RISK MANAGEMENT IN A DEVELOPING COUNTRY CONTEXT: STRUCTURED
DECISION MAKING FOR POINT-OF-USE WATER TREATMENT IN RURAL
TANZANIA
By

Kristianna Francesca Post

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Community, Agriculture, Recreation and Resource Studies
2010

ABSTRACT
RISK MANAGEMENT IN A DEVELOPING COUNTRY CONTEXT: STRUCTURED
DECISION MAKING FOR POINT-OF-USE WATER TREATMENT IN RURAL
TANZANIA
By

Kristianna Francesca Post

Research and practice in international development focuses on reducing risks and
improving the quality of life for people living in developing regions of the world. In
pursuit of these goals, development practitioners have confronted a number of
challenges including mistrust of outsiders, language and cultural barriers, and low levels
of education and literacy. As a result, many providers of support and aid to developing
countries have exported expert-driven decision support processes that have largely
failed to accurately capture the full spectrum of objectives and concerns that are of
importance and relevance to local stakeholders. With this as a backdrop, this thesis
reports the results of research conducted in East Africa. With support from the National
Science Foundation (SES 0924210), a series of interactive workshops were conducted
in Tanzania in 2010. The purpose was to help local villagers to identify and select
effective and culturally appropriate water purification systems for use at the household
level. To do so, we developed a decision support framework that merged concepts
from good practice in structured decision making, risk communication, public health,
water quality testing, and - as a matter of necessity - popular television cooking shows.

Acknowledgments

I would like to thank my advisor, Dr. Joseph Arvai, and my committee members, Dr.
Robert Richardson, Dr. John Kerr, and Dr. Maria Lapinski, for their guidance and
support throughout this process. I would also like to thank my fellow peers in the
Skunkworks Lab for their advice and comic relief, particularly Delanie Kellon.

Finally, I would like to thank Diane Ruonavaara, Patricia Peek, Ana Kelly, and the Water
Quality, Environmental and Molecular Microbiology Laboratory at Michigan State
University for their assistance, as well as the National Science Foundation, the
department of Community, Agriculture, Recreation and Resource Studies at Michigan
State University and the Graduate School at Michigan State University for supplying the
funding to carry out this research.

	
  

iii	
  

PREFACE

Research and practice in international development focuses on reducing risks and
improving the quality of life for people living in developing regions of the world. Much of
this work encompasses the closely related goals of poverty reduction, safeguarding
human health and natural resources, providing basic education, and encouraging social
justice and equality. But in pursuit of these goals, development practitioners have had
to confront a number of challenges.

Some of these challenges, which admittedly are not unique to international development
contexts, involve helping local people to: (1) recognize and understand the nature and
magnitude of the risks they face; (2) identify and characterize situation-specific
objectives intended to guide risk management decisions; and (3) become meaningfully
involved in the design, evaluation, and selection of a preferred risk management option.

Other challenges, however, are rather unique to development-specific contexts and
include a deep mistrust of outsiders, language and cultural barriers, and low levels of
education and literacy. In confronting these challenges, many providers of support and
aid to developing countries have simply exported expert-driven decision support
processes that, in our experience, have largely failed to accurately capture the full
spectrum of objectives and concerns that are of importance and relevance to local
stakeholders.

	
  

iv	
  

With this as backdrop, this thesis will report the results of research conducted in East
Africa. With support from the National Science Foundation (SES 0924210), a series of
interactive and interdisciplinary workshops were conducted in Tanzania in 2010. The
purpose of these workshops was to help local villagers to identify and select effective
and, importantly, culturally appropriate water purification systems for use at the
individual and household level. To do so, we developed a decision support framework
that merged concepts from good practice in structured decision making, risk
communication, public health, water quality testing, and - as a matter of necessity popular television cooking shows.
*

*

*

*

*

*

*

This thesis is structured around three chapters. The first chapter describes the
principles and methods that support decision making in international development
contexts. It focuses primarily on the approaches used to improve a project’s relevance
to targeted communities, and briefly summarizes the phases in development history that
set the foundation for locally sensitive development decision making.
The second chapter describes the theoretical framework behind the research reported
in the third chapter of this thesis. Specifically, it provides an overview of the history of
behavioral decision research, focusing on both normative and descriptive theories of
decision making.

It then discusses the frequently used decision aiding processes

known as Structured Decision Making (SDM). The third chapter describes the research
process in Tanzania and presents research results.

	
  

v	
  

Table of Contents
	
  
List of Tables .............................................................................................................. viii
List of Figures .............................................................................................................. ix
	
  
Chapter One: Supporting International Development Decisions: ......................... 1
Principals and Methods
1.0 Introduction............................................................................................................. 1
1.1 Guiding Principals of Development Decision-making............................................. 3
1.1.1 Principals for Including Local Stakeholders......................................................... 3
1.1.2 Principals for the selection of methods used in planning and decision processes
...................................................................................................................................... 4
1.1.3 Principals for facilitation....................................................................................... 5
1.2 Common Methods for Supporting Development Decision-Making......................... 6
1.2.1 Rapid Rural Appraisal (RRA)............................................................................... 6
1.2.2 Participatory Rural Appraisal ............................................................................... 9
1.2.3 Beneficiary Assessment .................................................................................... 12
1.2.4 Stakeholder Analysis ......................................................................................... 15
1.2.5 Stated Choice Methods ..................................................................................... 17
1.2.6 Holistic Management ......................................................................................... 19
1.3 Conclusion: Future Research to Improve Participatory Decision Making Practices
.................................................................................................................................... 21
Chapter Two: Behavioral Decision Research ........................................................ 24
2.0 Introduction........................................................................................................... 24
2.2 Descriptive Theories of Decision Making ............................................................. 29
2.2.1 Prospect Theory ................................................................................................ 29
2.2.2 Preference Construction.................................................................................... 31
2.2.3 Bounded Rationality and Satisficing .................................................................. 32
2.2.4 Representativeness........................................................................................... 33
2.2.5 Availability.......................................................................................................... 33
2.2.6 Anchoring and Adjustment ................................................................................ 34
2.2.7 Affect ................................................................................................................. 35
2.3 Structured Decision Making (SDM) ...................................................................... 37
2.4 Case Studies ........................................................................................................ 43
2.4.1. The Alouette River Stakeholder Committee ..................................................... 43
2.4.2 Integration of Biological and Sociological Needs in Northwestern Montana ..... 44
2.4.3 Mining in East Malaysia..................................................................................... 45
2.4.5. Experimental Tests of SDM Approaches.......................................................... 45
2.5 SDM in developing countries................................................................................ 46
	
  
	
  
	
  

	
  

vi	
  

	
  
	
  
	
  
Chapter Three: Risk management in a developing country context: Structured
decision making for point-of-use water treatment in rural Tanzania .................. 49
3.0 Introduction........................................................................................................... 49
3.1 Methods................................................................................................................ 55
3.1.1 Context: POU Water Treatment ........................................................................ 55
3.1.2 Study Locations and Participants ...................................................................... 58
3.2 Structured Decision Making.................................................................................. 60
3.4 Water Quality........................................................................................................ 66
3.4 Results.................................................................................................................. 67
3.4.1 Milola ................................................................................................................. 67
3.4.2 Naitolia............................................................................................................... 72
3.5 Discussion ............................................................................................................ 76
References ................................................................................................................ 82
	
  
	
  
	
  

	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  

	
  

vii	
  

	
  

List of Tables

Table 1: Milola Consequence Matrix………………………………………………………. 68
Table 2: Naitolia Consequence Matrix.….………………………………………………… 71

	
  

viii	
  

	
  

List of Figures

Figure 1: The evaluation of water samples by workshop participants in Naitolia……...64

Figure 2: Examples of objective sketches for POU methods........................................67

	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  
	
  

	
  

ix	
  

	
  
	
  
	
  
Chapter One
Supporting International Development Decisions:
Principals and Methods

1.0 Introduction
Decision making in international development has a dynamic history, encompassing
multiple phases. These phases have each been influenced by shifting perspectives
regarding how best to approach development activities. The earlier years of
development, roughly in the 1940s through the early 1960s, were characterized by a
belief that only through the transfer of Western scientific knowledge and technical
innovations could other less developed areas of the world develop. Many historians
pinpoint U.S. President Harry Truman’s 1949 inaugural address as the prominent call to
begin this stage of development (Lewellen 2003). At that time, improved production and
the transfer of capital were viewed as the solution for lesser developed countries (LCD)
to reach both peace and prosperity (Escobar 1995; Stiglitz 1989); thus, it was the hope
that strategies taken in the West that led to high rates of production could be a cure all
solution for other areas of the world. This resulted in strictly exogenous approaches to
development (Terluin 2003; Van Tatenhove & Leroy 2003) that exercised decision
making from the top down and integrated only scientific - or what was considered
‘expert’ - knowledge to be incorporated into development planning.

	
  

1	
  

In the 1970s, the first efforts to incorporate local perspectives into development planning
occurred (Reed 2008), later exemplified by a strong increase in the development of
methods enabling development researchers and practitioners to elicit local knowledge
and preferences for various development projects in the 1980s (Chambers 1981,
1994a; Ellis & Biggs 2001).

In the 1990s, development decision making took yet

another shift: This new discourse began calling for locals to serve as equitable partners
in the development process, from the research conducted to support decision making to
the decision making process itself (Chambers 1994a; Cooke & Kothari 2001; Ellis &
Biggs 2001; Kapoor 2002; Reed 2008). Those targeted for development were no longer
the last to be acknowledged in the development process; rather, they were steadily
moved forward in an effort to put the last first (Chambers 1997)1.

Today, the integration of multiple sources of knowledge and active partnership in
development initiative is a common occurrence. Including the perspectives of local
people has worked to enhance the relevancy of development projects to the people they
are targeting and make the overall process a more equitable approach. In fact, cases
exist where community members have rejected development projects offered to them if
they were not incorporated into the decision-making process (Guggenheim 2006).

Yet, the increase of multi-stakeholder involvement has come with a cost: the decision
making process itself has become more complicated. Development practitioners are
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
1

For a more thorough review of the phases in development
history see (Hickey & Mohan 2004) For an in-depth study on
competing development theories, see (Martinussen 1997).
	
  

2	
  

now pressured to include a wider variety of stakeholders in planning and are expected
to know how to integrate them into a fluid process.

Practitioners must also be able to

effectively elicit local knowledge, which may be difficult due to cultural and language
differences. For example, terminology used often in one culture may not be used at all
in another, or may take on different meanings. Cultural differences in social power
relations and gender may influence planning and decision processes. There is also
pressure to design and carry out this more complicated field research in a manner that
is both time and cost efficient.

To overcome these complications, research related to international development has
also focused on identifying guiding principals and methods to support integrated
stakeholder participation in development planning and decision-making.

These are

outlined in the following sections.

