balances all group

members’ perspectives; or there might be a government agency, which

333 Groups of decision makers

Page 15: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 15/29

has the responsibility and accountability for making the decision but

does want to take into account the views of stakeholders. In most cases,

however, the non-existence of a supra decision maker remains a problem:

she is a fiction, and that creates a fatal flaw in this approach. She has to be

constructed by agreement within the group, and this leads to a further

group decision, arguably as hard as the first, and an infinite regress.

Ultimately, we believe that any mechanistic concept of group decision

making is flawed and ill-defined. Beliefs, preferences, logical analysis –

all reside in each of the individual’s minds, not in some disembodied group

mind. We look upon groups not as some entity that possesses the power to

decide but as a social process, which translates the decisions of the individual

members into an implemented action (Dryzek and List, 2003; French, 1985;

French and Rıos Insua,  2000). For this reason, we prefer the expression‘groups of decision makers’ rather than ‘group decision making’. Our

approaches thus fall into the approach GDMFac ; or perhaps one other. The

four approaches discussed so far assume that the group  wants  to cooperate

and reach a consensus. It might be that they are more self-serving and wish

to negotiate a good end point for themselves.

GDMNeg    Bargaining, negotiation analysis or arbitration techniques are

deployed to define a process in which the group interacts and

discusses a series of solutions, converging through negotiations

on a deal or policy that all find acceptable (Raiffa, 1982; Raiffa

et al .,  2003).

We do not intend to discuss negotiation methods in this text, though we

do provide some comments and a guide to the literature in section  14.6.

11.4 Procedures for improving decision making in groups

If you really get the right people, and you’ve got them working together as team, whether it’s

in business, whether it’s in science, whether it’s in politics, you can make a big difference.

(Steve Case)

The research reviewed in section   2   provides some important insights

about the nature of the social processes that underpin decision making in

groups, indicating how and why these may be suboptimal. In this section

we briefly review procedures for improving group deliberation that

address and in some cases seek to modify the nature of these processesby providing social interaction guidelines, imposing a structure on

how members interact and choosing group members appropriately. In

334 Decision behaviour, analysis and support

Page 16: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 16/29

section 5 we discuss conferencing and facilitation techniques which can

provide further improvements.

Janis (1972) suggested several guidelines by which groups might counter

the negative effects of groupthink.

  The leader should encourage each member to air objections and

doubts. This requires the leader to be willing to accept criticism.

  The leader should initially remain impartial in discussion, stating his/

her own preferences and beliefs only after all the group members have.

  The group should form ‘breakout’ subgroups to discuss issues

independently, then reconvene to compare conclusions and hammer

out any difficulties.

  Outside experts should be included occasionally and encouraged to

challenge group views.

  At each meeting one member should be appointed to play devil’s

advocate.

While these seem eminently reasonable and address some of the group-

think limitations, there has been little research to test whether groups can

follow these guidelines or, indeed, whether they actually lead to an

improvement (see ’t Hart, 1998, for other suggestions about protecting

against groupthink).Hall (1971) developed a consensus approach built on a set of rules for

inducing a thorough, open and constructive discussion involving all

members of a group. These rules, presented in table  11.2, do appear to

facilitate behaviour likely to achieve some of the benefits of group decision

making identified in the introduction (e.g. rules 4 and 5 encourage a full

exchange of information between group members) and to limit damage

from some of the factors known to limit the effectiveness of group decision

making (e.g. rule 3 limits the damaging effects of conformity). Eils and

John (1980) showed that groups that had been trained to use these rules

made better decisions – in their experiment, a better prediction of whether

lenders would or would not default on loans. Although training groups to

follow rules of this kind seems to have great potential, this approach to

improving group decision making remains relatively underdeveloped.

A second approach to improving group decision making has been to

impose a structure on how groups interact. One such technique, devil’s

advocacy (Mitroff and Mason, 1981), entails assigning a particular role to

an individual or subgroup that involves them critiquing all assumptionsand recommendations advocated by the main body of the group in an

attempt to show why these should not be accepted or adopted. Proponents

335 Groups of decision makers

Page 17: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 17/29

of this approach believe that appropriate assumptions and recommenda-tions will survive such criticisms but inappropriate ones will not. A related

technique, dialectical inquiry, involves splitting the group in two with each

subgroup developing a different set of recommendations based on con-

trary assumptions. Then the two subgroups engage in an in-depth, critical

debate about the recommendations and their underlying assumptions.

Finally, they agree on those that have survived the scrutiny of debate, and

develop recommendations based on them. The detailed scrutiny of deci-

sions and their underlying assumptions is assumed to improve the quality 

of decision making.

