You are on page 1of 16

Decision Analysis informs ®

Vol. 4, No. 4, December 2007, pp. 211–226


issn 1545-8490  eissn 1545-8504  07  0404  0211 doi 10.1287/deca.1070.0098
© 2007 INFORMS

e-Participation and Decision Analysis


Simon French
Manchester Business School, University of Manchester, Manchester, M15 6PB, United Kingdom,
simon.french@mbs.ac.uk
David Rios Insua
Statistics and Operations Research, Universidad Rey Juan Carlos, Móstoles, E 28933 Madrid, Spain, david.rios@urjc.es

Fabrizio Ruggeri
National Research Council, Institute of Applied Mathematics and Information Technology, I-20133 Milan, Italy,
fabrizio@mi.imati.cnr.it

D ecision analytic methods are being increasingly used to help to articulate and structure debate and delib-
erations among citizens and stakeholders in societal decisions. Methods vary, but, essentially, a public
authority or agency, when faced with a significant set of issues, may organise one or more workshops with
stakeholders and citizens as participants. Such methods of public engagement and participation are, by and
large, conducted face to face. However, the advent of the World Wide Web brings the possibility of conducting
citizen and stakeholder interactions in a distributed, possibly asynchronous fashion. In this paper we discuss
the challenges that have to be addressed and overcome if such e-participation is to be a valid tool within a
modern democracy. The difficulties are many and varied, but the pressures towards e-government, and better
regulation in general, mean that such methods will be used in the near future. Thus, we outline a program of
research and debate in which we believe that the professional decision analysis community should engage.
Key words: decision analysis; decision conference; deliberative democracy; e-democracy; e-participation;
facilitation; negotiation analysis; public participation; stakeholder workshops
History: Received on September 25, 2007. Accepted by L. Robin Keller on October 23, 2007, after 1 revision.

1. Introduction automate standard democratic instruments. Typical


In this paper we discuss a range of issues that arise examples include technologies of e-voting, which
when using Web-based interactions to support par- refer mainly to facilitating voting through electronic
ticipation and deliberative democracy, which we call means (Krimmer 2006), and technologies of e-delib-
e-participation and e-democracy, respectively. Many eration, which facilitate debate of issues through the
exploratory studies and experiments are already Web; see Davies and Noveck (2007) for a review.
underway: See the reviews in French (2003b, 2007a) Browning (2002) provides a broad introduction to the
and Rios Insua (2007) that stem from the European use of information and communication technologies
Science Foundation program Towards Electronic to support standard democratic instruments.
Democracy: Internet Based Complex Decision Sup- Recently, Gregory et al. (2005) have described the
port (TED), which has been examining the potential use of decision analysis methodologies to structure
for decision analytic methodologies to provide frame- and facilitate public and stakeholder participation in
works to support e-participation and e-democracy societal decision making and deliberative democracy
(http://www.esf.org/ted or http://infodoc.escet.urjc. in a face-to-face manner. Similar ideas are offered
es/ted). Drawing on the experience accumulated by, inter alia, Renn (1999), Sheppard and Meitner
from this program, we concentrate here on how we (2005), or Winn and Keller (2001). Hämäläinen’s
might transform democratic processes using Web- group at Helsinki has been particularly active in
based decision and negotiation analysis to structure practical explorations; see, e.g., Mustajoki et al.
and articulate participative deliberations. This is in (2004) and many other references at http://www.
contrast with the bulk of research in the field, which decisionarium.tkk.fi. Moreover, many applications of
concentrates mainly on technologies to facilitate or decision analysis to societal issues have involved
211
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
212 Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS

stakeholder workshops or other similar participatory decision-analytic process is implemented within deci-
instruments; see Keefer et al. (2004) or Renn et al. sion conferencing. In §4, we suggest what the infor-
(1995). The broad motivation for this work is that pub- mation and communication technologies architecture
lic authorities, when faced with an important set of of an e-democracy system might look like. We make
issues, should draw together a range of stakehold- no claims that it is the only possibility, but we have
ers in a workshop or a series of workshops and use tried to take a comprehensive view of the generic
decision analysis to explore the issues and formulate requirements of such a system. In §5, we develop and
policies. Essentially, the ideas build on and broaden explore our concerns related to the potential lack of
the concept of a decision conference (Goodwin and validity of e-participation systems and the research
Wright 2003). This use of decision-analytic work- that is needed, giving substance to the issues raised in
shops should be seen as part of a much wider series the context of our proposed architecture. We conclude
of moves towards public engagement and partici- that there is need for a far-reaching research program
pation in many Western democracies (Beierle and on the use of decision analysis and support within
Cayford 2002, Rowe and Frewer 2005). Here we e-participation.
explore the natural step forward of implementing
such approaches through the Web, pointing out many
important challenges and questions that need timely
2. Deliberative Democracy and
answers from the decision analysis community, to Public Participation
achieve valid and coherent e-participation processes The Athenian ideal of deliberative democracy, some-
in the near future. In summary, we ask: How do we times called substantive or direct democracy, may
implement a decision-analytic process over the Web have never really existed. Women and slaves had no
so that participants are able to input their views, right to vote, and many of the poorer men could not
beliefs, and preferences, are informed by the process, leave their work to attend meetings (Crick 2002); it is
and are able to aid in developing a recommendation estimated that, in fact, only about 15% of the popula-
on a joint choice to be made? Moreover, because of the tion was involved. Nonetheless, to many it serves as
likely applications of these tools within modern mul- a model for societal decision making, in which all cit-
ticultural societies, we ask these questions in a context izens are able to input their views and have an influ-
in which there might be many participants from var- ence on policy.
ied backgrounds and educational experience. Across many Western democracies, disillusionment
We are all aware of political imperatives towards with central government and regulatory agencies’
e-government and public participation that, together handling of issues, now termed the democratic deficit
with pressures from the information and com- (Steffek and Kissling 2007), has led to an increas-
munications industry to sell their wares, mean that ing use of consultation and stakeholder participation
e-participation and e-democracy are coming, and in public decisions. Most politicians’ motivation to
coming fast, whether or not the processes involved promote this movement towards more participation
are meaningful and valid. Thus, we outline an urgent probably relates more to gaining greater public accep-
program of research and debate within the decision tance of the ultimate decision than to promoting
analysis community to identify valid mechanisms of democratic ideals. However, whatever the cause,
e-participation. there is undoubtedly wider use of participatory meth-
The paper is structured as follows. First we pro- ods in societal decision making, from citizen juries to
vide a brief review of the current state of the art deliberative polls, going through stakeholder work-
in the field of public participation and deliberative shops or participatory budgets, as reviewed in Rowe
democracy. Then, we turn to the subject of group deci- and Frewer (2005). Several studies have covered
sion making, reflecting on the implications of Arrow’s issues such as their philosophical underpinnings, cat-
Impossibility Theoremand related results for democ- egorisation of different types of activities, and case
racy and participation. We argue that group decisions studies (Beierle and Cayford 2002, Chess and Purcell
are best seen as social processes and discuss how the 1999, Renn 1998, Rowe and Frewer 2005, Slovic 1993,
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS 213

Webler 1999). There is also a relevant literature dis- Figure 1 The Phases of Decision Analysis: After Holtzman (1989)
cussing public engagement with science and, more Decision
Recommendations
generally, societal risk communication (Bennett and model