1.1 Guiding Principals of Development Decision-making
1.1.1 Principals for Including Local Stakeholders
Literature related to development planning asserts that those living in communities
selected for projects should be active participants in the project decision making
processes (Agrawal & Gibson 1999; Bradlow 2001; Chambers 1997; Grimble et al.
1995; Leeuwis 2000; Reed 2008; Rietbergen-McCraken & Narayan 1998). Some of the
proposed benefits of community member involvement include: increased reliability of
data (Agrawal & Gibson 1999); a larger number of solutions identified and increased
trust (Bradlow 2001; Richards et al. 2004); more accountability in projects (Ribot 2002);
	
  

3	
  

enhanced efficiency of development investments (Cleaver 2001); and empowerment for
local communities (Arnstein 1969; Bradlow 2001; Chambers 1994a; RietbergenMcCraken & Narayan 1998; Tippett et al. 2007).

The first step in this process is identifying the most relevant stakeholder groups to take
part. Stakeholder analysis is one of the most widely used methods to accomplish this
task (see section 1.2.4), although Grimble (1995) recommends utilizing multiple
stakeholder identification methodologies to ensure that all relevant groups are
considered.

Stakeholder groups most relevant for involvement also depend on the

degree to which they will participate. Scholars, most notably Arnstein (1969), have
categorized different levels of potential stakeholder involvement to help guide
practitioners in this area. Once selected, it is important that groups are integrated at the
earliest stage possible to ensure they have sufficient input on identifying potential
solutions to the development problems targeted (Chess & Purcell 1999; Grimble et al.
1995; Reed 2008; Tippett et al. 2007).

1.1.2 Principals for the selection of methods used in planning and decision
processes
As the introduction to this chapter implies, development decision making and planning
processes often incorporate more than just technical knowledge. In response, multiple
methods have been developed to help understand and elicit local knowledge. Deciding
which method is most appropriate can be a difficult task. One of the first method
selection principals to follow is to ensure that those chosen account for the sociocultural

	
  

4	
  

backgrounds of involved stakeholders (Bradlow 2001; Chambers 1994a, 1997). For
example, Chambers (1997) is widely regarded for promoting the use of visual tools in
order to ensure that illiterate stakeholders can equitably and effectively participate in
development planning. Methods that do not consider sociocultural factors are more
likely to produce inaccurate results due to insufficient elicitation and integration of local
knowledge.

Methods chosen may also have to be correlated with education measures to better
inform participants of the decision context if it is unfamiliar (Reed 2008).

Bradlow

(2001) emphasizes this by asserting that unless people have sufficient information on
the project, those facilitating the process cannot say with any confidence how
participants will respond when their chosen decision outcome occurs in real life. Finally,
methods cannot be chosen until the development objective is clearly identified, as this
definition shapes the overall planning process (Leeuwis 2000; Reed 2008; RietbergenMcCraken & Narayan 1998).

1.1.3 Principals for facilitation
Having trained and skilled facilitators is another strongly emphasized principal of multistakeholder development planning processes. Often, outcomes of these procedures are
more sensitive to the skill of the facilitator than they are to the tools employed to
conduct the process (Chess & Purcell 1999; Reed 2008; Richards et al. 2004; Tippett et
al. 2007). Facilitators must be able to build positive energy, develop good rapport with
each stakeholder group, be impartial to the different viewpoints involved, and be easily

	
  

5	
  

approachable (Reed 2008). Facilitators should also evoke a sense of trust, be
transparent in their actions, respect participants and be patient (Chambers 1994a, b;
Kapoor 2002).

Importantly, facilitators must also be able to effectively step back when it is appropriate
for participants to lead themselves (Chambers 1994a; Dogbe 1996; Francis 2001).
Finally, when selecting a facilitator project managers should consider their range of
experiences; although training guides have been developed (Pretty et al. 1995), field
experience is viewed as one of the most important components to building quality
facilitation skills (Chambers 2003).

1.2 Common Methods for Supporting Development Decision-Making
1.2.1 Rapid Rural Appraisal (RRA)
Developed in the 1970’s by Robert Chambers, RRA encompasses a suite of methods
that better enable outsiders to elicit information about various factors of local life and
environment from community members (Chambers 1994b). RRA was motivated by
practitioners’ overall disillusionment with the then current methods that were timeconsuming, expensive, and failed to gather valid and reliable data (Chambers 1994b).
Thus, the defining characteristic of RRA is that its associated methods aim to be the
quickest and least expensive to carry out while still producing reliable results (Beebe
1995; Wilkins et al. 2004).

	
  

6	
  

Rapid Rural Appraisal has been applied in multiple disciplines including agricultural
planning (Drinkwater 1993 ), watershed management (Kerr & Sanghi 1992) and health
(Cornwall 1992). RRA is generally a multidisciplinary approach that is a more intense
field process (Beebe 1995; Wilkins et al. 2004). RRA begins with data gathering
followed by a “filtering” of the data through the lens of the researcher to provide
thorough detail and insight (Wilkins et al. 2004). Primary methods used to gather data
include the use of secondary information, key indicators, scoring and ranking,
community mapping and direct observation.

Secondary information includes reports of surveys, government statistics, annual
reports, academic papers and other similar sources.

RRA recommends that

researchers utilize these sources to avoid unnecessarily collecting data that may
already exist, thereby saving time (Chambers 1981). Key indicators are markers that
represent progress toward a goal. Examples include using soil color as an indication of
soil fertility and particle size distribution; birth-weight to predict maternal health and
projected life spans of children; and available housing (with high or low quality housing
defined locally; indicators may include physical characteristics or ownership
opportunities for women) as an indicator of area prosperity or relative poverty
(Chambers 1981; Mayoux & Chambers 2005).

Scoring and ranking methods enable local respondents to produce quantitative
representations of preferences for project alternatives. Potential alternatives are first
displayed visually, most often on the ground. Following, respondents place a certain

	
  

7	
  

number of sticks or stones next to individual alternatives to indicate which they favor
(Chambers 1981; Drinkwater 1993 ). Community maps and models are used to locate
characteristics of ecological and social environments within communities. These serve
as guides to local perceptions of the areas in which community members live and work
(Cornwall 1992). Direct observation occurs when the researcher personally observes
community life in that environment. Other associated methods include key informant
and semi-structured interviews with various community members.

Generally, the RRA method chosen depends on the research objective. For example,
Quinn et al. (2003) utilized ‘risk mapping’ to illicit local perceptions of livelihood related
risks in semi-arid Tanzania.

Risk mapping combines qualitative information with

quantitative analysis to produce ‘risk maps’, which are a geographic depiction of the
problems communities face.

To conduct risk mapping, the authors utilized structured questionnaires to ask families
in twelve different villages about perceived obstacles related to their ability to provide for
their families.

Authors reported using carefully phrased, open-ended questions to

enable respondents to identify a wide range of concerns. Once identified, respondents
ranked them in terms of their importance.

Twenty-one prominent risks were documented and grouped into four categories:
natural, physical, financial and human/social.

Results were analyzed by applying

incidence and severity index measures. An incidence index represents the number of

	
  

8	
  

times respondents identified a certain problem.

The severity index quantifies the

severity of each identified problem using the following equation: 1 + (r – 1)/(n-1) where
‘r’ represents the rank of the problem based on the order in which it was identified, and
‘n’ represents the total number of problems recognized by the respondent. In this study,
water availability was identified as the most pressing risk.

Quinn et al. (2003) viewed the method as successful in its ability to identify and
understand local risk perceptions and suggested that it may be an asset to other
endeavors. The authors did acknowledge potential biases, such as that perceived risks
may have changed during different seasons and that respondents may have given
answers they felt were more likely to lead to direct project implementation. Still, Quinn
et al. (2003) assert that understanding perceptions of local risks is vital for projects
wishing to implement poverty alleviation strategies.

1.2.2 Participatory Rural Appraisal
Participatory Rural Appraisal (PRA) was developed in the 1980s (Chambers 1994a).
PRA’s key principles include: (i) participation in research where local people serve as
partners in the collection of data and its analysis; (ii) flexibility; (iii) teamwork; and (iv)
optimal ignorance (e.g., allowing ample time for analysis and planning but keeping time
and cost in mind) (Rietbergen-McCraken & Narayan 1998). PRA also emphasizes that
results of participatory research findings should be presented to other community
members and non-locals involved in the research to validate findings (Chambers
1994a).

	
  

9	
  

Though there are similarities with RRA, PRA is differentiated by its much more
participatory approach.

Rather than extracting information and taking it away for

analysis, PRA provides tools for local participants (not just respondents) to carry out
their own research collection and analysis (Asian Forest Network 2002). This emphasis
on participation is often viewed as a way to build rural empowerment (Asian Forest
Network 2002; Cooke & Kothari 2001; Ebrahim & Herz 2005; Mansuri & Rao 2004;
Motteux et al. 1999; Rietbergen-McCraken & Narayan 1998).

Participatory rural appraisal has been applied in a variety of fields including natural
resource management, agriculture, poverty, social programs, as well as health and food
security. PRA offers a multiple techniques, many of which overlap with RRA, including
ranking exercises (such as using local resources - sticks or stones -to indicate
preferences for alternative project interventions) and semi-structured interviewing
(Rietbergen-McCraken & Narayan 1998). PRA also utilizes methods such as: transect
walks; venn diagramming; free listing and card sorting; trend analysis; and
presentations of research findings given to all participants.

Transect walks are systematic walks taken with community members or key informants
through the targeted area. Particularly useful at the outset of a project, these methods
enable practitioners to formulate an understanding of the area where the project will
take place, especially in farming systems and natural resource inquiries. Walks are
done in a combing or sweeping method (Mascarenhas & Kumar 1991) and are recorded

	
  

10	
  

by drawing matrices where the area of the walk forms the top of the table and the
criteria studied are listed on the side of the matrix (Kirsopp-Reed 1994).

Venn diagrams help identify community institutions, key community members and their
relationships with one another (Kapoor 2002; Mascarenhas & Kumar 1991). Card
sorting enables people to express their knowledge, opinions and preferences first by
listing them and labeling them on multiple cards. Cards are then sorted into categories
and prioritized (Chambers 2003). To conduct trend analysis, PRA utilizes seasonal
calendars, daily activity charts and historical diagramming (Rietbergen-McCraken &
Narayan 1998).

The World Bank often utilizes participatory rural appraisal methods. For example, in
1993 the World Bank drew upon techniques from Participatory Rural Appraisal to
conduct a poverty assessment in four urban and six rural communities in Zambia. The
primary objectives of the research were to: (i) explore local perceptions of poverty and
well-being; (ii) identify local conceptions of the most effective strategies for poverty
reduction; (iii) explore each community’s main concerns related to poverty and
determine if and how these concerns changed over time; and (iv) to determine local
perceptions regarding policy changes undertaken to enhance economic freedoms.