Both techniques have the potential to reduce the chances of groupthink 

occurring, given the rigorous and critical evaluation of decision options

and their underlying assumptions. In addition, these procedures may also

reduce the likelihood that groups will be affected by some of the factors

known to reduce effectiveness – e.g. social loafing. It is less certain, how-

ever, that other factors, such as evaluation apprehension and conformity,

are managed effectively by these two techniques.

Direct evidence on the effectiveness of these techniques is mixed (seeSchwenk,   1990). Schweiger   et al . (1986) report a laboratory study that

involved giving groups of participants a relatively complex business

Table 11.2  Hall’s group interaction rules

(1) Avoid arguing blindly for your own assumptions and recommendations. Present your position clearly,

logically and persuasively, but consider carefully the comments and reactions of other group members.

If you present the same points again, take comments and reactions into account.

(2) Avoid making ‘win-lose’ statements in your discussion. Discard the notion that someone must win and

someone lose in the discussion. When impasses occur, look for the next most acceptable solution for all

parties.

(3) Avoid changing your mind simply to avoid conflict and reach agreement. Withstand pressures to yield

that have no logically sound foundation. Strive for enlightened flexibility; avoid mere capitulation.

(4) Avoid conflict-reducing techniques such as majority voting, tossing a coin, and the like. Differences of 

opinion indicate an incomplete exchange of relevant information on someone’s part; press for

additional sharing of task or emotional data when it seems in order.

(5) View differences of opinion as natural and helpful rather than as a hindrance to decision making.

Generally, the more assumptions and recommendations expressed the greater the likelihood of conflict,

and the richer the resources used in solving the problem at hand.

(6) View all initial agreements as suspect. Explore the reasons for the apparent agreement. Make sure that

people have arrived at similar recommendations either for the same reasons or for complementary 

reasons before incorporating such recommendations into the group’s final set.

Source:  Schweiger   et al . (1986).

336 Decision behaviour, analysis and support

Page 18: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 18/29

problem that required a strategic decision. They compared group per-

formance under dialectical inquiry, devil’s advocacy and the consensus

approach. The findings revealed that dialectical inquiry and devil’s advo-

cacy gave rise to higher-quality recommendations and assumptions than

the consensus approach, and that dialectical inquiry was better than devil’s

advocacy with respect to the quality of the assumptions used to make the

decision. Despite this, participants in the consensus groups expressed

greater satisfaction with their group experience, revealed a greater desire to

continue working with their group and were more accepting of their group

decisions than participants in the other two groups. This is an important

finding, in that is suggests that there may be a tension between improving

the quality of the decision, through confrontational approaches such as

devil’s advocacy and dialectical inquiry, and how pleasant group membersfind the process and their willingness to continue working together. Both

factors need to be considered, and the relative importance of each may 

vary from situations to situation. In some situations the need to keep a

group together may outweigh the quality issues – for example, when there

is a group of volunteers running a charity, maintaining the support of 

these volunteers may be of paramount importance.

A rather different technique for structuring the group process,  Delphi ,

was originally developed by Dalkey and his co-workers at the Rand Cor-poration. It is a procedure designed to ‘obtain the most reliable consensus

of opinion of a group of experts . . . by a series of intensive questionnaires

interspersed with controlled opinion feedback’ (Dalkey and Helmer, 1963:

458; see also Linstone and Turoff,  1978). The technique does not require

that groups meet face to face. Instead, they respond anonymously to a

series of questions, and their answers are synthesised and then fed back to

members for further comments and the generation of potential solutions;

these are again synthesised and sent out for further comment and evalu-

ation, and so on. There may be several such iterations. Thus the group

interacts remotely rather than face to face.

Rowe and Wright (1999) argue that the technique maintains many of 

the positive aspects of group decision making – e.g. increased knowledge

and experience as a result of including many individuals and the increased

creativity that occurs when people work together – while minimising some

of the negative social, personal and political processes – e.g. different

power relations are inhibited because contributions are not attributed to

particular members, and conformity effects are limited because contri-butions are made without knowledge of the views of others. In addition,

the technique allows a relatively large and geographically dispersed group

337 Groups of decision makers

Page 19: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 19/29

t o w o rk t o ge th er s in ce t he re a re n on e o f t he u su al c o ns tr ai nt s a ss oc ia t ed

w it h o rg an is in g a f ac e- to -f ac e m ee ti ng . H av in g r ev ie we d t he e xt an t

r es ea rc h, R ow e a nd W ri gh t c on cl ud e t ha t t he t ec hn iq ue i s e ff ec ti ve i n

i m pr o vi n g g r ou p d e ci s i on m a ki n g ( b ut s e e C o o ke ,   1991).