Calman 1999, Berry 2004, Langford et al. 1999, Leach Formulate Analyse Decide
et al. 2005). As we have noted, Gregory et al. (2005)
have reviewed the use of decision-analytic methods
within public participation. There are, however, two Refine
areas in which the literature is very sparse. First,
although there have been many empirical investiga- into real-world actions. There will typically be a need
tions about the advantages and disadvantages of spe- to assess whether the model(s) have brought enough
cific participatory instruments, there are few studies understanding to make the decision: i.e., is the analy-
comparing the relative merits of two or more instru- sis requisite, as defined in Phillips (1984)? Thus, there
ments; see Rowe et al. (2005) for an exception. Sec- may be a feedback or refinement path, either after
ondly, there is a relatively small literature on how to the analysis or the decision step. A point, often over-
design participatory processes appropriate to specific looked, is that the appropriateness of any instrument
contexts; see Bayley and French (2007) for a proposal. will depend upon the phase in which it is deployed.
Even terminology in the area has yet to become For instance, some argue against open Web discus-
standard. Indeed, some might say it is somewhat sion forums because anyone can express a view, and
confusing, with different terms being used by differ- the instrument therefore fails any test of legitimacy
ent authors for the same entities and, conversely, the and fair representation. However, if the site is used in
same term being used for different entities by dif- the problem formulation phase to identify factors that
ferent authors (Rowe and Frewer 2005). We believe may have a bearing on the matter or generate poten-
it is important to distinguish a participatory process tial alternatives for implementation, representation is
from participatory instruments (or techniques). A par- less important than the variety of issues raised.
ticipatory process is the entire series of interactions Last, but not least, there are many variants of partic-
between authorities, stakeholders, and citizens from ipation. For simplicity, within our discussion we will
the initial exploration of issues of concern, up to the consider just two. In the first, the participation pro-
conclusion of the deliberations and resolution of the cess is part of a direct democracy and concludes with
matter, through to the making and implementation a binding vote between all citizens on what should be
of the decision. During the process, several participa- done. In the second, the authority and responsibility
tory instruments (e.g., stakeholder workshops or open for the decision remains firmly within a government
meetings) may be deployed to enable the participants agency. They will make the decision. However, in the
to interact. In the specific context of this paper, some process leading up to that point, they interact with cit-
(or all) of the instruments would rely heavily on the izens and stakeholders. In the former case, there is a
information and communication technologies archi- clear imperative to ensure fair representation of all cit-
tecture to support their operation. izens in the process and the highest levels of security.
Following Holtzman (1989), we consider a decision In the latter case, these constraints may be softened
process—and, hence, a participatory process—to be by the concerned agency, modifying the decision if
broadly structured as in Figure 1. The first stage is to it feels that there has been systematic underrepresen-
formulate one or more decision models that reflect the tation of some groups. We recognise, however, that
decision makers’ perceptions of the decision problem. any such pragmatic modification is hostage to many
The next stage is to analyse the decision model(s), i.e., questions of democratic ideals, trust, and legitimacy.
to explore what insights they bring about the possi-
ble choice and see where the balance of advantages 3. Group Decision Theory and
and disadvantages lies. The third stage is to decide e-Participation
upon a policy to implement. This involves interpret- We are with Gregory et al. (2005) in believing that
ing the policies recommended by the analysis phase decision-analytic methodologies provide an effective
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
214 Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS

and valuable means of articulating and structuring GDMF ac : Gather the group together in a facilitated
deliberations within public participation. Moreover, discussion of issues. Through discussion between
we agree that the schools of decision analysis stem- participants, seek to agree on group probabilities
ming from the subjective expected utility (SEU) and utilities without formally eliciting individual
model, including multiattribute value and utility ones: i.e., seek to elicit group values directly with-
models, provide the most coherent forms of decision out any intermediate step for the individual mem-
analysis with which to do this (Clemen and Reilly bers. Develop a group analysis and explore areas of
2005, French and Rios Insua 2000, Keeney and Raiffa disagreement via sensitivity and robustness analysis.
1976). However, below we do note that proponents of Seek to reach a decision by consensus without formal
other schools of decision analysis might be effectively voting (Eden and Radford 1990, French 2003a).
disenfranchised if only SEU approaches are articu- Over the years, the many paradoxes and impos-
lated within a participatory process. For the present, sibilities stemming from Arrow’s theorem have led
nonetheless, we maintain an SEU perspective. most decision analysts to doubt the efficacy of
Participation inevitably means that we are dis- GDMGSEU and GDMvote . Although one can define algo-
cussing a group decision context, and the paradoxes, rithms to move the numbers and votes around so
social traps, and impossibilities that abound in group that, ultimately, a group ranking is mathematically
decision making and democratic systems are well defined, an examination of the assumptions under-
known (Bacharach 1975, French and Rios Insua 2000, pinning the algorithms finds inconsistencies. It should
Hodge and Klima 2005, Raiffa et al. 2002). The fol- also be noted that the generality of Arrow’s and sub-
lowing modes of group decision analysis have been sequent authors’ analyses means that the difficulties
proposed as follows. apply not just to SEU approaches, but essentially to
GDMGSEU : Essentially, this assumes that the subjec- all decision-analytic methodologies.
tive expected utility model applies at the group level, Approach GDMSupraDM is more promising at first
and moreover that group probabilities and utilities are sight: all interpersonal comparisons are made within
constructed from those of its members. Thus, elicit the mind of the supra decision maker, and it is
each group member’s subjective probabilities and the issue of defining valid interpersonal compar-
utilities. Combine the individual probabilities and isons objectively that tends to cause the paradoxes
utilities into group probabilities and utilities, respec- and inconsistencies leading to the problems with
tively. Form the corresponding group expected utili- GDMGSEU and GDMvote . In some cases, the supra deci-
ties and choose according to their ranking (Bacharach sion maker actually exists: there may be an arbiter,
1975, French 1985). formally responsible for recommending a decision
GDMvote : Work with each individual and develop that balances all stakeholder perspectives. Also, in
a personal decision analysis to guide their choice. In circumstances in which a government agency has
the light of this understanding, each individual votes the legal responsibility and accountability for mak-
within the group and a group choice is made accord- ing the decision, but does want to take into account
ing to the vote. In variants of this, the numerical val- the views of citizens and stakeholders, the assump-
ues of the individuals’ expected utilities are used to tion of the GDMSupraDM approach may become plausi-
indicate strength of preference, and this information ble: the supra decision maker is the agency. In these
is incorporated into the voting (Rios and Rios Insua circumstances, we can imagine e-participation sys-
2007). tems in which individual citizens interact with a Web-
GDMSupraDM : A supra decision maker is imagined to based decision support system, leaving their personal
exist. He observes the entire elicitation and decision preferences and beliefs for the agency to be drawn
analysis process for each individual and altruistically together later into an analysis that will drive their
uses this knowledge to construct a single decision decision. However, in true democratic approaches to
analysis for the group. The choice is made according e-participation, the nonexistence of the supra deci-
to the supra decision maker’s analysis (Keeney and sion maker remains a problem: it is a fiction that
Raiffa 1976). creates a fatal flaw in this approach. He has to be
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS 215