The PRA study was conducted over a 10-week period. The principal methods utilized
included wealth and well-being ranking activities, seasonal diagramming, livelihood
analysis, thematic mapping of village resources and institutional mapping of social

	
  

11	
  

organizations and agencies available to community members. Research findings were
reported to the World Bank and highlighted the priorities identified by local people for
government policy in health, water and sanitation, transport infrastructure, garniture,
natural resource management and education (Rietbergen-McCraken & Narayan 1998).

1.2.3 Beneficiary Assessment
Similar to RRA and PRA, Beneficiary Assessment (BA) is a consultative methodology
used to inquire about people’s perceptions of planned projects and policies that aim to
build positive social and economic change (Salmen 1999; Salmen 2002). Designed in
the early 1980s, BA is the World Bank’s most widely used methodology (Francis 2001).
The defining characteristic of BA is that its research methods target mainly those people
directly benefitting from a project; results are used to inform the managing decision
makers who are normally not located at the project site (Francis 2001; Salmen 2002).
Technical specialists are selected to guide the approach. Practitioners utilizing BA hope
that it will lead to a more successful project implementation process and a more
sustainable development initiative overall (Salmen 1989).

Within the World Bank, BA has been applied in a number of sectors including health,
agriculture, energy, environment, transportation, and industry (Rietbergen-McCraken &
Narayan 1998). The methodology draws upon both market and social research and
emphasizes qualitative research by combining direct and participant observation,
conversational interviewing, and focus groups. The emphasis on qualitative information
is meant to balance and give deeper, well-rounded meaning to quantitative data results

	
  

12	
  

(Salmen 2002). However, qualitative results are most often converted into quantitative
data (Rietbergen-McCraken & Narayan 1998).

Direct observation in BA is viewed as the simplest technique, requiring the researcher to
simply note the behavior and expressions of potential project beneficiaries through
direct observation of them in their environment. This is sometimes done in a more
participatory manner where the researcher takes part in beneficiaries’ daily lives.
Conversational

interviewing,

otherwise

described

as

well-guided,

naturalistic

interviewing, is a structured interview that aims to aid respondents in revealing their
feelings about the issue at hand. Interview standards include (i) that the time frame
should not exceed forty-five minutes to one hour at a time; (ii) that interviews should be
conducted away from possible distractions; and (iii) that interviews should be adapted to
the culture of interest (e.g. gender should be taken into consideration, and recorders
and cameras, if used, should not be intrusive). Focus group discussions are similar to
conversational interviewing except that they include more than one person (Salmen
2002).
	
  
In 1995 the World Bank hired a Canadian Non-Governmental Organization to conduct
a Beneficiary Assessment to evaluate current water supply and sanitation strategies for
two different projects in Luanda, Angola. The primary objective of the research was to
consult the community beneficiaries of both projects regarding the potential for urban
upgrading of both supply and sanitation systems. More specifically, the assessment
sought to increase community knowledge about the current water distribution system

	
  

13	
  

and gain information regarding existing water and sanitation practices, as well as
willingness to pay for improved services and suggestions on what those services might
be.

The Luanda project was conducted over a period of six weeks and divided into three
phases. In the first phase, researchers worked with key informants to designate their
sample population. In the second phase, conversational interviews were completed
with community water distributors to investigate the current distribution system
availability, including water sources, profits for distributors and water truck availability.
In phase three, 60 focus groups were conducted with local beneficiaries regarding
current water and sanitation practices as the focal topic.

Each group discussion

followed a guide that drew upon methods associated with RRA and PRA to stimulate
discussion. Water and sanitation themed photographs were used to begin the
conversations, drawings stimulated community priority discussions, card sorting
categorized current sanitation techniques into categories of inadequate and adequate,
and participatory mapping determined community risks to poor sanitation.

The resultant findings had important impacts on the design of the water and sanitation
project. Initial assumptions about community priorities for water and sanitation were
completely reversed, which directly affected the ultimate decision regarding how to
upgrade the $1 million project (Rietbergen-McCraken & Narayan 1998; Salmen 2002).

	
  

14	
  

1.2.4 Stakeholder Analysis
Stakeholder Analysis (SA) is a method that aims to understand a system by identifying
the stakeholders within that system and investigating their related interests and
concerns (Grimble et al. 1995). It evolved from a recognition that the diverse interests
of all relevant stakeholders need to be included in management processes and that the
differences in problem perceptions ought to be explored. The primary objective of SA is
to improve the effectiveness of development policies and projects by making those
interests an explicit consideration in a project’s design, implementation, and monitoring.

The SA method begins by dividing potential stakeholders into macro and micro level
groups. Macro level stakeholders include national institutions, national departments
and/or any international organizations.

Micro level stakeholders are those who are

more locally based, such as local farming groups or environmental lobby groups.
Following, the SA methodology recommends: (i) identifying the analysis’s purpose; (ii)
developing a holistic understanding of the system and its decision makers; (iii)
identifying primary stakeholders for involvement (within the different levels); (iv) eliciting
stakeholder interests; (v) determining interactions between stakeholders and the context
in which interactions occur; and (vi) identifying management options.

SA uses a variety of methodologies to complete these steps. Those found to be most
beneficial include informal and semi-structured interviews that involve checklists of key
topics to examine stakeholder interests, as well as oral case histories (e.g. to
investigate changes over time and the internal dynamics of the system). Matrices to

	
  

15	
  

visually represent the conflicting objectives and potential tradeoffs between different
levels of stakeholders and their interests are also used, as is preference ranking to
determine stakeholder values (Grimble et al. 1995; Grimble & Wellard 1997;
Rietbergen-McCraken & Narayan 1998).

When 43 coal mine closures were anticipated to occur in Ukraine in 1996 as part of the
country’s structural readjustment phase, the World Bank undertook SA to identify those
most likely to be affected by the closures. The study aimed to determine baseline
conditions of affected communities, and, most importantly, to elicit miners’ perceptions
on the closures and determine their priorities for assistance.

Fourteen stakeholder

groups were identified, of which four were selected for initial consultations regarding
assistance priorities. Yet, only two were eventually carried out due to time constraints.
Mine managers and representatives from the mines’ union were also selected to
participate.

The SA was divided into two phases. Methods used in the first phase were in-depth
interviews and focus group discussions during consultations with individual miners.
Interview guides that consisted primarily of open-ended questions directed at the mining
families’ current livelihood status were utilized. Focus group discussions were held with
groups of miners and union representatives.

Findings from the SA revealed that the initial assumptions about expected social costs
were much lower than previous estimates predicted. This determined that the originally

	
  

16	
  

proposed mitigation package for affected miners was extremely inadequate.
Interestingly, the analysis also revealed that many miners were skeptical that the
closures would actually occur due to the Ukrainian government falsely reporting
closures in the past. As a result, it became imperative to inform the affected groups of
the reality of the situation (Rietbergen-McCraken & Narayan 1998).

1.2.5 Stated Choice Methods
Stated Choice Methods (SCM) offer a flexible approach to collect preference data
related to proposed changes in attributes of existing or hypothetical goods, policies, or
services (Adamowicz et al. 1998; Louviere et al. 2000).

They employ survey

instruments to present respondents (stakeholders affected by the context within which it
is employed) with different alternative outcomes characterized as multi-attribute
scenarios representing potential changes in policies or management strategies
(Adamowicz et al. 1998). The primary objective of SCM is to investigate, estimate and
predict the behavior of those who would be affected by such changes (Louviere et al.
2000).

Of the approaches that make up SCM, the most commonly used are attribute-based
methods, contingent valuation and paired comparison.

There are slight differences

between each: contingent valuation estimates willingness to pay only; attribute-based
methods and paired comparison reveal willingness to pay and the preference order of
alternatives (Brown 2003).

Yet, despite their differences, each is a survey-based

method that combines qualitative and quantitative data to produce results. They are

	
  

17	
  

used to examine potential implications of changes in policies or management practices;
results inform decision makers responsible for managing such changes.

The main steps in choice experiment design include: (i) specification of the valuation
approach; (ii) definition of attributes, levels, and customization; (iii) theoretical
construction of measurement objective, or, the experimental design and context; (iv)
questionnaire development; (v) choice of sample and sampling strategy; and (iv) mode
of response; and appropriate statistical models to scale responses. Bergmann et al.
(2004, as cited in Porras & Hope 2005), emphasize that included attributes ought to be
relevant and applicable to the problem being analyzed, realistic, and easy for
respondents to understand. Brown (2003) suggests additional steps, such as including
enough background information in the surveys to ensure respondents have adequate
information to make an informed choice.
	
  
Porras and Hope (2005) used Conjoint Analysis (CA) to investigate local residents’
willingness to engage in different land use scenarios aimed at conserving the local
watershed. In this study, participants were provided with information regarding the
watershed management problem and the different implications for upstream and
downstream users. Next, each participant was presented with four different land use
options, with the current management strategy retained as one option. The researchers
hoped that the CA would enable a more thorough understanding of the opportunities
and benefits available to farmers if they were to engage in a Payments for
Environmental Services (PES) project. Thus, all options included different levels of PES

	
  

18	
  

involvement – including length of PES contract, electricity payments, access to state
benefits, and road investments. Respondents rated their likelihood to support a given
scenario on a scale of 0-10, with 0 being the least likely to support.

Results demonstrated that participants were more likely to take part in a given scenario
if it included road investments, land titling and subsidies access, pasture existence and
land area investment.

Engagement likelihood decreased when the percentage of

required forest cover and the PES contract length increased. Payment levels had little
effect on willingness to engage.

While these findings were informative, the authors suggested that they also presented a
greater challenge to identify why participants were so opposed to changing current land
use practices. Questions were raised as to whether or not PES monetary payments
were too little or if there were political factors or other institutional challenges that
negatively influenced the involved participants’ reaction to the government-related
management options.

1.2.6 Holistic Management
Holistic management (HM) is described as a decision-making framework that identifies
and evaluates management plans against a desired future state (Savory 1991; Tippett
et al. 2007).

It focuses primarily on environmental, rangeland and livestock

management and considers economic, social and ecological sustainability.

The

incorporation of local ideals and goals in all stages of planning is emphasized, as is a

	
  

19	
  

more structured decision process. Allen Savory, the practitioner who developed (and is
most often associated with) HM, believes that individuals often assume they have come
to the right decision, but cannot be sure until they experience the outcome first-hand.
Thus, in order to better ensure people will be satisfied with outcomes, they must arrive
at a choice through a more structured approach that recognizes that many desired
goals may conflict with one another (Savory 1991).

Holistic management recommends following three steps to implement a decision
making process. First, decision makers should define the quality of life they wish to
achieve. Second, they should indentify the modes of production required from the land
or resource they are managing to sustain that quality of life. Finally, the individuals
responsible for managing a specific environmental area or resource should define that
entity in terms of management responsibilities and management tools available (Savory
1991; Tippett et al. 2007)

One of the most well known applications of holistic management was conducted at the
Dimbangombe Ranch in Zimbabwe.	
   	