T he D el ph i t ec hn iq ue h as p ro vi de d t he s ti mu lu s f or a m or e g en er al

d is cu ss io n a bo ut w he th er i t i s b et t er f o r g ro up i nt e ra ct i on s t o b e f ac e t o

f ac e o r c om pu te r- me di at e d. T he e vi de nc e s ug g es t s t ha t, f o r d iv er ge nt

d i sc u ss i on s u ch a s t h e f o rm u la t i on p h as e o f d e ci s i on a n al y si s , c o mp u te r -

m ed ia te d d is cu ss io n i s m or e e ff ec ti ve , p er ha ps b ec au se i t a ll ow s t he

m em be rs t o t hi nk a lo ng t he ir o wn l in es w it ho ut c on st an t d is tr ac t io n b y    

t he s ug ge st io ns o f o t he r g ro up m em b er s. F o r c on ve rg en t d el ib e ra t io n i n

t h e e v al u at i o n a n d a p pr a is a l p h as e s, h o we v er , t h e b a l an c e s h if t s t o w ar d s

f a ce - t o- f ac e m e et i ng s ( K er r a n d M u rt h y,   2004 ).S o me h av e a rg ue d t ha t g ro up d el ib e ra t io ns c an b e i mp ro v ed b y d el ib -

e r a t e l y s e l e c t i n g g r o u p m e m b e r s i n s u c h a w a y a s t o e n s u r e t h a t t h e y h a v e

h et er og en eo us r at he r t ha n h om og en eo us v ie w s – i . e. i ni t ia ll y d is ag re e

r a th e r t h an a g re e . C h oo s in g m e mb e rs i n t h is w a y i n cr e as e s t h e l i k el i ho o d

t ha t g ro up s w il l d eb a te a nd d is cu ss a ss um pt io ns a nd p os si bl e a ct io ns .

R es ea rc h h as s ho w n t ha t h et er og en eo us g ro up s s ho w r ed uc ed l ev e ls o f                  

ov e r c onfi d e n ce ( S n i e ze k ,   1992 ) , a re l ess p ro ne t o u nd ere st im at e r is ks

( W i l l i a m s a n d T a or m i n a ,   1992 ) , r e a c h m or e a c c u r a te j u d g e m e n ts ( S n i e z e k  a nd H en ry ,   1989) , g en er at e m or e h yp ot he se s ( Cr ot t   et al .,   1998) and

e xc ha ng e m or e i nf or ma ti o n, i nc lu di ng u ni q ue i nf o rm at i on ( Br od be ck  

et al .,   2002) . I n d e e d , S c h u l z - H a r d t et al . (2002 ) s h o w t h a t s e l e c t i n g g r o u p

m em be rs i n t hi s w ay m ay b e a m or e e ff ec ti ve w ay o f i mp ro vi ng g ro up

d ec is io n m ak i ng t ha n s t ru ct ur ed t ec hn iq ue s s uc h a s d ev i l’ s a dv oc ac y.

T he se fi nd in gs h ig hl ig ht t h e i mp or t an ce o f r ec ru it i ng i nd iv id ua l s w i th

d if fe re nt b ac kg ro un ds , b el ie fs a nd v al ue s i f w e a re t o m ax im is e t he f ul l

p ot e nt i al f ro m w o rk in g w it h g ro up s o f D Ms .

11.5 Facilitated workshops and decision conferencing

W h e n y o u a r e a B e a r o f V e r y L i t t l e B r a i n a n d y o u T h i n k o f T h i n g s , y o u fi n d s o m e t i m e s t h a t a

T h i n g w h i c h s e e m e d v e r y t h i n g i s h i n s i d e y o u i s q u i t e d i f f e r e n t w h i c h i t g e t s o u t i n t o t h e o p e n

a n d h a s o t h e r p e o p l e l o o k i n g a t i t . ( A . A . M i l n e )

In the p r ev i o us s e ct i o n w e i de nt ifi ed g ui de li ne s a nd s tr uc tu re d p ro c ed ur es

t ha t c an c o un te r s om e o f t he e rr or s a nd b i as es t ha t a ri se i n g ro up d el ib -

e r at i o ns . I n t h i s s e ct i o n w e d i sc u ss f a ci l i ta t e d w o rk s ho p s, o f t en k n ow n a s

decision conferences    ( Ed en a nd R ad fo rd ,   1990; F re nc h,   1988; P hi l li ps ,

1984 ) . T h es e d r aw s u p on t h re e k e y m e th o do l o gi e s:

338 Decision behaviour, analysis and support

Page 20: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 20/29

  facilitation, in which a facilitator, who has no responsibility or

accountability for the consequences of the decision, joins the group to

structure, smooth and enhance the deliberative processes;

  decision analytic models, to help the DMs understand themselves, the

context and the issues before them; and

  interactive software, to explore and display the implications of the

models.