constructed by agreement within the group, and this provide user-friendly facilitation. A third issue refers
leads to a further group decision, arguably as hard as to the availability of time and the willingness to use it
the first, and an infinite regress. Thus, most decision for a democratic good, because there is a clear under-
analysts have looked to GDMFac , developing the use lying assumption that citizens will contribute in sub-
of facilitated workshops or decision conferences in stantial ways to the deliberations. This could perhaps
which behavioural aggregation of individual perspec- be solved through participation incentives, perhaps at
tives is driven by sensitivity analysis on a putative the risk of biasing the process, or through delegation
model (Dryzek and List 2003, French 2003a, Phillips of participation to software agents if one assumes that
1984). The approach of Gregory et al. (2005) to public the latter will achieve sufficient sophistication in the
and stakeholder participation is largely driven by this foreseeable future. We mention, finally, communica-
viewpoint. tion and coordination issues. A question, still largely
Modes GDMGSEU , GDMvote , GDMSupraDM , and unanswered, is how decision analyses should be com-
GDMFac essentially assume that the group wants to municated to the general public. Moreover, no coor-
cooperate and reach a consensus. A further mode dination approaches are available for such potentially
recognises that individual citizens and stakeholders large groups. In §5, we discuss these issues in greater
may be more self-serving and wish to negotiate a detail, but first let us give more substance to the dis-
good end point for themselves. cussion by considering a possible architecture for an
GDMNeg : Bargaining, negotiation analysis, and e-participation system.
even, arbitration tools and methods are deployed to
define a process in which the group interacts and dis-
cusses a series of solutions, usually generated to con-
4. A Decision-Analytic Based
verge to a point on the Pareto boundary that corre- Architecture for e-Participation
sponds to a “deal” or “policy” that all participants Support
find acceptable (Raiffa et al. 2002). What, then, might a decision-analytic based e-partic-
Variants of GDMNeg can lean more to the algorith- ipation system look like in terms of the architecture
mic or more to a structuring of a softer facilitated of modules, databases, etc.? We offer one view here,
social process that can be run in workshops, in many recognising that other researchers might offer a dif-
ways similar to the GDMFac approach. ferent vision, albeit to provide the same functional-
We note briefly a number of issues that partici- ity. However, we believe that our conception of the
patory democracy brings to standard group decision architecture is sufficiently generic to illustrate many of
support. Some of them will be described in further the points that we wish to make. Moreover, in purely
detail below. First, we have the issue of scalability: software-engineering terms, we believe that it could
The GDM modes were initially conceived for small essentially be constructed with today’s technologies.
groups of participants, but not for the many thou- Indeed, many components already exist.
sands that one could expect in a participatory process. The architecture that we sketch is designed to sup-
GDMGSEU and GDMvote would scale better to such con- port decision-analytic based e-participation processes
texts, provided that the analyses were supported by a that fit the GDMFac perspective, but pick up elements
system. GDMSupraDM , GDMF ac , and GDMNeg could per- of GDMvote , GDMSupraDM , and GDMNeg where they
haps be implemented in a hierarchical fashion, possi- make sense. It is generic because it seeks to accom-
bly with the aid of several facilitators working with modate various participatory instruments by appro-
different subgroups. Second, there is the issue of capa- priate choice of algorithms and definition of processes
bility, because the GDM modes were designed for par- within the modules. Different participatory processes
ticipants who have some analytical inclination and may be accommodated by removing or repeating
who are willing to express their judgements numeri- some of the steps, drawing on different instruments,
cally. Note, however, that full analytical sophistication etc. Moreover, at some points different participants
is only expected of the facilitators and analysts sup- might interact with different instruments to achieve
porting the processes. The idea, therefore, would be to the same end. It would allow some participants to
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
216 Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS

delegate their participation in some interactions to alised quantitative analyses of the issues and draw
an agent, virtual or real. For example, some partici- their own conclusion on what they believe is the way
pants might interactively explore and perhaps mod- forward. Moreover, they could see their own posi-
ify a quantitative analysis, whereas others would just tion against a context of analyses conducted for other
want to discuss alternatives qualitatively in a discus- participants, enhancing their understanding of each
sion forum. Some might contribute detailed judge- other’s positions.
ments through elicitation processes, whereas others • Decide. This might be done by arbitration, negoti-
might be interested in delegating their input to a more ation, or voting, or several possible combinations and
generic agent. sequences thereof. As an example, we could propose
The architecture is illustrated in Figure 2 and articu- a process that involves a negotiation, and, if this fails,
lates a participatory process as suggested in Figure 1: undertake a voting session. If arbitration, voting, or
• Formulate. At this stage, the issues to be deliber- negotiation fails to find a mutually acceptable out-
ated would be formulated and the problem structured, come, then the process would have to iterate back
identifying uncertainties, alternatives, their interrela- through earlier stages, refining the elements of the
tions, constraints, criteria with which to evaluate con- model until a further stage finds an acceptable one.
sequences, and so on. The participants would discuss, There might also be a need to iterate or introduce
consolidate, and evolve the basic structure, aided by some further discussion and negotiation if the chosen
real or virtual facilitators to promote and enhance strategy fails to be Pareto optimal, as can sometimes
creativity. This stage would lead to a family of deci- happen in some negotiation or voting processes.
sion models capable of spanning all perspectives and The suggested architecture is illustrated in Fig-
allowing all views to be articulated within them. ure 2 with a unified modelling language diagram that
• Analyse. At this stage, the participants’ judge- reflects its key elements. The system would need elic-
ments of uncertainty and value would be elicited. itation interfaces focused on naïve users, maybe using
Through these, the participants could explore person- some of the ideas pioneered in MAUD (Humphreys

Figure 2 A Generic IT Architecture for an e-Participation System

Interface
subsystem
Log Main control
database subsystem
Agent
repository

Security manager Kernel subsystem


subsystem

Problem Arbitration Preference


structuring Explanatory
Decision- manager modelling
Census module module
making module module
database
database

Participatory
process control
module

Information Negotiation
Debate manager Voting manager
resource manager module
module module
manager module
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS 217

and McFadden 1980). Indeed, a compromise might be Noveck 2007) are very relevant, as in, e.g., tools to
sought between ease of use and understanding and support distributed synchronous and asynchronous
too much sophistication in the methods deployed. collaborative writing over the Web. However, most
There might be a need to “lock” parts of the model of this work assumes an already established, rela-
so that, say, exploration could be focused on the val- tively small and cohesive group, in contrast with our
ues of the participants with subjective probabilities context of large disperse groups. We include in rela-
locked at values set by previous discussion. In other tion with this subsystem an agent repository, which
words, there would need to be model management would provide various types of software agents on
systems such as those developed by Nunamaker et al. demand, which a participant might download to del-
(1988). egate aspects of his participation in some stages,
We now describe the key component databases should he lack time to actually participate.
and modules, and point out elements already avail- • Main Control Subsystem. This module would coor-
able that could be drawn on to implement this dinate computationally the rest of the subsystems. It
architecture. would receive user requests from the interface sub-
First, we identify three main databases: system, query the security manager subsystem to ask
• The census database would contain the data of whether the user has sufficient permissions to do it
all users allowed to participate in the process, their and, if positive, call the appropriate module within
roles, permissions to take part at various stages, etc. the kernel subsystem. Finally, when the results were
The system would check against this database when- available, it would submit them to the interface sub-
system to show them to participants. In addition, this
ever a participant seeks to undertake a given action
subsystem would periodically obtain the individual
with the system.
execution traces from the rest of the subsystems, and
• The log database would contain the complete
record them in the log database. Should the system be
execution trace of every system interaction, so as to
able to offer automatic facilitation, then this subsys-
be verified whenever a political party or an autho-
tem, together with the participatory process control
rized citizen demands it. Individual execution traces
module described below, would provide this.
of each subsystem should be constantly recorded here
• Security Manager Subsystem. Using the census and
for verification and validation purposes, as a way to
decision-making databases, this would authenticate
increase trust in and legitimacy of the system but, of
whether a user has sufficient permissions to do some-
course, risking perceptions of the loss of anonymity. thing he or she has requested and grant the corre-
• The decision-making database would contain all sponding execution permission. Standard work on
data related to each decision-making process, such computer and network security (as in Stallings 2005),
as the internal state of the process (debating, negoti- and on secure Internet voting (see Krimmer 2006 for
ating, voting, terminated, etc.); probabilities, utilities, a review), would be relevant. Note, however, that if
etc., of each participant; votes cast; messages submit- users provide their preference information in a utility
ted to debates; alternatives posted; concessions made; function, the information would require more com-
or offers accepted at negotiations, etc. plex treatment than “simple” votes.
The architecture includes four main subsystems: • Kernel Subsystem. This would be the system com-
• Interface Subsystem. This part would be used ponent in which requests to vote, negotiate, debate,
to interact with and between the users and, possi- structure a problem, communicate preferences, etc.
bly, their agents. To mitigate the “digital divide,” it would be executed, without the need to worry about
should be intuitive and easy to use. Depending on security issues, handled by the security manager sub-
the user sophistication level, we could employ differ- system. It would query and update the decision-
ent types of interfaces, as required depending on the making database and provide support for individuals
stages and tools adopted in the participatory process and groups, as described below.
at hand. Here, developments in computer-supported The kernel subsystem would be divided into sev-
collaborative work (Coakes et al. 2002, Carstensen eral modules. All of them would use the decision-
and Schmidt 1999) and online debating (Davies and making database and be directly activated by the
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
218 Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS

main control subsystem. The main modules that refremov/rgmas/pp_gecd/index.html). Some Web-
should be included are: based equivalent of the gmaa sensitivity analysis sys-
• Participatory Process Control Module. This mod- tem (http://www.dia.fi.upm.es/∼ajimenez/GMAA)
ule would be involved in creating, specifying, and should also be devised. In the related area of consumer
structuring the participatory process, by defining the preferences for purchases, Butler et al. (2008) describe
stages and, for each of them, the tools, iterations, the facilitation of e-transactions using multiattribute
time allowed, and so on. Once the participatory pro- utility theory.
cess was specified, agreed upon, and scheduled, this • Voting Manager Module. This would manage
module would be in charge of controlling its correct all voting processes at the various stages of the
execution and its state flow, either manually through participatory process. It would implement a range of
the system administrator or other user requests, or voting protocols, including some requiring utility
automatically through an algorithm. Depending on and value functions to be submitted. For example, it
the specific participatory process, it would have sev- could handle approval voting from a list of candidate
eral states, such as acquiring information, struc- options or majority voting on the current alterna-
turing the problem, debating, negotiating, voting, tives in contention in a negotiation. Internet voting
etc. In principle, this should support any decision- has attracted a great deal of attention, and many
making process, whatever the decision domain is implementations of different rules already exist.
(budget elaboration, law elaboration, etc.). Model and Some examples are Opinions-Online (www.opinions.
process management tools would be relevant here hut.fi/introduction.html); Vote-pro (http://www.
(Nunamaker et al. 1988, Papamichail and Robertson vote-pro.com/); 2ask (http://www.2ask.net/); the
2005). VoteSecure Project (http://www.votesecure.org); and
• Problem-Structuring Module. This would be used KOA (http://sort.ucd.ie/projects/ucdkoa/).
to support issue formulation and problem structur- • Negotiation Manager Module. This would manage
ing (Rosenhead and Mingers 2001, Franco and Shaw all negotiation processes taking place at various
2006) and should allow for various problem struc- stages of the participatory process. A range of nego-
tures from a simple list of alternatives, possibly with tiation methodologies and algorithms from simple
constraints, to an influence diagram or decision tree (say, negotiation by posting) to sophisticated (say,
(Clemen and Reilly 2005). Formal structuring meth- negotiation through the balanced increments method)
ods could be supported. The module might initially would be implemented; see Raiffa et al. (2002) for
be used by the problem owner and his analysts, but a review. There is an important body of work on
in later stages it might be used by all participants generic e-negotiation systems (Benyoucef and Verrons
to explore and debate their perception of the prob- 2007). Many examples of Web-based negotiation
lem, assess consequences, etc. There are some tools to systems stem from the Interneg project (http://www.
help in this, but many more are needed (Chen and interneg.org). Other systems include IBM’s SilkRoad
Lee 2003). Note that many of the soft methods used (http://www.zurich.ibm.com/csc/ebizz/oldprojects/
in problem structuring also are extremely effective in silkroad.html) and Joint Gains (http://www.jointgains.
communicating issues (French et al. 2005). Thus, in hut.fi/).
the consolidation and discussion phase, there might • Arbitration Manager Module. This would be
also be a need to call this module to support debate. invoked if arbitration is selected as the conflict reso-
• Preference-Modelling Module. This would be lution method. It should support various arbitration
used, if invoked, to model the participants’ pref- algorithms (Thomson 1994) and be able to interact
erences to be used later to find the partici- with a real arbiter as well as to generate and sug-
pants’ preferred alternatives and support conflict gest arbitrated solutions automatically. Examples of
resolution. It should facilitate various preference- Web-based arbitration systems may be seen at http://
modelling modes, from value/utility functions as in interarb.com/vl/p806396718.
the decisionarium (http://www.decisionarium.hut.fi), • Debate Manager Module. This would be used
to goal-setting methods (http://bayes.escet.urjc.es/ to handle, generate, structure, and compare user
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS 219

opinions in all phases required by the participa- compiled to facilitate wider use, together with train-
tory process. For instance, in the structuring phase ing and software manuals aimed at a wide variety
it would support the development, comparison, and of users of varying levels of sophistication. The total
modelling of different perspectives; in the negotia- complexity of the whole system would be compara-
tion phase, it would handle discussion about posted ble to a software suite such as SAS. Even allowing for
offers. Here there are many noteworthy contribu- the existence of many parts of the system in stand-
tions in the area of online debate tools, as may be alone form, making them work together in a secure,
seen in Davies and Noveck (2007), from which we robust, coherent way would be a task requiring many
could draw upon PHPbb, which is a freeware engine person-years of effort.
to implement forums (http://www.phpbb.com/). An An example of a site with many potential com-
important issue for which there is little work would ponent modules is the decisionarium at Helsinki
be to incorporate tools to detect clusters of users University of Technology (Hämäläinen 2003, www.
with similar interests or opinions, to create coali- decisionarium.tkk.fi). It is aimed at somewhat sophis-
ticated users of decision analysis tools. It is built on a
tions. A description of such a system may be seen in
functional model, where the user presses a button to
Lourenco and Costa (2007).
initiate an operation or analysis, crucially lacking pro-
• Information Resource Manager Module. This would
cess management tools to integrate the components
save and manage information resources related to a
into a well-structured participatory process. Thus,
decision-making problem, such as Internet links, book
the user needs to know which buttons to press and
references, electronic documents, previous related
in what order. For less sophisticated users, a work-
debates, sounds, e-learning material, etc. It would also
book approach such as the one used by Mathemat-
include computer-supported collaborative work tools,
ica, (Wolfram 1991), leading them through an analysis,
for example, wikis, to share information cooperatively might be more appropriate.
among several participants, such as Carstensen and Of course, rather than building a generic tool, we
Schmidt (1999). Debates generated in the debate man- could try to build specific algorithmic tools to solve
ager module could be related to the information man- particular instances of specific problems. One exam-
aged here. Therefore, the debate manager module ple is parbud (see Rios and Rios Insua 2007), aimed
would use the information resource manager module. at supporting participatory budgeting experiences.
• Explanatory Module. Because the majority of users
would not be able to interpret the output of the anal- 5. Decision Analytic e-Participation:
ysis and develop a qualitative understanding of the
Some Practical Issues
factors driving the output, the system could benefit
The previous section suggests that building a system
from an automatic explanation module. Klein (1994),
capable of supporting a range of participatory pro-
Papamichail and French (2003), Bielza et al. (2003),
cesses securely and robustly is a major task, albeit one
and Geldermann et al. (2007) provide illustrations. that seems technically achievable. However, building
These might further benefit from sensitivity analysis a system is only part of the task. If we are to create
and robustness tools as described in Rios Insua and a valid e-participation and e-democracy system, we
Ruggeri (2000). need to use it to implement and support the social
We believe the majority of the technology, algo- processes of GDMFac or GDMNeg . Although build-
rithms, and tools needed to build such a system ing the technology will be difficult, recognising and
exist. However, that does not mean it can be built addressing all the behavioural, cognitive, cultural,
instantly—the tools must be interconnected and inte- legal, political, and psychological issues that must
grated. At present, this would require a federation of be solved to establish valid e-participation will, we
complex sites involving substantial programming and believe, be much harder. Below, we catalogue some
technically complicated interfaces, whose use needs of the most relevant issues that we have discovered—
to be transparent to end users. In particular, tests, and there will be more!—using the generic informa-
case studies, and training examples also need to be tion and communication technologies architecture to
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
220 Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS

structure and illustrate our points. Unless these points to the minimum time to be given to each stage so that
are successfully addressed, the validity and meaning- all participants have both a chance to interact among
fulness of any e-participation system might be ques- themselves and to read the input of others.
tionable. We formulate the issues through research Building upon current understanding of the social
questions that could be addressed by the decision processes in group decision support related to small
analysis community within a relevant research pro- group synchronous interactions:
gram, providing pointers to possible answers. —How do such processes work in asynchronous
interactions in much larger groups?
5.1. The Web vs. a Room —What methodologies are needed to support
How do we extend our knowledge of how to support them?
group discussion of a decision analysis within face- In a Web-meeting, almost everybody has a com-
to-face environments to support asynchronous explo- puter in front of them, and therefore can conduct side
rations of quantitative decision models? analyses of the issues or use the Web to gather fur-
Meeting on the Web is not the same as meeting in ther information not available to all. In workshops,
a room. Of course, videoconferencing (Schuman 2003) it is unusual to let participants behave that inde-
or virtual environments—e.g., Second Life (http:// pendently. The emphasis is on sharing information,
secondlife.com/)—may be used to recreate some ele- models, and ideas. There is a long and extensive lit-
ments of a room, but one of the key arguments erature on managing and facilitating such processes
for e-participation is the possibility of asynchronous in computer-supported collaborative work and group
usage of the Web. Indeed, this is the first and most decision support (Ackermann and De Veerde 2001,
obvious difference between the Web and a room: Coakes et al. 2002, Morton et al. 2003). However, at
Interactions on the Web can be, and usually are, spa- the very least, the methodologies would need modifi-
tially and temporally dispersed. There is much less cation to cope with much larger groups. Thus, a key
chronological ordering of interactions in Web meet- issue, still to be studied in depth, is that there is no
ings, and many more people can “speak” at the same prima facie reason to expect that the processes, pro-
time. To some extent, each participant can explore cedures, or agendas developed for face-to-face par-
the material in his or her own chosen order, whereas ticipation workshops and other implementations of
in face-to-face meetings all participants hear and see GDMFac will translate simply in a one-to-one fashion
the same interactions in the same order. Building the into processes, procedures, or agendas for e-participa-
main control subsystem and the participatory process tion events.
control module to be sufficiently powerful, secure,
and robust to manage such asynchronous processes 5.2. The Diversity of the Population
for large numbers of participants may be a challenge How does one organise, structure, and facilitate
in itself. There are many sites that can manage hun- debate around decision models when some of the par-
dreds and thousands of two-way and group interac- ticipants share few values with others, or perhaps
tions using e-mail, chat, and discussion forums. Such even disagree on the underlying facts of the situation?
technologies could support qualitative deliberations, Decision conferences were developed to support
but we know of none that can handle a similar range groups of decision makers faced with a common
of asynchronous interactions in relation to, e.g., the set of concerns and issues. Such groups of decision
exploration of a quantitative decision model. Asyn- makers do not form randomly, but rather are already
chronous working in large groups also places a sub- working together because of some common interests
stantial burden on individual participants who truly and objectives. For instance, members of a corporate
wish to deliberate and debate with others. The more board of directors may represent many individual
participants, the more time each needs to review inter- perspectives, but they will also share many objec-
actions and assimilate arguments, points, etc., made tives, such as ensuring profitability of the firm. Even
by others. In the context of societal decision making, in stakeholder workshops run as part of a public
it is conceivable that legal requirements may arise as participation process, participants are not usually as
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS 221

diverse as the whole of society. In our experience, the same information, and a need to verify whether
their willingness to attend such events, their selection trust is to be engendered in the participants.
for participation by the authorities, and other factors How does one deal with participants who reject the
imply that they will share many objectives as well validity of the decision methodology embodied in the
as a willingness to reach consensus. All the GDM decision models?
models assume that there is a well-defined group. In Even within decision analysis, there is far from uni-
contrast, e-participation events will draw their par- versal agreement on the validity of the SEU method-
ticipants from across society and they may be self- ology. This is not merely an esoteric issue that fuels
selecting, therefore being much less likely to have academic debates, but may also relate to cultural dif-
common interests or even common understandings ferences. For instance, the differences between the
of science, economics, etc. Consider an e-participation French outranking school and the primarily Anglo-
event designed to discuss medical research fund- American school based on SEU arises, in part, because
ing. Animal rights activists are unlikely to share any of subtle cultural differences in understanding what
key objectives with representatives of pharmaceu- a decision is (process versus point of choice) and,
tical companies. In discussions on flood defences, arguably, in the understanding of intentionality (Roy
there may be a minority who do not accept current 1996, Belton and Stewart 2002). Thus, structuring
views on climate change. In short, the diversity of e-participation around a particular decision analy-
the participants’ objectives is likely to be greater in sis methodology may effectively disenfranchise those
e-participation than in face-to-face workshops and the who find it neither transparent nor, perhaps, ratio-
difficulty in exploring decision models and converg- nal. Therefore, we believe that an important research
ing to a consensus may be correspondingly larger. issue within this area is to analyze how we may aid in
How does one define interfaces that meet the their decision processes to large diverse groups, pos-
diverse skills and cultural understandings in society sibly supported by heterogeneous decision-analytic
in a fair and effective manner? methodologies. Note that this is acknowledged within
Societies are becoming more multiethnic and mul- our architecture, because we allow various modelling
ticultural. This too has serious implications. Differ- approaches within the preference modelling module.
ent cultures have very different attitudes towards
decision making (Hofstede 1994, Wright and Phillips 5.3. Communication and Understanding
1980). Such differences may occur in decision con- How does one construct effective communication,
ferences, but previous joint working experience often possibly automatically within the interface subsystem?
softens the edges. Cultural backgrounds also affect French (2007b) has argued that a key to making the
the understanding of words and images. Geograph- SEU paradigm a methodology for decision support in
ical information systems (GIS) interfaces are being large groups dispersed over the Web is the develop-
used in some explorations of e-participation (Carver ment of better means of communicating the reasoning
et al. 2001); however, Walsham and Sahay (1998) indi- and import of an analysis. Similar arguments may be
cate that even the cultural understanding of maps can made for other decision paradigms. Much of the work
affect the usefulness of GIS-based decision support. on public engagement has identified the importance
Offence is always possible between different cultures of addressing communication issues, particularly with
by the unintentional choice of words. In face-to-face respect to risk (Bennett and Calman 1999, Fischhoff
encounters, body language can help anticipate and 1995, Gigerenzer 2002). Despite an enormous research
recognise such issues as they happen; in a large Web effort, much more is understood about the problems
meeting there may be fewer safety nets. If the inter- of communication than is known about how to con-
face subsystem offers different interfaces tailored to struct effective communication (Maule 2004). True
the needs of different user groups, some more tech- communication, in the sense of building shared men-
nically sophisticated than others, say, there is still a tal models, is difficult even when it is performed face
question of whether all the interfaces fairly convey to face with all sorts of cues passing between listeners
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
222 Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS

and speakers. On the printed page, in a media inter- per se, who joins the group to structure, smooth,
view, or through a Web-based discussion, it seems and enhance the deliberative processes. He should
even more difficult. be skilled in decision-making tools, group dynam-
First, in spite of the work that has gone on over ics, psychology, cultural issues, and communication,
many years about effective presentation of data and and he would use this knowledge to support a group
analyses, see, e.g., Tufte (1997), practice is still far from in their work. Facilitation has been fairly well stud-
perfect. Within e-participation, data presentation will ied in face-to-face contexts, but is less well under-
need to be much better. The Web is essentially graph- stood in Web-based meetings (Niederman et al. 1996,
ical, yet one only has to look at a few Websites to Macauley and Alabdulkarim 2005). The facilitator’s
realise how poorly we may use its power. We need to art relates to his or her skill in selecting an effec-
improve, and not only in terms of data presentation. tive intervention to move the group’s work on in
Good decision analyses provide understanding, not a productive manner and draw on content arising
simply optimal alternatives (French and Rios Insua from the group. Shaw (2004) provides one example
2000, Phillips 1984). While most decision analysts are of e-facilitation in problem structuring. Asynchronous
adept at explaining an analysis and what it is saying involvement of thousands of participants is likely to
either at a meeting or in a report, their audiences have be particularly challenging. For instance, one cannot
typically been well educated. How decision analyses just watch the effect of a participant’s statements as
should be communicated to the general public is still they are made, but must maintain a watch for any
largely an unanswered question. effect in the hours and days following.
Secondly, words are not used just with their dictio- Apart from the lack of physical cues to help the
nary definitions. They have local and cultural mean- facilitator in Web meetings, there is a problem of scale.
ings that some participants may not understand. For Decision conferences and stakeholder workshops typ-
instance, while conducting an environmental deci-
ically have 15 to 50 participants; 100 or more partici-
sion conference, one of us (SF) had difficulty eliciting
pants would be exceptional. However, e-participation
value judgements from a representative of an envi-
could involve thousands of participants. This is sig-
ronmental group until he discovered that this person
nificant because the potential for misunderstandings
objected to using the word “scoring” to refer to elici-
in participant interactions rises as some power of their
tation, because of connotations of success (scoring in a
number. Moreover, it is unlikely that there would be
game) when the environmental impacts being consid-
funding to maintain a ratio of facilitators to partici-
ered were usually detrimental. In face-to-face discus-
pants in the order of 1 to 30 or so. Inevitably, one will
sions, particularly facilitated discussions, the meaning
probably be working with teams of facilitators, which
of words is negotiated. How that is achieved on the
brings its own issues of coordination and coherence
Web is far from clear, particularly when the partici-
of approach.
pants are drawn from a wide, politically diverse, mul-
One possibility is to develop artificial intelligence
ticultural society.
or agent technologies to facilitate the process auto-
5.4. Facilitation matically. However, even recent work in collaboration
How does one facilitate large-scale, asynchronous engineering on describing interventions as thinklets
Web deliberations, both in qualitative discussion and (see Kolfschoten et al. 2006) has not addressed the
in exploring and learning from one or more decision sophistication of facilitation needed for issues as com-
models? plex as found in societal decisions. Moreover, facili-
In a very real sense, many of the points made tation requires much tacit knowledge that cannot be
in §§5.2 and 5.3 relate to the need for facilitators easily articulated in software (Polyani 1962, Walsham
in a decision conference (Ackermann 1996, Macauley 2001). Therefore, to develop automated facilitation
and Alabdulkarim 2005, Phillips and Phillips 1993). and encode it within the interface and control subsys-
A facilitator is someone with no responsibility or tems would require both a major research program
accountability for the consequences of the decision and a substantial software-engineering task.
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS 223

5.5. Legitimacy and Trust in some neutral way, discarding irrelevant details
How does one engender trust in e-participation sys- to ensure that the decision analysis is tractable. To
tems, including a great extent, they fix the deliberation rules, but
—in the process, that leaves potential for mistrust. Politicians would
—in the software, also have to accept that such systems could lead to
—in the facilitation and analyst team, and changes in political power structures. That may be
—in the methodology and its simplifications? a step too far for some. However, like citizens, they
Let us suppose that all the issues above are success- have to trust the processes and systems.
fully addressed and one can create an e-participation How does one ensure that e-participation processes
process to the satisfaction of the facilitators and ana- are legitimate and perceived to be so?
lysts. Would the public trust it? Would they perceive There is also the issue of legitimacy. Even if society
it as legitimate? There are several levels of trust that trusts an e-participation process to reflect the delib-
we should consider. Would they believe that all the erations of the participants, it does not necessarily
interactions on the Website are genuine? Would they follow that it will see it as a legitimate reflection of
accept the analyses on the Website as reflecting a set the whole of society. This relates to issues of the dig-
of beliefs and preferences or would they perceive the ital divide (Browning 2002): Does every citizen have
system as distorting or oversimplifying their views? equal access to the Web? We should note, however,
In a workshop, participants see and hear all the inter- that such issues exist for any participation process.
actions and can count shows of hands themselves; Many face-to-face interaction processes disadvantage
what would give them the same assurance over the rural communities because they are generally held
Web? Any democratic system is subject to some risk in cities. They also impose costs such as the need
of manipulability (Gibbard 1973, Hodge and Klima to get time off from work, or travel costs. Therefore,
2005). Will citizens trust other citizens to reveal their any participatory process imposes costs and requires
preferences honestly in the manner that a decision overcoming barriers. There are also issues relating to
analytic approach would assume? Face-to-face inter- “fair” representation and the potential for a pressure
action provides many cues to support trust; the Web group to hijack an e-participation process. Even if the
does not provide as many. e-participation is used to inform an agency’s deci-
sion rather than to determine the decision itself, the
There is also a trust issue in the sense of trust-
agency must assess whether the deliberations reflect
ing the system to implement an algorithm correctly.
the views across society. If the agency gives incentives
Even in the relatively simple area of counting votes,
to promote active engagement, will these further dis-
debates in the literature of e-voting suggest that guar-
tort the representativeness of the participants? This
anteeing that the system transparently adds up the
is particularly important because participatory pro-
votes correctly is a nontrivial task. How much greater
cesses may shift the focus of deliberation from issues
is the challenge with systems that solve influence
of science to issues of values: in itself, a move for
diagrams with complex probability and utility struc-
the better (Fischhoff 1995), but a move that places
tures? What quality control is needed to ensure that
further demands on the representativeness of those
the results are trusted? Would the code need to be
participating.
published, even when the models rely on copyrighted
We partly acknowledge the above problems by
or patented decision-analytic software? Open source
including the log database and the security manager
code may well be needed, but it takes time and effort
subsystem.
to write and assure quality. In the previous section, we
noted that the scale of an e-participation system is as 5.6. Designing the Process Itself
great and as complex as some of the largest software How does one design a participation process that is
systems developed to date with many person-years of either entirely Web based or blends Web and face-to-
effort. face interactions?
Would the facilitation team be trusted by citizens We noted earlier the paucity of advice on the
and politicians? They need to steer the deliberation design of participatory processes. This remark applies
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
224 Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS

even more to the design of e-participation processes. so that society will use e-participation wisely, possibly
Moreover, we have been writing as if a participa- through the type of architecture we have proposed.
tion process would be Web based. It is likely, how-
ever, to be a hybrid or blended process arranging a Acknowledgments
Much of the thinking in this paper grew from workshops
number of Web-based and face-to-face participation
and discussions within the tedproject: The authors are
instruments throughout the deliberation process indi- grateful to the European Science Foundation for funding
cated in Figure 1. We have noted that there have this, and to their many friends and colleagues who took
been few comparative studies on the effectiveness part. Other relevant funding has been provided by the
of different instruments. There have been virtually UK Research Councils, the Spanish Ministry of Science
and Education, and the government of Madrid, through
no studies investigating interaction effects between
the e-Democracia-CM program. Comments on earlier drafts
different participation instruments. Does running a from Greg Kersten, the editors of Decision Analysis, and
Web discussion before a stakeholder workshop elicit three anonymous referees have been particularly helpful.
a broader, more embracing range of issues to be con- The latest version of the paper was prepared while the
sidered in the deliberations? Once such studies have second and third authors were visiting the Statistical and
Applied Mathematical Sciences Institute, NC.
been undertaken, how does one use the knowledge
obtained to design an effective process? Bayley and
French (2007) have discussed the issue, offering a References
resource allocation model to help structure the design. Ackermann, F. 1996. Participants’ perceptions on the role of facil-
itators using group decision support systems. Group Decision
Once we have designed the process, we would define Negotiation 5 93–112.
it through the participatory process control module. Ackermann, F., G.-J. De Veerde, eds. 2001. Special issue: European
research on group decision support systems. Group Decision
Negotiation 10 1–94.
6. Conclusions and Recommendations Bacharach, M. 1975. Group decisions in the face of differences of
opinion. Management Sci. 22 182–191.
We are concerned that political and commercial
Bayley, C., S. French. 2007. Designing a participatory process for
imperatives towards the adoption of e-participation stakeholder involvement in a societal decision. Group Decision
might lead to their use before we understand what Negotiation. Forthcoming.
Beierle, T., J. Cayford. 2002. Democracy in practice: Public partic-
they are actually achieving. Thus, our aim here has
ipation in environmental decisions. Resources for the Future,
been to bring a number of research questions to the Washington, D.C.
attention of the decision analysis community. These Belton, V., T. J. Stewart. 2002. Multiple Criteria Decision Analysis: An
Integrated Approach. Kluwer Academic Press, Boston.
questions should be addressed before the validity
Bennett, P. G., K. C. Calman, eds. 1999. Risk Communication and
and legitimacy of decision analytic e-participation Public Health: Policy Science and Participation. Oxford University
can be assessed. There is a need to develop tools Press, Oxford.
and methodologies: to explore, formulate, and model Benyoucef, M., M. H. Verrons. 2007. Configurable e-negotiation sys-
tems for large scale and transparent decision making. Group
issues; to allow unsophisticated users to conduct, or Decision Negotiation. Forthcoming.
at least explore, decision analyses; to discover and Berry, D. C. 2004. Risk Communication and Health Psychology. Open
build consensus. We need to understand how to facil- University Press, Maidenhead UK.
Bielza, C., J. A. Fernández del Pozo, P. Lucas. 2003. Finding and
itate large, diverse, dispersed, multicultural groups, explaining optimal treatments. Lecture Notes in Computer Sci-
many of whom may share few common values, may ence, Vol. 2780. Springer Verlag, Berlin, 299–303.
be interested in different decision methodologies, and Browning, G. 2002. Electronic Democracy: Using the Internet to Trans-
form American Politics. Information Today, Medford NJ.
may prefer nonquantitative methods. Without such Butler, J. C., J. S. Dyer, J. Jia, K. Tomak. 2008. Enabling e-transac-
developments and the answers to many lesser ques- tions with multiattribute utility theory. Eur. J. Oper. Res. 186(2)
tions raised above—and, doubtless, many more ques- Forthcoming.
Carstensen, P., K. Schmidt. 1999. Computer supported collaborative
tions that we have missed, it is unlikely that any work: New challenges to system design. K. Itoh, ed. Handbook
deliberative e-democracy or e-participation process of Human Factors/Ergonomics. Asakura, Tokyo, 619–636.
can be deemed valid and worthy of trust. It behooves Carver, S., A. Evans, R. Kingston, I. Turton. 2001. Public participa-
tion, GIS, and cyberdemocracy: Evaluating online spatial deci-
the decision analysis community to lead the debate sion support systems. Environment Planning B: Planning Design
and conduct the research to address these questions, 28 907–921.
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS 225