   The Dimbangombe Ranch is located 22 kilometers
from Victoria Falls and encompasses 8000 hectares of private and state land. The
nearby community Wange is separated from Dimbangombe by a road a small patch of
forest state forest owned land. In 1994, Dimbangombe was donated to the African
Center for Holistic Management, which is now located on the property. The center
partnered with Wange community members to undertake Holistic Management in an
effort to decrease land degradation.

	
  

This was done by increasing the number of

20	
  

animals herding on the ranch and planning their grazing patterns to avoid re-grazing of
plants prior to their recovery from previous grazing.

Neely and Butterfield report that this change in grazing patterns mimicked more natural
patterns and that the increased hoof action and animal dung improved soil fertility.
Also, the livestock brought from the Wange community - that was starving at first regained health2 (Neely & Butterfield 2004).
	
  
1.3 Conclusion: Future Research to Improve Participatory Decision Making
Practices
Attitudes and practices regarding the incorporation of local preferences and knowledge
in international development decision making have changed dramatically since the
1960’s. Efforts to better integrate local viewpoints have improved the relevance and
longevity of many projects. Some projects adopted major revisions in approach after
consulting with their targeted communities, enabling them to avoid misdirected project
objectives (Mosse 2001). However, altohugh significant improvements in methodology
and philosophy have been made, additional progress is necessary to better ensure that
stakeholders’ opinions are thoroughly integrated.

For example, the process of eliciting and analyzing local preferences for alternative
development projects is one area that would benefit from increased attention. As early
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
2	
  

For more information on the Dimbangombe Ranch see:
Ecoagriculture Partners (Ecoagriculture Partners 2009).
	
  
	
  

21	
  

as 1995, inadequacies in this area were identified, characterized particularly by the
failure to include respondents’ weights in their rankings of project alternatives (Maxwell
and Bart 1995). There has been some discussion about how to better elicit and apply
ranks (Chambers 1993; Maxwell & Bart 1995), but overall, the question of structuring a
better approach to this problem has not been conclusively answered within the
development discourse as calls for better understandings of local preferences continue
(Albert et al. 2010a).

A focus on more explicitly confronting tradeoffs that occur across multiple project
objectives is also largely missing from the development literature. Implicitly, it has been
discussed by referring to the importance of understanding such tradeoffs. However, to
the best of my knowledge, little concrete discussion has occurred in a way that actually
helps participating decision makers to manage and assess tradeoffs in the decision
making process.

Recently, behavioral decision researchers working in developing communities have
asserted that more explicit attention needs to be paid to the decision making process as
a whole process.

Kellon and Arvai (2011) believe that the lack of comprehensive

processes sends an inadvertent message that the largest implementation obstacles
faced when integrating stakeholders into a decision are first, convincing managers to
use a participatory process and second, ensuring that all stakeholders work together in
a fluid manner.

While some development literature offers specific frameworks for

decision making (Ebrahim & Herz 2005), the overall structure appears to be more an

	
  

22	
  

assemblage of guiding principles and less a rubric for applying the intricate details of
decision making (such as weighting objectives and actively confronting tradeoffs) to
ensure that participants have the means to reach their optimal decision outcomes.

Meanwhile,

insights

from

behavioral

decision

research

have

increased

our

understanding about how individuals make choices and offered structured ways to
improve decision making.

These approaches enable decision makers to better

understand their values and translate them into objectives, as well as make necessary
tradeoffs (Hammond et al. 1999; Keeney & Raiffa 1976).

Thus, it is my view that

behavioral decision research represents an area of scholarship and practice that ought
to be better integrated with the current discourse on international development.

	
  

	
  

23	
  

Chapter Two
Behavioral Decision Research

2.0 Introduction
Behavioral decision research (BDR) seeks to understand how individuals make
decisions. It investigates human information processing needs and cognitive limitations
inherent in decision making that affect the quality of individual and group decisions. It
also develops techniques to improve decision making, particularly when decisions are
associated with risk or uncertainty (Keeney & Raiffa 1976). BDR has been conducted
in, and applied to, a number of different fields including education, medicine,
economics, political science, and geography (Slovic et al. 1977). Theoretical research
in this area is generally divided into two areas of study: normative decision theory and
descriptive decision theory.

2.1 Normative Behavior Theory
In 1713, Swiss professor Nicola Bernoulli conducted research that would later set the
stage for behavioral decision research. Bernoulli had developed a coin flipping game
that would enable participants to win a certain amount of money depending on the
number of subsequent times ‘tails’ appeared when the coin was tossed. For example,
the participant would win two dollars if ‘tails’ appeared the first time, four dollars if it
appeared the second, and eight dollars if it appeared the third. The amount would
continue doubling as long as ‘tails’ continued appearing. By not deducting money if
heads appeared, Bernoulli theorized that the expected utility, meaning the probable
value, of this game was infinite.
	
  

24	
  

Bernoulli’s question then was, how much a participant would be willing to pay in order to
play this game? Interestingly, he found that players were not willing to pay more than a
small amount to play the game despite its potentially large payoffs. This experiment
was subsequently dubbed the St. Petersburg Paradox, and it raised several questions
for BDR to answer. Years later, Bernoulli’s younger cousin Daniel re-worked the St.
Petersburg paradox and theorized that the value people place on the amount of money
won declines as the amount increases. He then questioned whether or not the value of
the game was really infinite, at least in the eyes of the decision makers. His question
influenced many subsequent theories of human choice behavior. Of those, expected
utility theory is one of the most famous (Plous 1993).

Following the work of Bernoulli, expected utility theory (EUT) was coined by Von
Neumann and Morgenstern in 1947 and represents a normative theory of decision
making. Normative theories describe how an individual should behave in order to be
considered a rational decision maker (Plous 1993). Expected utility is comprised of six
axioms of rational decision making that, if followed, ought to guide a decision maker to
an optimal choice. Plous’s (1993) explanation of these axioms is included below.

1: Ordering of Alternatives
The first axiom of rationality is the ordering of alternatives. This axiom states that a
rational decision maker should be able to effectively compare two different choice
outcomes and, after evaluating the different attributes within each, ought to prefer one

	
  

25	
  

to the other. If not, a decision maker should be indifferent to the outcome of the choice
being made.

2: Dominance
The second axiom is dominance, which claims that a rational decision maker should
never accept a decision outcome that is ‘dominated’ by another. A dominated outcome
is one that, when compared to an alternative outcome, has a lower expected utility.
Dominance can be strong or weak, depending on the difference between the expected
utilities. An outcome is strongly dominant if it generates a better outcome in multiple
respects. An outcome is weakly dominant if it generates a better outcome in only one
respect.

For example, if the decision at hand is ‘what apartment to move into’,

Apartment A is strongly dominated by Apartment B if Apartment B has cheaper rent and
more square footage, but is in an equally desirable location. Apartment A is weakly
dominated is Apartment B has only more square footage, but is equal in rent and its
location. Expected utility theory states that a decision maker should never choose an
option that is dominated, even if it is only weakly dominated (Plous 1993).

3: Cancellation
Cancellation, the third axiom, states that attributes of different decision outcomes that
are identical to one another should not be considered in the decision. For example, if a
person were choosing from among airlines on which to fly and the price among all
options were the same, price should not be a consideration in the final choice. Instead,
a rational decision maker should only focus on the attributes that are different. In this

	
  

26	
  

example, the flight departure and arrival times may be different or airline safety ratings
might be contrasting.

4: Transitivity
Transitivity, the fourth axiom, asserts that, if Alternative A is preferred to Alternative B,
and Alternative B is preferred to Alternative C, then Alternative A should be preferred to
Alternative C. A decision maker’s choice should not be intransient by, according to the
given example, choosing Alternative C instead of Alternative A (Plous 1993).

5: Continuity
Continuity declares that, for any grouping of outcomes, a rational decision maker should
prefer to gamble for an outcome with the best expected value instead of taking a sure
outcome that has a lower expected value, as long as the probability of gaining the better
outcome is good enough.

6: Invariance
The final axiom, invariance, specifies that the choices of rational decision makers
should not be affected by the way a decision is presented or framed. Plous (1993)
explains this using an example of a lottery game that is presented in two different ways.
The first way presents the lottery as a two-stage game in which the person playing has
a 50% success in each stage, with the payoff of $100.00 if the player is successful both
times. The second way condenses it into a one-stage game, with a 25% chance of a

	
  

27	
  

$100.00 prize. According to invariance, a decision maker should have no preference
between the two games if they are presented as two different lottery options.

Since the development of EUT, several studies have shown that people are actually
relatively poor decision makers, consistently violating the six axioms of rational choice.
For example, framing effects are found to often build barriers to effective decision
making (Frisch & Jones 1993; Gregory et al. 1993). Seminal work in this area was
conducted by Tversky and Kahneman (1981). They conducted a series of experiments
that found significant changes in preference when minor changes were made in the
framing of their different choice outcomes. Specifically, participants were risk seeking
when the options presented to them were framed as losses and risk averse when
options were framed as gains.

For example, gambles with a 25% chance to lose

nothing and a 75% chance to lose $1,000 were preferred to an alternative sure loss of
$750. Yet, when participants were presented with a sure gain of $240 or a 25% chance
to win $1,000 and a 75% chance of winning nothing, they preferred the sure gain. The
experiments demonstrated a clear violation of rationality.

This and other violations of rationality do not result from ignorance on the part of the
decision maker. Instead, research on descriptive decision making that addresses how
people actually make decisions rather than how they should has documented that
people are prone to many cognitive limitations when making decisions.

These

limitations inhibit their ability to fully comprehend the multi-dimensional aspects of a
decision, including difficulty addressing and comprehending the tradeoffs that occur

	
  

28	
  

across multiple decision outcomes (Arvai et al. 2006; Gregory et al. 1993; Kahneman et
al. 1982; Keeney 1992).

Decision makers are also often unable to recognize the role outside influences have on
their decision processes (Payne et al. 1992). As a result, there is often a heavy reliance
on heuristics and other shortcuts to arrive at decision outcomes.

This reliance

contributes to the common failure to consider all factors relevant to the decision at hand,
which often leads to suboptimal decisions (Arvai et al. 2006; Bohnenblust & Slovic
1998; Gintis 2000; Gregory et al. 1993; Kahneman et al. 1982; Slovic 2000; Tversky &
Khaneman 1981).

Descriptive theories of decision making and the most common

heuristics and biases of decision making are discussed next.