We have noted that groups can fall prey to various biases and other

dysfunctional behaviours. One of the key roles of a facilitator  is to counter

these. A facilitator is skilled in the process of group discussion, but may 

have little expertise in the context of the issues at hand. His role is to

smooth the group’s work, to help the process and make the team more

productive and creative. Phillips and Phillips (1993) summarise the key 

functions of a facilitator as observing, attending, maintaining awareness of 

feelings and intervening. The content of the discussion, however, comes

entirely from the group itself. The group members ‘own’ the problem,

have knowledge of it, have access to relevant data and experts and are

responsible for its resolution.

In a sense, a facilitator is no more than an impartial chairperson or

group leader, but in practice his ‘distance’ from the group is far greater

(Maier, 1967). Because he does not share in the ownership of the problem,he may concentrate on:

  encouraging members of the group to contribute ideas and listen to

those of others;

  assuming responsibility for accurate communication between the

members, perhaps cutting through jargon or simply making sure that

no one is too shy to say ‘I don’t understand’;

  protecting minority views and ensuring that they are debated fairly;

  being sensitive to people with unexpressed feelings and views andhelping them enter the discussion;

  calming conflict by keeping the group task-oriented rather than

personality oriented;

  summarising the position at appropriate points in the discussion; and

  generally keeping the discussion moving and focused on the task 

in hand.

By following these guidelines the facilitator is limiting the extent to which

the group falls prey to the factors known to inhibit group decision making

discussed earlier in this chapter. Indeed, Ackermann (1996) has surveyed

over 100 managers who have taken part in facilitated workshops. She finds

that the process was positive for all the reasons given above.

339 Groups of decision makers

Page 21: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 21/29

Facilitators concentrate their attention on the   process , leaving the

members of the group free to contribute, explore, shape and understand

content . Miranda and Bostrom (1999) have suggested that content facili-

tation can be more effective, however. In truth, both content and process

interventions are necessary, but the point that we would make is not that

facilitation should concentrate on process or content but that process and

content interventions should be explicitly separated so that the group

perceives the process facilitation as neutral and unbiased. It is important

that group members play the primary role in providing the content of the

discussion, identifying opportunities, creating and evaluating the options

and generating the action lists to implement their decisions. Through their

total involvement in the creation of strategy they become fully committed

to its implementation. They ‘own’ the strategy. Moreover, because of theirshared understanding of the reasons behind its adoption, they can explain

it to others.

All the above suggests what  a facilitator should do: how  he should do it is

another matter. How should he intervene to enhance the work of the

group? There are some tricks of the trade. Generally, the facilitator should

raise issues neutrally, asking open questions – although sometimes, when

the group is drawing together behind a single viewpoint, he may play 

devil’s advocate and press an alternative view to test whether groupthink is rearing its ugly head. Ignorance can be a great advantage: a facilitator

may often move a group forward by asking a very naıve question and

uncovering hidden, perhaps contentious assumptions or misunderstand-

ings. Because he is an outsider, he often questions jargon and so clarifies

discussion for the whole group. Not everyone in the group may be au fait

with all the jargon used in an organisation, but some may lack the courage

to indicate their ignorance.

Not all interventions require the facilitator to distance him calmly from

the issues. Occasionally, there may be benefit in his pressing a point forcibly,

particularly if a member of the group is trying to take control of the process

or if the general level of stress has fallen below that needed for productive

activity. Generally, though, a facilitator is wise to hold his temper and

intervene gently, catalysing rather than directing or confronting.

The process of decision analysis and its tools can of themselves provide

very effective interventions. The prescriptive decision analysis cycle

organises the general flow of discussion, moving through issues one by 

one, concentrating on each in turn. This avoids the confusion that can becaused by simultaneously considering many issues and darting between

them. Soft modelling methods can help foster creativity, providing a

340 Decision behaviour, analysis and support

Page 22: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 22/29

framework to brainstorm effectively. The model structuring, elicitation,

evaluation and sensitivity analysis cycle helps to move the discussion for-

ward productively, focusing on one issue at a time. Sensitivity analysis often

defuses heated but irrelevant debates, concentrating attention on the real

issues that matter to the problem at hand. Finally, the developing model

provides a very effective vehicle for communication.