Chen, J. Q., S. M. Lee. 2003. An exploratory cognitive DSS for strate- Hofstede, G. 1994. Management scientists are human. Management
gic decision making. Decision Support Systems 36 147–160. Sci. 40 4–13.
Chess, C., K. Purcell. 1999. Public participation and the environ- Holtzman, S. 1989. Intelligent Decision Systems. Addison-Welsey,
ment: Do we know what works? Environ. Sci. Tech. 33(16) Reading, MA.
2685–2692. Humphreys, P. C., W. McFadden. 1980. Experiences with MAUD:
Clemen, R. T., T. Reilly. 2005. Making Hard Decisions. Thomson Aiding decision structuring versus bootstrapping the decision
Learning, Duxbury, Pacific Grove, CA. maker. Acta Psychologica 45 51–69.
Coakes, E., D. Willis, S. Clarke, eds. 2002. Knowledge Management Keefer, D. L., C. W. Kirkwood, J. L. Corner. 2004. Perspective on
in the SocioTechnical World. Computer Supported Co-Operative decision analysis applications, 1990–2001 with discussion. Deci-
Work, Springer Verlag, London. sion Anal. 1 4–34.
Crick, B. 2002. Democracy: A Very Short Introduction. Oxford Univer- Keeney, R. L., H. Raiffa. 1976. Decisions with Multiple Objectives: Pref-
sity Press, Oxford. erences and Value Trade-Offs. John Wiley and Sons, New York.
Davies, T., B. S. Noveck. 2007. Online Deliberation: Design, Research, Klein, D. A. 1994. Decision-Analytic Intelligent Systems: Automated
and Practice. CSLI Publications/University of Chicago Press. Explanation and Knowledge Acquisition. Lawrence Erlbaum Asso-
Dryzek, J. S., C. List. 2003. Social choice theory and deliberative ciates, Hillsdale, NJ.
democracy: A reconciliation. British J. Political Sci. 33 1–28. Kolfschoten, G. L., R. O. Briggs, G. J. de Vreede, P. M. Jacobs,
Eden, C., J. Radford, eds. 1990. Tackling Strategic Problems: The Role J. H. Appelman. 2006. A conceptual foundation of the thinkLet
of Group Decision Support. Sage, London. concept for collaboration engineering. Internat. J. Man-Machine
Stud. 647 611–621.
Fischhoff, B. 1995. Risk perception and communication unplugged:
Twenty years of process. Risk Anal. 15 137–145. Krimmer, R. 2006. Electronic voting 2006. 2nd International Work-
shop, GI Lecture Notes in Informatics. Castle Hofen, Brĕgenz,
Franco, A., D. Shaw, eds. 2006. Special issue on: Problem structur- Austria.
ing methods. J. Oper. Res. Soc. 57 757–878.
Langford, I., C. Marris, T. O’Riordan. 1999. Public reactions to
French, S. 1985. Group consensus probability distributions: A crit- risk: Social structures, images of science and the role of trust.
ical survey. J. M. Bernardo, M. H. DeGroot, D. V. Lindley, P. G. Bennett, K. C. Calman, eds. Risk Communication and Public
A. F. M. Smith, eds. Bayesian Statistics 2. North-Holland, Ams- Health: Policy, Science and Participation. Oxford University Press,
terdam, 183–201. Oxford, 33–50.
French, S. 2003a. Modelling, making inferences and making deci- Leach, M., I. Scoones, B. Wynne, eds. 2005. Science and Citizens. Zed
sions: The roles of sensitivity analysis. TOP 11 229–252. Books, London.
French, S., ed. 2003b. Special issue: The challenges in extending Lourenco, R., J. P. Costa. 2007. Incorporating citizens’ views in local
the MCDA paradigm to e-democracy. J. Multi-Criteria Decision policy decision making. Decision Support Systems 43 1499–1511.
Anal. 12 61–233.
Macauley, L., A. Alabdulkarim. 2005. Facilitation of e-meetings:
French, S., ed. 2007a. Special issue: Interface issues in e-democracy. State-of-the-art review. IEEE Internat. Conf. e-Technology, e-Com-
Internat. J. Tech. Policy Management 7 105–208. merce and e-Service, Hong Kong.
French, S. 2007b. Web-enabled strategic GDSS, e-democracy and Maule, A. J. 2004. Translating risk management knowledge: The
Arrow’s theorem: A Bayesian perspective. Decision Support Sys- lessons to be learned from research on the perception and com-
tems 43 1476–1484. munication of risk. Risk Management 6 15–27.
French, S., D. Rios Insua. 2000. Statistical Decision Theory. Arnold, Morton, A., F. Ackermann, V. Belton. 2003. Technology-driven and
London. model-driven approaches to group decision support: Focus,
French, S., A. J. Maule, G. Mythen. 2005. Soft modelling in risk research philosophy, and key concepts. Eur. J. Inform. Systems
communication and management: Examples in handling food 12 110–126.
risk. J. Oper. Res. Soc. 56 879–888. Mustajoki, J., R. P. Hämäläinen, M. Marttunen. 2004. Participatory
Geldermann, J., V. Bertsch, M. Treitz, S. French, K. N. Papamichail, multi-criteria decision analysis with Web-Hipre: A case of lake
R. P. Hämäläinen. 2007. Multi-criteria decision support and regulation policy. Environ. Model. Software 19 537–547.
evaluation of strategies for nuclear remediation management. Niederman, F., C. M. Beise, P. M. Beranek. 1996. Issues and concerns
OMEGA—The Internat. J. Management Sci. Forthcoming. about computer-supported meetings: The facilitator’s perspec-
Gibbard, A. 1973. Manipulation of voting schemes: A general result. tive. MIS Quart. 20 1–22.
Econometrica 41 587–601. Nunamaker, J. F., L. M. Applegate, B. R. Konsynski. 1988.
Gigerenzer, G. 2002. Reckoning with Risk: Learning to Live with Uncer- Computer-aided deliberation: Model management and group
tainty. Penguin Books, Harmondsworth, UK. decision support. Oper. Res. 36 826–848.
Goodwin, P., G. Wright. 2003. Decision Analysis for Management Papamichail, K. N., S. French. 2003. Explaining and justifying the
Judgement. John Wiley and Sons, Chichester, UK. advice of a decision support system: A natural language gen-
Gregory, R. S., B. Fischhoff, T. McDaniels. 2005. Acceptable input: eration approach. Expert Systems Appl. 24 35–48.
Using decision analysis to guide public policy deliberations. Papamichail, K. N., I. Robertson. 2005. Integrating decision making
Decision Anal. 2 4–16. and regulation in the management control process. Omega 33
Hämäläinen, R. P. 2003. Decisionarium—Aiding decisions, negoti- 319–332.
ating and collecting opinions. J. Multi-Criteria Decision Anal. 12 Phillips, L. D. 1984. A theory of requisite decision models. Acta
101–110. Psychologica 56 29–48.
Hodge, J. K., R. E. Klima. 2005. The Mathematics of Voting and Elec- Phillips, L. D., M. C. Phillips. 1993. Facilitated work groups—The-
tions: A Hands-On Approach. American Mathematical Society, ory and practice. J. Oper. Res. Soc. 44 533–549.
Providence, RI. Polyani, M. 1962. Personal Knowledge. Anchor Day Books, New York.
French, Rios Insua, and Ruggeri: e-Participation and Decision Analysis
226 Decision Analysis 4(4), pp. 211–226, © 2007 INFORMS

Raiffa, H., J. Richardson, D. Metcalfe. 2002. Negotiation Analysis: The Sheppard, S. R. J., M. Meitner. 2005. Using multi-criteria analy-
Science and Art of Collaborative Decision Making. Harvard Uni- sis and visualisation for sustainable forest management plan-
versity Press, Cambridge, MA. ning with stakeholder groups. Forest Ecology Management 207
Renn, O. 1998. The role of risk communication and public dialogue 171–187.
for improving risk management. Risk, Decision Policy 3 3–50. Slovic, P. 1993. Perceived risk, trust and democracy. Risk Anal. 13
Renn, O. 1999. A model for an analytic-deliberative process in risk 675–682.
management. Environ. Sci. Tech. 33 3049–3055. Stallings, W. 2005. Cryptography and Network Security. Prentice Hall,
Renn, O., T. Webler, P. Wiedermann, eds. 1995. Fairness and Compe- Englewood Cliffs, NJ.
tence in Citizen Participation: Evaluating Models and Environmen- Steffek, J., C. Kissling. 2007. Civil Society Participation in European and
tal Discourse. Kluwer, Dordrecht, Holland. Global Governance: A Cure for the Democratic Deficit? Palgrave
Rios, J., D. Rios Insua. 2007. A framework for participatory budget MacMillan, New York.
elaboration support. J. Oper. Res. Soc. Forthcoming.
Thomson, W. 1994. Cooperative models of bargaining. R. Aumann,
Rios Insua, D. 2007. Special issue: eDemocracy. Group Decision Nego- S. Hart, eds. Handbook of Game Theory, Vol. 2. North Holland,
tiation. Forthcoming. Amsterdam, 1238–1277.
Rios Insua, D., F. Ruggeri. 2000. Robust Bayesian Analysis. Springer, Tufte, E. R. 1997. Visual Explanations. Graphics Press, CT.
New York,
Walsham, G. 2001. Knowledge management: The benefits and lim-
Rosenhead, J., J. Mingers, eds. 2001. Rational Analysis for a Problem-
itations of computer systems. Eur. Management J. 19 599–608.
atic World Revisited. John Wiley and Sons, Chichester, UK.
Walsham, G., S. Sahay. 1998. GIS for district-level administration in
Rowe, G., L. J. Frewer. 2005. A typology of public engagement
India: Problems and opportunities. Management Inform. Systems
methods. Sci. Tech. Human Values 30 251–290.
Quart. 23 39–66.
Rowe, G., T. Horlick-Jones, J. Walls, N. Pidgeon. 2005. Difficulties
in evaluating public engagement initiatives: Reflections on an Webler, T. 1999. The craft and theory of public participation:
evaluation of the UK GM Nation? Public debate. Public Under- A dialectical process. J. Risk Res. 2 55–71.
standing Sci. 14 331–352. Winn, M. I., L. R. Keller. 2001. A modelling methodology for
Roy, B. 1996. Multi-Criteria Modelling for Decision Aiding. Kluwer multi-objective multi-stakeholder decisions: Implications for
Academic Publishers, Dordrecht, Germany. research. J. Management Inquiry 102 166–181.
Schuman, S. 2003. Information technology for groups. Group Facili- Wolfram, S. 1991. Mathematica: A System for Doing Mathematics by
tation (Spring). http://findarticles.com/p/articles/mi_qa3954/ Computer. Addison Wesley, Redwood City, CA.
is_200304. Wright, G., L. D. Phillips. 1980. Cultural variation in probabilistic
Shaw, D. 2004. Creativity and learning through electronic group thinking: Alternative ways of dealing with uncertainty. Inter-
causal mapping. Internat. J. Innovation Learn. 14 364–377. nat. J. Psych. 15 239–257.

You might also like