2.2 Descriptive Theories of Decision Making
Descriptive theories of decision making work to uncover how people actually make
decisions rather than how they should make them (Slovic et al. 1977). These theories
explain the impracticality of normative models of decision making and discuss the
cognitive factors that have a strong effect on individual and group processes. Among
the numerous alternatives to normative decision making’s Expected Utility Theory,
Prospect Theory is one of the most widely acknowledged.

2.2.1 Prospect Theory
Daniel Kahneman and Amos Tversky devised Prospect Theory in 1979. In prospect
theory ‘utility’ is referred to as ‘value’ and value is defined in terms of gains and losses
	
  

29	
  

rather than wealth. Its main tenet is that people feel losses stronger than they do gains.
When depicted visually, the loss or gain is represented as an S-shape value function
with the curve appearing steeper in the loss region than in the gains area (Tversky &
Khaneman 1981). This theory provides an explanation for the previously mentioned
documentations of invariance violations such as loss aversion.

Another example of loss aversion was studied by Gregory et al. (1993).

In their

research, participants were presented with two different alternatives for an
environmental policy: the first was to restore a forest that was previously lost; and the
second was to improve upon the existing state of a current forest (e.g., ensure no
further trees would be cut down). In both cases, the same number of trees was going to
be affected (i.e. restored or improved). Thus, in consideration of the invariance axiom
(in that Alternative A is equal to Alternative B, in this case), both projects presented the
same outcome and rationally speaking, decision makers should have been indifferent to
either approach. However, Gregory et al. found that participants in the study much
preferred restoring the forest that was previously lost to improving upon the state of the
forest that was currently in existence.

Prospect theory’s account of loss aversion has influenced other theories in descriptive
decision making, one of which is the endowment effect. The endowment effect states
that a person values a good more strongly when it becomes part of that person’s
personal property.
theory.

	
  

The certainty effect is a second theory influenced by prospect

Developed by Tversky and Kahneman, the certainty effect states that a

30	
  

reduction in the probability of an outcome has a stronger effect when the outcome is
certain compared to one that is merely probable.

A common example used to explain the certainty effect is the relative importance of
certainty involved with Russian roulette. In this example, the majority of people would
pay more money to remove the only bullet from a gun used for Russian roulette than
they would to remove one of four bullets. Even though the probability of being shot is
reduced by the same margin (one bullet) the certainty of zero bullets remaining is felt
more strongly than if only one among many is removed. Both the endowment effect
and the certainty effect influence violations of expected utility theory (Plous 1993).

2.2.2 Preference Construction
The nature of constructed preferences is another important finding from descriptive
decision research. In opposition to common thought in behavioral economics, people
do not always come to a choice situation with previously determined preferences (Gintis
2000; Slovic et al. 1990). Instead, research demonstrates that preferences for decision
outcomes are often developed on the fly during the decision process itself (Arvai et al.
2006; Payne et al. 1992; Slovic 1995; Slovic et al. 1990).

Early writing on constructed preferences states that this reality of preference
construction is due to human inability to have unwavering and thoroughly evoked
preferences that are tied to precise goals (March 1978).

Often referred to as an

“architect versus archaeologist” perspective, decision science states that preferences

	
  

31	
  

need to be carefully built, rather than simply uncovered during the decision process
(Gregory et al. 1993).

2.2.3 Bounded Rationality and Satisficing
Herbert Simon’s ‘bounded rationality’ (Simon 1990) is another descriptive theory that
directly contrasts normative decision thought. Bounded rationality is based on Simon’s
argument that, in order to understand actual decision making behavior, research needs
to focus on the cognitive and perceptual limitations inherent in human decision behavior
that lead people to diverge from enacting in a normative, more rational, fashion (Payne
et al. 1992).

Simon’s theory asserts that because optimal decision making

circumstances are seldom available (e.g. time to make decisions may be lacking or all
relevant information may not be readily accessible), humans consistently generate
different techniques to solve problems in a more approximate manner (Simon 1987,
1990).

One of the most common techniques employed under bounded rationality is satisficing.
Also coined by Simon, satisficing occurs when a decision maker settles for a decision
outcome that meets their objectives adequately, rather than searching for an optimal
outcome even if one exists (Plous 1993).

Simon’s work has influenced others to

research heuristics people use to come to a decision outcome. Three of the most
common heuristics are representativeness, availability, and anchoring and adjustment.

	
  

32	
  

2.2.4 Representativeness
The representativeness heuristic describes decision makers’ tendency to judge the
probability or frequency of an outcome occurring based on readily available data or by
using resemblance. Plous describes resemblance as “by the degree to which A is
representative of B, that is, by the degree to which A resembles B” (Plous 1993).
Tversky and Kahneman (1982) were the first to study this heuristic. In their research,
they conducted an experiment that presented participants with a passage describing a
woman. The passage portrayed the woman as a 31-year-old named Linda who was
single, very bright, and outspoken about issues regarding racial and social justice, as
well as nuclear energy policies.

Participants were then asked whether Linda was a (i)

bank teller or if Linda was (ii) a bank teller and active in the feminist movement.

Interestingly, the vast majority of respondents chose the second option.

This

demonstrated that Linda’s stated interests were thought to be more representative of
the second option even though the probability of her being both a bank teller and active
in the feminist movement is less than the probability of her simply being a bank teller.
This violated a fundamental rule of probability: that the probability of two events
occurring together is less than the probability of either event occurring alone (Plous
1993).

2.2.5 Availability
The availability heuristic is similar to representativeness.

Availability describes the

human tendency to evaluate the probability of an event occurring depending on how

	
  

33	
  

easily one is able to bring to mind instances in which those events took place
(Kahneman et al. 1982). For example, generally, events that are more publicized in the
media are thought to have higher probabilities of actually happening. The probability of
death by falling airplane parts is a common example. When asked, most people would
predict the probability of being killed by a shark at a higher rate than that of being hit by
a falling airplane part when, in reality, the opposite is true (Plous 1993).

Attempts to counter this heuristic are evident in many social outlets, particularly travel
and health awareness campaigns, that give explicit attention to the truth behind
common misconceptions. Specifically, many travel companies may reassure nervous
flyers that they are more likely to die in a car accident than they are in a plane crash
even though air crashes are more available than car crashes.

2.2.6 Anchoring and Adjustment
Anchoring and adjustment occurs when people make estimates by starting at an initial
value and adjusting upward or downward to determine their final answer. The values –
or anchors - may be irrelevant to the decision at hand, which can lead to insufficient
judgments.

For example, in a study by Tversky and Kahneman (1974) participants were asked to
provide a median estimate for the number of African countries in the United Nations.
Before providing their estimates, participants received the number 10 or the number 65
as an arbitrary number – in this case, the anchor. Results demonstrated that those who

	
  

34	
  

received the number 10 provided a median estimate of 25, whereas those who received
the number 65 provided a median estimate of 45.

In this case, participants anchored

heavily on the arbitrary number assigned and insufficiently adjusted for those numbers,
which influenced their final estimate.

Since Tversky and Kahneman first studied this heuristic, anchoring and insufficient
adjustment has been found to occur when people are asked to provide a diverse array
of estimates, including the proportion of chemistry professors that are female and the
number of Iranians who practice Islam (Plous 1993). All studies point out that decision
makers rarely adjust to a sufficient degree when anchors are present, regardless if the
estimate is something more familiar (such as average prices for renting an apartment)
or unfamiliar (the chances of an asteroid hitting your city). Suggested ways to defend
against this bias include thinking about what an alternative anchor may be and how one
would react to that instead.

2.2.7 Affect
In decision research, affect is referred to as the quality of ‘goodness’ or ‘badness’ that is
associated with either (i) a state of feeling or (ii) the demarcation of a stimulus. Affective
responses to stimuli often occur automatically and rapidly.

A reliance on such

responses in decision making is known as the ‘affect heuristic’ (Slovic et al. 2004).

The affect heuristic is employed when judgment and decision making processes are
influenced by affective responses to stimuli (Slovic 2005; Slovic et al. 2002; Slovic et al.

	
  

35	
  

2004; Wilson & Arvai 2006b). Essentially, the affect heuristic describes when a decision
maker consults their “affect pool” to come to a decision outcome. The affect pool is
made up of numerous images that have either positive or negative correlations.
Particularly used with decisions under risk or uncertainty, decision makers take cues
from their affect pool to guide their decision making process (Finucane et al. 2000;
Slovic 2000).

Damasio explains affect using the term “somatic markers”, which he states are stored
images marked positively or negatively that, when linked to a future outcome, serve as
either alarms or incentives (Damasio 1994; Finucane et al. 2000). Acknowledging the
importance of the affect heuristic has been on the rise ever since decision scientists
asserted that individual ability to utilize information available during decision making
relies highly on the affect responses we attach to that information (Slovic et al. 2002).

The aforementioned heuristics and biases describe some of the ways boundedly
rational decision makers arrive at decision outcomes.

Following heuristics has its

advantages, such as reducing time in normative decision processes, and its
disadvantages, including that their utilization often leads to systematic biases (Plous
1993).

Moreover, Payne et al. (1992) found that people in general are unable to

recognize the role these outside influences have on their choices. Thus, another
primary focus of decision research is developing different techniques to structure
decision processes in an attempt to help decision makers overcome biases and make
more informed, better decisions. One of these approaches is discussed next.

	
  

36	
  

2.3 Structured Decision Making (SDM)
Structured Decision Making (SDM) is an organized approach intended to identify and
evaluate creative options for decision outcomes.

The overall goal of SDM is to

decompose the decision into parts that are more cognitively manageable for individual
or group decision makers (Keeney 1992; von Winterfeldt & Edwards 1986). This is
done by providing decision makers with a step-by-step process that accounts for the
multiple cognitive biases and limitations one can face in a complex decision situation
(Arvai et al. 2001a).

This structured process helps decision makers utilize powerful, systematic methods in
place of weak heuristics and biases when there is no framework in place (Simon 1990).
SDM also combines technical expertise with public values and preferences and frames
individual and collective thinking in a way that enriches elicited objectives and informs
decision makers more thoroughly of the potential tradeoffs they will need to consider.
This approach is said to provide the links needed between analysis and deliberation, or
dialogue and assessment, when the public is incorporated into policy or management
decisions (Gregory & Keeney 2002; National Research Council 1996).

The development of SDM was motivated by increasing calls to better incorporate public
values and concerns into decision processes that are primarily expert driven, but have
outcomes that affect the public (NRC 1996). SDM was created to make the insights
from decision analysis that are most salient to such decision processes more available

	
  

37	
  

and applicable to a wider array of stakeholders and decision makers. Findings from
decision analysis findings, coupled with principals of value-focused thinking (Gregory
2002b), provide the foundation for structuring decisions.

There are five basic steps to a structured decision process (Hammond et al. 1999;
Kellon & Arvai 2011):

1. Clarify the decision context;
2. Define objectives and corresponding measurement criteria ;
3. Develop decision outcome alternatives;
4. Estimate consequences of potential alternatives; and
5. Evaluate tradeoffs that occur and select appropriate outcome

The first step, focused on clarifying the decision context, means to carefully define what
specific question or problem is being addressed and why.