In a decision conference the DMs responsible for the decision meet

together, ideally away from the distractions of their normal working

environment, to discuss and explore the issues. The entire group responsible

for a decision, including relevant experts and perhaps key stakeholders,

should take part in the conference. Ideally, the size of the group should be

between seven and fifteen people, so that there are sufficient participants to

stimulate wide-ranging discussions yet not so many that it is difficult foreach to contribute. The meeting concentrates entirely upon the issues that

led to it being called. There are no time-outs to consider peripheral matters

‘while the team are together’. The facilitator is usually assisted by one or

more decision analysts. The facilitator leads the meeting, guiding the

discussion forward in a constructive fashion. The analysts build and run

decision models, generating representations of the issues as they arise. The

models are projected and the group can look into their implications

immediately, modifying the models as they explore different assumptionsand inputs, informed by sensitivity analyses of previous models. The

analysts also record the development of the debate and the reasoning

behind the judgements and decisions made. At the conclusion of the

conference the group members are able to take away a record of important

conclusions and an action list with them, along with a record of how and

why these evolved.

A decision conference is generally a two-day event. Other time scales are

possible (see, for example, Hamalainen and Leikola,   1996), but the

inclusion of a night is of considerable advantage. In the evening the group

members are able to relax together and reflect on the progress and dis-

cussion so far. This reflection, together with the distance from the day’s

deliberations that a night’s sleep brings, helps members acquire a more

mature perspective on the issues that concern them. Without the overnight

break some may have second thoughts soon after the conference ends,

perhaps on the journey home, and much of the value of the event will be

dissipated as their commitment to its conclusions evaporates.

Each decision conference is different. It evolves according to the needsof the group and not according to some fixed agenda. There are common

themes and patterns, however. The facilitator is always careful to ensure

341 Groups of decision makers

Page 23: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 23/29

that the opening discussion is as wide-ranging as possible, encouraging

divergent thinking. Decision conferences are demanding on the time of 

busy DMs and so are called only to deal with complex, unstructured

strategic issues. During the initial phase the facilitator may simply allow 

the discussion to develop or he may use soft modelling techniques to add

structure to the perspectives being developed. Sometimes the development

of such models can occupy the entire event, particularly those dealing

with the broad strategic intent and mission of an organisation (Ackermann

et al ., 2004; Eden and Ackermann, 1998).

With the issues and context defined, the facilitator moves into a decision

modelling phase, encouraging convergent thinking on key issues that soft

modelling has identified. He generally aims to build an initial ‘quick and

dirty’ model encapsulating the issues and concerns well enough to give afirst indication of ‘an optimal course of action’. With good scheduling by 

the facilitator, this will occur just before the overnight break, giving the

DMs the opportunity of an extended period to reflect upon the analysis to

date. By the next morning they will usually be clear on one thing: whatever

the model is doing, it is   not  reflecting an optimal course of action. They 

will note flaw upon flaw with the analysis – forgotten issues, ill-formed

 judgements and so on. The associated feelings of frustration – ‘Did we

waste yesterday entirely?’ – will provide added impetus to revitalise theirdiscussions. The facilitator and his team will rebuild the model, adding

additional features as necessary, and explore it via sensitivity analysis. The

decision analytic cycle continues until the model is requisite.

The final stage of a decision conference is to work with the group

members to provide a summary of the conclusions and strategy that they 

can take with them, along with an action list to implement the decision.

One of the reasons for the success of a decision conference is that, because

it is   their   model and   their   analysis, the DMs are highly motivated to

implement   their   conclusions. There is considerable evidence that, when

decisions are implemented by managers who neither own nor fully under-

stand the decision, the outcomes may not be as successful as anticipated

(Borges   et al .,  2005, 2006). This final stage of summarising and allocating

tasks ensures that the DMs may return to their usual working environment

and colleagues able to communicate the conclusions succinctly. Decision

conferences ensure that the participants own the strategy, are committed to

its implementation and can communicate their arguments and enthusiasm

effectively.In case vignette 11.1 we reflect on the decision conferences run during

the International Chernobyl Project.

342 Decision behaviour, analysis and support

Page 24: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 24/29

Case vignette 11.1 Reflections on the Chernobyl decision conferences

In section 7.8 we described the multi-attribute analysis in the International Chernobyl

Project. This analysis was developed over a sequence of five decision conferences.

The conferences were rather large: all had more than thirty participants, not including

the support team and observers. This did not cause the anticipated difficulties, however.

Having more than thirty people in the room was no less manageable than fifteen. The

presence of observers, albeit bound by strict rules of confidentiality, might also have

been expected to cause problems, yet no difficulty was encountered. Discussion ebbed

and flowed, developing many different perspectives on the issues.

Another standard rule of decision conferences is that no papers or prepared material

are allowed on the first morning, so that participants are forced to present their views,

composing their words on the spot. This encourages rapid and wide-ranging discussion

of the main issues and concerns. It also stops individuals hiding behind prepared data

and entrenched positions. Many of the participants, true to their political backgrounds,

came prepared to read speeches. They were allowed to: several were ministers of 

sufficient authority that it would not have been possible to stop them! The facilitator 

simply sat quietly, politely listening but not reacting to their words, and then let the

silence drag out after they had finished, until they had to continue in their own words –

and live discussion developed.

Some interesting aspects of working in a different society and culture emerged. For 

instance, there is no Russian word with a meaning corresponding to ‘trade-off’. This

made the development of multi-attribute value models difficult, but, since we used

swing weighting techniques, not impossible. In fact, explaining the principles of the

models without relying on the term ‘trade-off’ is arguably a very good discipline for 

decision analysts.

In the past Soviet society hid many value judgements within appeals to scientific

objectivity. Moreover, the State had promised the absolute safety of all its projects,

denying the need for any risk analysis: if it wasn’t safe, the State wouldn’t have decided to

build it, would it? In dealing with the after-effects of Chernobyl, those involved had to

reappraise many of their working assumptions and beliefs, and to tease out and examine

many value judgements. Science cannot remove all the contamination from the envir-

onment and return it to an absolutely safe condition. Not that it had ever been absolutely

safe before the accident: there is always some background radiation and plenty of other 

non-radiological risks. It is not for science to place a relative social cost on the occurrence

of a radiation-induced genetic effect compared with a death from a radiation-induced

cancer. That is a value judgement for a society to make through its political and other 

institutions. The discussion at the conferences time and again returned to and examined

such value judgements.

It is often said that people are the same the world over. In a very small sense, these

conferences provided some evidence to support this. Once the concerns and issues had

been brought to the fore and the discussion had begun in earnest, there was no

noticeable difference in the interactions and group behaviour of the participants

343 Groups of decision makers

Page 25: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 25/29

11.6 Game theory, negotiation and bargaining

In principle, every social situation involves strategic interaction among the participants. Thus,

one might argue that proper understanding of any social situation would require

game-theoretic analysis. (John Harsanyi)

In 1944 von Neumann and Morgenstern published their seminal text

Theory of Games and Economic Behavior  (second edition, 1947). It was and

is a monumental work, which developed utility theory from a simple set of 

axioms, essentially completing the definition of rational economic man

and, above all, laying the foundations of game theory. Game theory is thestudy of conflict and cooperation between a number of independent

players. It differs from the direction that we have taken earlier in this

chapter. There we confined our attention to supporting a group of DMs

working together to resolve some set of issues. We recognised that the

import of Arrow’s theorem was that it was not possible to construct a

theory of a democratic group of DMs who together acted as some rational

entity. Thus we needed to view the group as a social process and our role as

analysts and facilitators was to smooth and support this process.

Game theory takes a quite different tack. It posits several independent

players, each with his/her own preferences and each able independently to

choose one of a set of actions to take. The combined effect of their indi-

vidual actions leads to the overall outcome and the ‘share’ each receives.

The purpose of game theory is to investigate what outcomes might arise if 

one assumes that each plays rationally to achieve his/her own ends. There

are variants depending on whether the players compete or cooperate in

coalitions and upon the definition that is taken of rationality. Originally 

von Neumann and Morgenstern suggested that a  maximin  definition of rationality might be adopted. This is a very pessimistic way of taking

decisions. For each possible action, the DM identifies what the worst is that

compared with similar events in the West. This was fortunate, because there had been

concern that the interventions that the facilitator might make would have effects other 

than those intended. They did not. To give one example: on one occasion he was having

difficulty getting the group members to ‘own’ the problem and admit that they had the

knowledge to discuss it. Repeatedly a need was expressed to consult others outside theroom. It was a hot afternoon, and the doors and windows were open. The facilitator used

this to make an intervention to the effect that no one was going to come through the

door and windows and ‘locked’ the team in with their problem. The effect was dramatic.

Within minutes the group took responsibility and discussion moved forward rapidly.