Although basic, this step is

important as it frames the decision and determines the alternatives that may be
considered.

For example, while ‘which apartment to rent’ is a decision generally

confronted when a person is moving to a new city, the decision itself may be whether or
not to rent a condo instead of an apartment, or whether to rent a furnished apartment for
some time while learning where one would want to settle in the city (Hammond et al.
1999). A weak definition, or one that is incorrectly defined, may result in a suboptimal
decision outcome due to the decision process being based on the wrong problem or

	
  

38	
  

framed in an incorrect manner. This step also identifies the stakeholders to be involved
and what role they should play.

The second step is to define objectives and create measurement criteria. Objectives
are of primary importance to the decision maker as they establish the foundation for
subsequent evaluation of the potential decision outcomes. Decision scientists state that
thorough identification of objectives will ensure that unbalanced decisions can be
avoided (Hammond et al. 1999). Suggested ways to uncover objectives include: 1)
listing all concerns; 2) converting those concerns into objectives; 3) separating them into
ends and means; 4) clarifying the meaning behind each objective; and 5) assigning
evaluation criteria.

The third and fifth recommendations are perhaps the most important. The third
recommendation is to determine which objectives are a means to an end (means
objectives) and which are ends in themselves (ends objectives). Essentially, means
objectives “represent way stations in the progress toward a fundamental objective”
(Hammond et al. 1999). The recommended way to distinguish whether an objective is a
means or an end is to ask why that objective is important. When the answer reaches a
point where an explanation is not needed besides ‘it just is’, that objective is most often
determined as an ends objective. Tools often used to help organize the results of this
step are objective hierarchies, as well as ends-means networks meant to visually
represent the structure of objectives.

	
  

39	
  

In this step, evaluation criteria are identified and applied.

Evaluation criteria, also

known as attributes and measures, are used to assess the degree to which potential
alternatives will fulfill objectives.

There are three categories of evaluation criteria:

natural, proxy and constructed. Natural measures are those that align most closely with
the attribute in a direct sense, such as actual price for a “cost” objective.

Proxy

measures are more indirect, such as the acres of available habitat to denote the health
of a particular species. The third type - constructed scales - are developed specifically
for the decision at hand (Gregory et al. 2001).

The third step is to develop alternatives. Alternatives are the “raw material of decision
making” and thus bear considerable importance (Hammond et al. 1999). Two rules of
thumb should be kept in mind when identifying alternatives: 1) on cannot choose an
alternative that has not been identified; and 2) no matter the number of alternatives, the
chosen alternative can be no better than the best of those available.

The most common pitfalls associated with alternative identification include choosing an
alternative that represents business as usual, choosing the first possible solution
available, or choosing among only those presented by others. To avoid these pitfalls,
incorporate your own thinking first and ask others for suggestions later. Also, challenge
constraints (both mental and real, such as time constraints) and let your objectives
guide your alternative identification process (Hammond et al. 1999).

	
  

40	
  

The fourth step in SDM is to identify consequences of the potential alternatives. This is
an analytical task that uses the previously established evaluation criteria to determine
how the alternatives might perform (or are expected to perform) in terms of each
objective. For example, if the decision at hand is deciding ‘which office to rent’ and one
of the objectives is to ‘minimize distance to home’, the consequences of each
alternative in terms of that objective may be shown in miles or in the amount of time
spent in transit between a potential office and one’s home.

To assure well-defined and understood consequences, decision scientists recommend
building consequences tables. These tables are essentially a table of contents, such as
a chart, that summarizes the estimated consequences of the potential alternatives and,
in doing so, exposes the key tradeoffs to be considered across all considered
alternatives. A clear understanding of the consequences is important; if understood
thoroughly, the optimal decision may then become obvious, negating the need for the
fifth and final step (Hammond et al. 1999).

The fifth and final step is to address the tradeoffs that appear across multiple objectives.
It is unlikely that one alternative will meet all stated objectives. Therefore, it is important
to discover which alternative best meets the objectives identified as most important.
Often, this involves discarding the objectives that are either uninformative or are simply
dominated by other objectives.

	
  

41	
  

In the book Smart Choices, Hammond et al. (1999), state that the ability to thoroughly
understand and make wise tradeoffs is one of the most difficult, but more important,
challenges a decision maker faces. A commonly used tool to aid in the tradeoff analysis
process is even swaps. As the name indicates, an even swap occurs when the value of
one alternative is increased in terms of a one objective and decreased by an equivalent
amount in another objective. This enables decision makers to adjust the consequences
of attributes to make an attribute in a given objective equivalent to the others. That
attribute is then uninformative and able to be cancelled out. Essentially, even swaps is
a form of bartering that forces decision makers to think about the value of one objective
in relation to the value of another.

In highly complex decisions, it is sometimes necessary to go beyond the initial steps of
SDM. One of them is to identify and clarify the uncertainties involved in the decision.
A second is for the decision maker to think through their risk tolerance: Namely, one
should choose the alternative that also does not exceed a risk level that is appropriate.
A third step is to consider how the decision made today, may affect decisions in the
future (Hammond et al. 1999). And a fourth step is to employ a utility function. This
step is akin to decision analysis and is part of advanced SDM. It enables a decision
maker to apply a numerical value to individual attributes within each alternative. These
values represent their relative importance to the decision maker. The alternative with
the highest calculated utility represents the optimal decision.

	
  

42	
  

Finally, an important distinction to make about structured decision making is that its
point is not to reach consensus among all involved stakeholders.

Many decision

scientists feel that focusing on consensus may hinder the development of creative
alternatives for decision outcomes and lead to the adoption of mediocre policy choices
(Gregory et al. 2001b). Instead, decision aiding tactics embrace differences in opinion
and encourage thoughtful deliberation of conflicting values and objectives.

These

differences are viewed as the foundation needed to reach acceptable agreements
overall (Gregory et al. 2001b).

2.4 Case Studies
Structured decision making has been applied to a wide variety of problems, and many
examples are documented in actual policy and management decision processes. Some
apply step-by-step SDM techniques; others incorporate certain components that are
most appropriate for the context of the decision.

The quality of a SDM approach

depends mostly on the intentions and goals of all involved stakeholders, as well as
whether or not the included SDM steps were carried out from start to finish.

The

following examples successfully employ one or more of the fundamental elements of a
structured approach.

2.4.1. The Alouette River Stakeholder Committee
McDaniels et al. (1999) utilized SDM to facilitate a multi-stakeholder planning process
aimed at designing a new operating plan for a BC Hydro dam in Canada. The highly
controversial plan for managing water flow arose out of concerns for the dam’s effect on
	
  

43	
  

the environmental health of the Alouette River. Stakeholders included representatives
from BC Hydro, First Nations, local residents, provincial and government agencies and
other key user groups. The main objective was to determine the best management plan
for the Alouette facility that balanced conflicting objectives such as increased water flow
for the health of fish populations and costs of foregone power production for residents.

The facilitators used a combination of value-focused thinking and decision aiding
techniques to elicit objectives, create potential alternatives, and address involved
tradeoffs. McDaniels et al. (1999) deemed this venture a success overall due to the
ability of involved stakeholders to develop and recommend a highly effective
management alternative. Key elements contributing to the success of the SDM process
were the transparent use of stakeholder input, the thorough exploration of stakeholders’
values and affective responses and finally, the inclusion of learning over time in the
management strategy.

2.4.2 Integration of Biological and Sociological Needs in Northwestern Montana
In this example, Maguire and Servheen (1992) explain how they utilized decision
analysis and tradeoff analysis techniques in combination with expert opinion to design a
program that met the biological needs of a grizzly population and the sociological needs
of the area’s residents. In this case, the researchers originally thought that the elicited
objective of maximizing grizzly bear population could not be met while also meeting the
objective to minimize human-bear conflicts. However, the tradeoff analysis process
revealed that the maximum limit for conflict and the minimum limit for retention could be

	
  

44	
  

met by reintroducing only female grizzly bears that were between four and eight years
of age.

2.4.3 Mining in East Malaysia.
Gregory and Keeney (1994) utilized a structured approach to develop policy alternatives
for a proposed mine site in the Sabah Maliau Basin in East Malaysia. The primary
objective was to use stakeholder values to build the decision objectives in hopes that it
would lead to better policy alternatives.

This process, which spanned three days,

developed six alternative policies, compared to only two alternatives that stakeholders
had previously considered.

Gregory and Keeney (1994) state that, along with the

successful process resulting in increased alternatives, the structured approach enabled
stakeholders to have open communication and work together as equals on a very
important risk management decision, which, in the future, they believe will aid enhanced
communication and negotiation among the stakeholder groups.

2.4.4. Experimental Tests of SDM Approaches
In addition to the applied SDM cases mentioned previously, several studies have been
aimed at determining the efficacy of structured approached in comparison to nonstructured approaches.

In one study, Arvai et al. (2001a) tested whether or not SDM would lead to more
thoughtful and well-informed decisions.

To do so, Arvai et al. provided study

participants with a hypothetical decision context that involved a hydroelectric facility’s
	
  

45	
  

management practices and its effect on salmon habitats in British Columbia.
Participants were then placed in a decision workshop that utilized either a SDM
approach or a more typical “alternative focused approach” (AF). At the end of the
workshops, participants self-rated the decision process. Results showed that those in
the SDM approach felt they had received more information and better understood the
aspects of the decision when compared to those in the AF approach.

Moreover,

participants in the SDM condition reported higher satisfaction in terms of the decision
process’s efficacy in aiding them to make a decision that reflected their personal
concerns and values.

In another study, Arvai and Gregory (2003) compared different approaches to facilitate
the integration of non experts into complex, technical decisions regarding radiation
cleanup at contaminated sites.

While both approaches presented participants with

factual, scientific information, they were framed differently: one used a value-based
SDM approach while the other relied on a more common, technical presentation.
Although results demonstrated that both ways resulted in higher acquisition of decisionrelated knowledge, participants in the value-based approach stated that the information
was more constructive in aiding them to make a decision based on their own values.
Moreover, participants in the value-based approach demonstrated a reliance on
affective responses to a lesser degree than those in the more standard approach.

2.5 SDM in developing countries
While SDM approaches have been utilized extensively in developed countries, it has

	
  

46	
  

rarely been applied in developing countries. This is surprising considering that many
risk management or, more generally, many participatory decision processes in
developing countries are ideally suited for the application of SDM. There are many
pressing issues that involve multiple stakeholders, have multiple conflicting objectives
and difficult tradeoffs, and require clearer thinking and communication about potential
decision alternatives. Moreover, with many development projects being undertaken by
international agencies, some assert that the suitability of these projects would be
enhanced by including the interests and concerns of the local population in the project
assessments (Albert et al. 2010a; Reed 2008).