344 Decision behaviour, analysis and support

Page 26: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 26/29

could happen and then chooses the action that has the best worst outcome –

i.e. maximin   ¼   ‘maximise the minimum pay-off’. In strictly competitive

games, in which all the players are out to maximise their own personal gains

whatever the cost to the others, this might be a sensible criterion; for

any games with some element of cooperation, however, the rationality of 

maximin is questionable.

Since von Neumann and Morgenstern’s original work, other approaches

to rationality have been investigated, notably the Bayesian (Harsanyi,

1977; Smith, 1996). For general reviews of game theory, see, for example,

Colman (1995), Osborne (2003) or the classic introductory text by Luce

and Raiffa (1957). We would emphasise that, while game theory does

provide some guidance for individual players, its primary concern is to

explore conflict and cooperation between players. This is a subtle dis-tinction, but it is important. Players looking for advice on their decision

making in a game might be better served looking to the general literature

on decision analysis rather than game theory itself (French, 1986; Kadane

and Larkey,   1983; Kadane   et al .,   1999; van Binsbergen and Larx,   2007).

There have many behavioural studies of the strategies actually adopted by 

players in games (Camerer,  1997, 2003; Colman, 1995). Finally, we note

two game theoretic approaches that can help DMs formulate issues in

game-like contexts, hypergames (Bennett, 1977) and metagames (Howard,1971), which are both related to drama theory (Rosenhead and Mingers,

2001).

Closely related to game theory and its exploration of competition and

cooperation between several players is the literature on bargaining and

negotiation. Raiffa  et al . (2003) provide a comprehensive introduction to

these topics, exploring normative, descriptive and prescriptive approaches.

Two earlier texts by Raiffa are also still very relevant: Luce and Raiffa

(1957) and Raiffa (1982). Raiffa et al . (2003) adopt an approach that – like

ours – explores the normative, prescriptive and descriptive issues. They 

argue that negotiation is similar to other kinds of decision making: there is

a normative theory determining what people should do if they wish to be

rational, but this is not descriptive of what people actually do. Instead,

people engage in heuristic forms of thinking, which can lead to error and

bias; error and bias in this context can include such negative outcomes as

not reaching a settlement, or ending up with a settlement that is sub-

optimal for one or both parties.

Bazerman (2006) devotes a chapter to describing some of these heur-istics, many of which we have already met in earlier chapters of this book.

For example, he considers framing effects (see section   2.4) that show 

345 Groups of decision makers

Page 27: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 27/29

people are more likely to reach a settlement when the negotiation is framed

in gains rather than losses; overconfidence and optimism (see sections 2.6,

2.7 and 3.8) that lead people to believe too strongly that they are right and

that their protagonist is wrong; and anchoring effects, whereby possible

settlements presented early in a negotiation have a constraining effect

upon the discovery of better ones later.

In response to these potential mistakes, Raiffa   et al . (2003) outline a

procedure to follow that, from users’ standpoints, is prescriptive for them

and descriptive of their protagonists. In particular, the procedure prescribes

a series of steps that users should follow to establish their interests, the

relative importance of these interests and their bottom line. It then outlines a

procedure for developing a description of these same elements from their

protagonists’ standpoints. They suggest a series of strategies for achievingthis – e.g. by building trust so that information is shared, asking questions

that are likely to reveal this information and strategically disclosing infor-

mation as a means of encouraging the protagonist to do the same.

The procedure emphasises the importance of users gathering infor-

mation to understand better their own objectives and values as well as

those of their protagonists. From this standpoint, the process of negoti-

ation is about information gathering rather than influencing. This is an

important point to note, since in negotiations it is unlikely that the peopleinvolved will change their views. Even when these views are in conflict,

however, it is often possible to find a settlement that is of mutual benefit.

Globalisation, the growth of international supply chains and the

increasing power of the web have led to much recent work on electronic

negotiations to help companies and their suppliers negotiate contracts

(Bichler et al ., 2003). Modern agent technology means that more and more

of these negotiations are being automated (Chen   et al ., 2005; Maes  et al .,

1999). Another topical application of negotiation theory is in the arena

of public participation (Kersten, 2003; Rıos Insua   et al .,  2003). Rıos and

Rıos Insua (2008) provide an application to the development of a council’s

budget.

11.7 Concluding remarks and further reading

In this chapter we have continued the discussion of strategic decision

making in the corporate strategic and general domains, but begun to

acknowledge that such decisions are usually, though not inevitably, takenby groups of DMs. We have not discussed the group context in the

operational or hands-on domains. Generally, these problems are better

structured and the potential for differing perspectives is less. Organisations

346 Decision behaviour, analysis and support

Page 28: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 28/29

have an agreed approach to such decisions, and decision support is usually 

very similar to that for a single decision maker.

As we have seen, the provision of group decision support has to address

the potential for dysfunctional group behaviour, on the one hand, and the

nebulous concept of ideal rational and democratic behaviour, on the other.

There is much more we could have said on descriptive studies of the

behaviour of groups. The classic text is that of Janis and Mann (1977) (see

also Argyris,  2001, Baron and Kerr,  2004, Brehmer, 1986, Forsyth, 2006,

Hall and Williams,   1970, Kerr and Tindale,   2004, Phillips and Phillips,

1993, and Sudweeks and Allbritton, 1996). Closely related topics on which

we were silent are the roles that members play in the group and how, with

this knowledge, one can build effective management groups (Belbin, 1991;

Naude et al ., 2000). Turning to concepts of ideal behaviour, the literatureon voting systems, group decision making, social choice and game theory 

is peppered with discoveries of paradoxes, inconsistencies and impossi-

bility results (Bacharach, 1975; Bacharach and Hurley,  1991; Dryzek and

List, 2003; French, 1985, 1986; French and Rıos Insua, 2000; Hodge and

Klima, 2005; Kelly, 1978; Luce and Raiffa, 1957; McLean and Urken, 1995;

Raiffa  et al .,  2003; Taylor, 1995, 2005; Tulloch, 1998; von Neumann and

Morgenstern, 1947). We have presented Arrow’s classic result, but, as we

have indicated, there are many other issues, such as the manipulability of systems through tactical voting and agenda rigging (Gibbard, 1973; Hodge

and Klima, 2005; Patternaik, 1978).

We have explored various approaches to group decision support that

resolved these tensions from the perspective that groups are social pro-

cesses, not entities that can decide in the same sense that individuals can.

This led us to consider facilitated workgroups, also called decision con-

ferences (Eden and Radford, 1990; Hamalainen and Leikola, 1996; Phillips,

1984). Discussions of the role and effectiveness of facilitation can be found

in Ackermann (1996), Antunes and Ho (2001), Eden and Radford (1990),

Griffith   et al . (1998), Macaulay and Alabdulkarim (2005), Ngwenyama

et al . (1996) and Seifert (2002). A current development is the growth of 

collaboration engineering , a multidisciplinary approach to designing pro-

cedures and systems to support specific types of group decisions (Briggs et al .,

2003; de Vreede, 2006).

11.8 Exercises and questions for discussion

(1).   N & T is a retail organisation of very long standing that has been

losing market share every year for each of the last ten years, in

contrast to its major competitors, which have increased their market

347 Groups of decision makers

Page 29: Chapter 11 Groups of Decision Makers

8/10/2019 Chapter 11 Groups of Decision Makers

http://slidepdf.com/reader/full/chapter-11-groups-of-decision-makers 29/29

share. Over this period the personnel on the board have remained

largely the same, have worked together very closely to try to rectify 

the situation, but have had little success. There is now some concern

that the quality of decision making of this strategic group may be

suboptimal, and they are seeking expert advice from you about

possible limitations in their group decision making and how these

might be overcome. Write a report for this group outlining some of 

the potential advantages of group decision making, some of the

factors that may inhibit the achievements of these advantages, what

can be done to improve this activity and the relevance of these for

the situation at N & T.

The report should include:

 a review of theory and research on the potential advantages anddisadvantages of group decision making, highlighting the factors

that limit the effectiveness of group decision making, and empha-

sising those that are particularly relevant to the situation at N & T;

  a critical review of the approaches designed to improve group

decision making, outlining possible strengths and weaknesses of 

each, including which aspects of the limitations described above

that each approach addresses, and the particular relevance of these

for the situation at N & T; and  some specific recommendations for the group involved.

We suggest that you structure your report in the following way.

  A brief introduction outlining the background to the report

(including why these kinds of issues are important and why they 

come about), the aims and objectives of the report and its

structure – i.e. the major sections and their purpose.

  A section reviewing the extant literature on the effectiveness/

ineffectiveness of group decision making, highlighting those

aspects likely to be particularly relevant to the group in question.

  A section reviewing ways of improving group decision making,

pointing out which aspects of poor decision making each addresses

and their strengths and weaknesses in the context of their appli-

cation to the strategic situation at N & T.

 A conclusion outlining some recommendations and limitations in

the report.

348 Decision behaviour, analysis and support


Recommended