Perhaps, the lack of SDM implementation in these areas is due to the many constraints
facing SDM facilitators. These include a lack of technical support systems, a lack of
trained facilitators, time constraints, and cultural differences, among other reasons.
However, these challenges do not weaken the assertion that international development
decision making processes may be enhanced if insights from the decision sciences are
incorporated. Namely, the structuring of preferences, tradeoff analysis and weighting of
objectives offer particularly useful insights to help decision makers better explore their
decision alternatives.

To the best of my knowledge, these factors are only discussed in an implicit manner in
development literature and are in need of a more explicit focus. Therefore, the next
chapter in this thesis will report the results of a pilot study aimed at adapting and
applying SDM to the pressing risk management problem of point of use water treatment

	
  

47	
  

in the developing country of Tanzania.

	
  

48	
  

Chapter Three
Risk management in a developing country context: Structured decision making
for point-of-use water treatment in rural Tanzania

3.0 Introduction
More than one billion people—or one out of every eight worldwide—lack basic access to
clean water for domestic use ((Mintz et al. 2001; Unicef/Who 2008). The vast majority
of these people live in the developing world. In the East African nation of Tanzania, for
example, extreme water shortages are the norm for much of the rural population living
in the interior of the country. Despite the presence some of the world’s largest lakes
(e.g., Lake Victoria and Lake Tanganyika), this region of Sub-Saharan Africa receives
an average annual rainfall of less than 800 mm. As a result, people in this area—most
of them living in extreme poverty—typically obtain whatever water they can from
transient sources. These include seasonal ponds and streams, and in some extreme
cases, puddles.

Much of the water that is available for domestic use in this region of Tanzania is
contaminated with an array of viruses, bacteria, and protozoa. Associated with these
agents are water-borne diseases, including cholera, typhoid, shigellosis, and a range of
other diarrhea-causing illnesses. Seventeen percent of under-five mortality in Tanzania
can be attributed to diarrheal diseases. From a global perspective, more than 5,000
people die from diarrheal diseases linked to contaminated water daily, with the highest
fatality rate again observed among children under the age of five. To wit, diarrhea kills
more children than AIDS, malaria, and measles combined (Sobsey et al. 2008;

	
  

49	
  

UNICEF/WHO 2009) and is responsible for more deaths worldwide than all forms of
violence, including war (WHO 2002).

In addition to the unacceptably high mortality rate, the lack of readily available clean
water also comes at a significant cost to the fledgling economies and social structures
of developing countries. These costs have been linked mainly to the incidence of waterrelated illnesses, as sick people cannot contribute effectively to economic and social
growth, and the large amounts of time that people must spend looking for and hauling
clean water over long distances. In sum, water associated diseases affect poor people
in developing countries in a disproportionate way with extreme poverty linked to ill
health, and ill health leading to further impoverishment (Bloom & Canning 2001; WHO
1999).

For these reasons, finding and promoting effective and sustainable solutions for the
provision of reliable clean water in developing nations has become a focus of several
public health and international development efforts. One of the most effective ways to
ensure that people have access to clean water is to provide a reliable source of safe
water near communities (WHO 2004); this can be achieved by constructing a
combination of conveniently located wells, water tanks, and tapped pipes near
populated areas. But, as any development practitioner knows only too well, this is far
easier to state as an abstract goal than to achieve in the context of specific developing
communities with limited resources, the need to respect cultural traditions and local
customs, not to mention what are often significant political and institutional barriers.

	
  

50	
  

Even if this infrastructure can be provided, there’s still no guarantee that people in
developing communities will consume clean water. Recontamination of water between
the point of collection and the point of use is widespread (Wright et al. 2003); for
example, containers that people often use to transport water from a storage tank or
centrally located tap to their homes are often contaminated themselves, thereby
negating the benefits of an uncontaminated source.

Likewise, the homes that

encompass many rural villages are spread out over vast distances making it difficult for
everyone to have easy access to a centrally located clean water source. As a result,
many people still end up collecting water from whatever source, contaminated or not,
that is closest to them. For this reason, development agencies and sanitation experts
strongly advocate the use of point-of-use treatment systems alongside whatever source
of water people regularly use (Sobsey et al. 2008).

Point-of-Use (POU) water treatment systems—which rely upon physical, chemical and
biological processes—have been shown to effectively reduce the incidence of many
water-borne diseases (Arnold & Colford 2007; Clasen et al. 2007). But despite the
efficacy of these approaches, adoption rates of POU systems remain low in many parts
of the world. The reasons for this are manifold. On the one hand, many people simply
do not know that the water they routinely use is contaminated. On the other, there is
widespread uncertainty about the treatment methods that are available and how to
properly use them. There are also significant shortcomings in terms of the reliability of
distribution networks to reach communities with reliable and effective water treatment

	
  

51	
  

systems (which only serves to heighten the risks faced by many). And, importantly, the
POU systems that are often made available do not adequately address users’
preferences. For example, several POU systems meet health-related objectives but do
not address potential users’ preferences for other attributes such as convenience, odor,
and taste (Anderson et al. 2007; Clasen 2009).

We view these problems through the lens of the decision sciences (Arvai 2007;
Hammond et al. 1999; Kellon & Arvai 2011; Kleindorfer et al. 1993).

From this

perspective, many of the challenges encountered with respect to providing people with
suitable POU systems (and the required knowledge about their use) stem from the
absence of a comprehensive framework for involving affected stakeholders in the
process of decision making about water treatment.

Such a framework would help

people to clarify and articulate their risk-specific values and concerns; have a hand in
setting technical agendas aimed at characterizing the nature of both the risks they face
and the efficacy of the available alternatives; and be involved meaningfully in the
evaluation of the costs and benefits of competing risk management alternatives
(Gregory et al. 2001).

To this end, research reported here focused on the development of a deliberative risk
management framework for involving affected stakeholders in decisions about POU
water treatment systems. Previous studies of POU devices have focused on identifying
the appropriate price that people ought to be charged (Ashraf et al. 2011), the role
information in facilitating behavior change (Madajewicz et al. 2007), POU adoption rates

	
  

52	
  

(Albert et al. 2010b), and a ranking of systems deemed ready for widespread
distribution and adoption (Sobsey et al. 2008). However, no studies to our knowledge
have undertaken an up-front and systematic analysis of stakeholders’ values and
objectives about POU water treatment, and what these mean in terms of people’s
preferences across competing options.

The starting point for this research is our previous work on structured decision making
(SDM). The general goal of a SDM approach is to place the values and concerns of the
potentially affected individuals squarely in front of policy makers so that they lend
maximum insight to decisions that will be made about risk management options. A
typical SDM approach engages people in the following steps (Hammond et al. 1999;
Kellon & Arvai 2011):

1. Defining and clarifying the context for the impending decision;
2. Characterizing what matters to stakeholders in the form of clearly articulated
objectives;
3. Identifying a set of attractive alternatives that address stakeholders’ objectives;
4. Establishing the projected consequences of the alternatives; and
5. Directly confronting the value tradeoffs that arise when objectives conflict.

SDM has been used extensively in a variety of mostly western risk management
contexts (e.g., see Arvai & Gregory 2003; Gregory & Long 2009; McDaniels et al. 1999;
Wilson & McDaniels 2007). However, few cases have focused on risk management in

	
  

53	
  

developing countries.

The reasons for this are understandable.

In rural areas,

participants often have to travel great distances—often on foot—in order to take part in
SDM efforts, and have only very limited time that they can devote to multi-party
initiatives. From the standpoint of SDM facilitators, other obstacles exist. Among them,
there is general lack of facilities where people can interact with increasingly common
computer-based decision support tools; in the case we outline below, even something
as simple as a flip chart was impossible to come by.

At the same time, political,

cultural, and language barriers between facilitators, policy makers, and local participants
can further hobble the best intentions of researchers and practitioners.

Despite these challenges, however, many of the risk management problems faced by
communities in developing countries are ideally suited to the application of SDM.
People are faced with many pressing problems that require quick and clear solutions.
And, there is pressure from aid and donor agencies to obtain input about alternatives
from multiple stakeholders, and to confront the tradeoffs that arise as a result of
conflicting objectives. With this as background, we report the results from research
aimed at developing and testing a SDM framework for rapid deployment in an
international development context. Thus, the guiding objectives were to develop an
SDM approach to fit the context of a developing country with the added constraint of
time pressure, and to use the adapted SDM approach to help people identify a water
treatment system (or suite of systems) that stood the best chance of seeing daily use in
rural households while also ensuring that the potential water systems also proved
effective with the local water supply.

	
  

54	
  

3.1 Methods
3.1.1 Context: POU Water Treatment
We worked under the auspices of the Center for the Advanced Study of International
Development at Michigan State University, which is overseeing a multi-year and private
donor-funded development effort in Tanzania. An important element of this effort was
addressing health risks by ensuring that people would have sustainable access to clean
water at the household level. According to our research partners in Tanzania, fewer
than 10% of households in rural Tanzania disinfect their water prior to using it for
drinking or cooking. Thus, the focus of our work was on helping people living in rural
communities to make decisions about POU water treatment systems.

In terms of the practical implications of our research, we had two fundamental
objectives in mind. First, we sought to help people identify a water treatment system (or
suite of systems) that stood the best chance of seeing daily use in rural households.
Second, we had to ensure that the water systems that we were working with proved
effective with the local water supply.

In consultation with both our in-country partners and several recognized experts in the
areas of international development and microbiology, we identified five alternative POU

	
  

55	
  

systems that were both widely available and technically feasible (in that they did not
require electricity or batteries to operate3) in Tanzania.

The first of these was boiling, which relies on prolonged exposure to heat to neutralize
bacteria, viruses, and parasites. Unlike the situation in the developed world, boiling is
not an easy or straightforward process among the rural poor in Tanzania. It requires
first collecting firewood or charcoal (which may be made or purchased). This process
alone can take an individual, usually the woman in the household, several hours. After
next building and then maintaining a fire, which also may require hours, a family can
obtain approximately four liters of boiled water in 30-60 minutes. (The few families that
possess a kerosene stove can do this in approximately 20 minutes.) According to our
contacts in Tanzania, it could take as many as six hours to obtain four liters of
disinfected water by boiling, including the time it takes to collect the wood, build and
manage the fire, and boil the water.

The second method, termed solar water disinfection (SODIS) involves first placing
collected water (with a turbidity of less than 30 NTU4) in a clean, transparent 1-3 liter
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
3

Though electronic devices utilizing solar battery chargers (e.g.,
mobile phones with solar panels affixed to the back of the device
for charging the internal battery) are gaining prominence
throughout Tanzania—including the small hamlets and villages in
inhabited primarily by the rural poor—a reliable solar-powered
POU water treatment system is not yet widely available.
4
Nephelometric Turbidity Units; water that has a turbidity of >30
NTU must first be filtered prior to use with the SODIS method.
	
  

56	
  

PET water bottle. Next, the capped bottles are placed in full sunlight for 10 consecutive
hours.

On days with >50% cloud cover, the bottles must be left outside for two

consecutive days. This method is not effective during periods of rain. Using SODIS,
bacteria, viruses, and parasites are neutralized by UV-A radiation present in sunlight.
After the requisite time, water should be consumed directly from the bottles. When
empty, the bottles must be cleaned with soap prior to reuse.

Method three involves using a 1% sodium hypochlorite solution (dilute bleach branded
locally as WaterGuard), which at this concentration is effective at neutralizing bacteria
and most viruses.

However, it is not effective at inactivating certain protozoa such as

cryptosporidium. To use WaterGuard, an individual simply adds one standard capful
(approximately 8 mL) of the WaterGuard solution to 20 liters of water. WaterGuard may
also be used with turbid water, however two capfuls of the solution must be used. In
both cases, the water must be stirred for approximately five minutes and then allowed to
rest for 30 minutes before it is consumed.

Similar to WaterGuard, the fourth (and newest in Tanzania) method involves using a
disinfectant branded as the PUR Sachet by its manufacturer, Procter & Gamble. Like
WaterGuard, this method also relies upon a time-release hypochlorite—Ca(ClO)2 in this
case—to deactivate microbes.

The PUR sachet also contains a flocculant, ferric

sulphate, which acts to remove suspended materials (through settling) from water. The
effect is of the flocculant is quite dramatic as it quickly renders the most turbid (i.e., dark
brown) water clear. About the size of a sugar packet, each sachet treats 10 litres of

	
  

57	
  

water. However, unlike WaterGuard, using the PUR sachet is more labor intensive.
One adds the powder from the PUR sachet to 10 litres of water in a mixing bucket and
stirs for five minutes so that it may fully dissolve.

Next, five additional minutes are

allowed for the flocculating agent to act in the suspended solids. The water is then
transferred to a second vessel while filtering it through a tightly woven cloth. Finally, 20
additional minutes are required prior the water being ready for consumption.

The final method involved using a large clay filter (approximately 40 cm in both diameter
and depth), which rests inside a larger collection receptacle fitted with a spigot. The
clay filter, manufactured in Tanzania, is made primarily of terra cotta that has been
coated with antimicrobial colloidal silver. Water is poured into the filter by the user and,
at a rate off approximately 2 litres per hour, moves through the small pores in the terra
cotta and into the collection receptacle. Filtered water can then be served via the spigot
on the collection receptacle. Water can be continuously added to the filter so that there
is always a supply of approximately 10 liters in the collection receptacle. If properly
cared for, each clay filter has a useable lifespan of approximately five years.

3.1.2 Study Locations and Participants
Our research was conducted near two small rural villages in Tanzania: Milola and
Naitolia.

Milola, which is located in the Lindi region of southeastern Tanzania, represents one of
the poorest areas in the country. Approximately 10,000 people live in Milola split across

	
  

58	
  

four sub-villages (termed Milola-A, Milola-B, Milola C, and Milola D by the Lindi District
Office) and a series of smaller hamlets. Our work concentrated on a small hamlet with
a few hundred5 inhabitants, which was located between Milola A and Milola B.

The primary source of water for domestic use in this area is a centrally located tap to
which water piped form a nearby natural spring (the Chipwapwa) may be obtained.
Alternatively, local residents may also collect water from a secondary source, the
nearby river (which local residents refer to as the Ninu River).

Naitolia, which is also very poor, is located is located in the Monduli District of northcentral Tanzania. This village consists of 245 households and a total population of
approximately 1,300 spread out over several square kilometers.

As in Milola, we

worked with residents of a small hamlet3 located near the center of Naitolia, which
comprised mainly of members from the Maasai ethnic group (the Maasai make up the
majority of the population in Naitolia). The primary source of water in Naitolia is a well
and adjacent water storage tank located approximately 8km from the village center
(which consists of nothing more than a small administrative office and classroom). The
secondary source of is a pond approximately 2km from the village center where the
water has been designated as for domestic use only.

	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
5

We were unable to obtain an accurate estimate of the population
for either study area from either the Lindi (Milola) or Monduli
(Naitolia) district offices.
	
  

59	
  

3.2 Structured Decision Making
A typical SDM process for POU water treatment in North America or Europe might take
several days; this would include time for several deliberative elements, including (i)
defining the decision context, including key stakeholders and constraints; (ii) several
rounds of eliciting objectives (including appropriate attributes and measures; i.e.,
measurement criteria) from key stakeholders and decision makers; (ii) tests of water
samples obtained from each POU system by both stakeholders and experts to
determine how well each performs across each of the stated objectives; and (iv) formal
tradeoff analysis aimed at informing either a rank order of options or a decision to
implement a single alternative.

In Tanzania, however, we were faced with the constraint that we would only have 3.5
hours to conduct each individual SDM workshop. As a result, a third objective of our
research focused on the development of an SDM approach that could be implemented
in a developing country context—with all of the additional challenges that this kind of
work introduces—under significant time pressure.

All three of the workshops at a given site (Milola and Naitolia) were conducted over two
consecutive days. On day one at each site, we conducted two workshops with the two
groups of five women only (one in the morning and one in the afternoon). On day two,
we worked in the morning with five members of the village water committees. The
afternoon was reserved for a discussion of the workshop results with the village water
committees. There was a primary focus on women because women are responsible for

	
  

60	
  

gathering and using household water. Village Water Committee members were also
emphasized due to their more in-depth knowledge of the local water supply and system
(in comparison to other villagers).

So as to use our limited time wisely, each workshop followed the same basic protocol.
Each workshop began with a 30-minute introductory section, where we introduced
ourselves as well as the members of our team (our research assistant and two paid
translators from the National University in Dar es Salaam); we also asked that
participants introduce themselves at this time.

Following these introductions, we

provided a description of the nature of our work and our objectives for the workshop.
We also obtained informed consent from each participant.

Next, we provided an

overview of the health risks associated with untreated water, as well as the expected
health benefits of using POU water treatment devices. This aspect of the introductory
session was prepared in advance and was developed with insights from experts in
human health (a registered nurse from Michigan State University who accompanied us
to Tanzania) and microbiology (based at Michigan State University and the Centers for
Disease Control and Prevention in Atlanta, GA).

Finally, we introduced workshop

participants to the concept of SDM. To make the concept of SDM salient for them, we
used an example—decisions about daily activities—to guide our discussion.
Participants were encouraged to ask questions at any time during the introductory
section, and at all times during the remainder of the workshop.

	
  

61	
  

Immediately following the introductions, the focus of the workshops turned to eliciting
objectives about water for domestic use and POU water treatment from each workshop
participant. This process took approximately 45 minutes. Because of the need to be
efficient, and because of the low levels of education among the workshop participants,
we used boiling (with will all participants were familiar) as a reference point in the
discussion of objectives. Time was taken during this phase of the workshop so that
each participant could articulate their objectives concerns, and to separate means from
ends objectives. As part of this session, time was also taken to identify locally relevant
measurement criteria associated with each objective. Workshop participants set their
own measurement criteria for each objective with the exception of one: safe water (see
Section 3.4, below).

It was decided that constructed scales were the best way to

measure objectives. A 0-5 scale, with 5 indicating the highest satisfaction, was used for
all.

Participants provided context for the scales (e.g. with taste, a measure of 0

indicated very bad or “medicinal” taste; a measure of 5 indicated very good, or “cold”
and “more sweet” taste).

During the discussion of objectives, the methodology in Milola and Naitolia differed
slightly. In Milola, the workshop participants had not worked as frequently with outside
researchers or development practitioners. As a result, the nature of this stage of the
workshop—where people were encouraged to talk about what was important to them
vs. to researchers and practitioners—was quite foreign to them. Therefore, to make this
part of the workshop easier, fun, and more intuitive, we asked participants in Milola to
first draw pictures on paper brought to the workshops that characterized their objectives

	
  

62	
  

and concerns about water and water treatment; these were then discussed by the
group. We also used pictures to represent objectives in Milola because many of the
participants could not read; as a result pictorial symbols understood by the group were
use to characterize objectives during the tradeoff analysis and discussion periods (see
below).

In Naitolia, by contrast, participants have worked more frequently in the past with
outside researchers on a variety of issues (e.g., agricultural development, primary
education, and emergency medical interventions). As a result, a more straightforward
discussion of objectives and measurement criteria took place at this site. Moreover,
Naitolia participants specifically requested that more time be spent in the interactive
portions (see below) of the workshops so they could gain a more thorough technical
understanding of how each system works.

After discussing objectives in both sites, we moved into 60-minute interactive
demonstration of each of the POU methods outlined above.

First, the authors

demonstrated each of the five POU methods; this was followed immediately by the
participants given the opportunity to test each of the POU methods themselves6. This
interactive session was designed to be quite lively, with open discussion among the
facilitators, participants, and translators. It also enabled participants to add to their list
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
6

Because time was limited, the demonstration of boiling was
undertaken using a portable kerosene stove. However, the
demonstration was discussed in terms of the prominent local
context, which involves boiling over a fire.
	
  

63	
  

of previously objectives, now that they had more context for the potential alternatives for
water treatment and better understood the similarities and differences among them.

Next, workshop participants in both Milola and Naitolia evaluated post-treatment
samples (see Section 2.4, below) from each of the available POU systems in terms of
their ability to meet their stated objectives of taste, odor and sight (Figure 1). Each
sample was scored on a scale of 0 to 5, with participants placing the desired number of
tokens in cups (Milola) or by showing a corresponding number of fingers (Naitolia).
Then, non-sensory objectives were scored.

When the scoring across all of the

objectives was completed, the results across all of the participants were summarized in
a consequence matrix. This process lasted for approximately 30 minutes.

Following evaluation of post-treatment samples, we undertook a 45-minute discussion
of tradeoffs across the different POU systems. Because we worried that a discussion of
tradeoffs focused on even swaps (Hammond et al. 1999) or swing weighting (Clemen
1996) would have been too complex7, we approached a discussion of tradeoffs in a
different lexicographic way.

In Milola, all drawn objectives were pinned to the wall of the workshop. Participants
then discussed the consequences of all objectives pertaining to each different POU
system. For example, extra materials needed to run each system (e.g. two buckets are
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
7

Our previous experience in North America suggests that people
often need a length description of these methods, as well as time
to practice them, prior to use in a specific decision context.
	
  

64	
  

needed to use PUR) was one objective. After using each system during the interactive
session and gaining an understanding of the material needed, it was decided that all
materials were readily available in the local area and not too expensive. Therefore, this
objective was removed from the wall – physically – to demonstrate that it now had
lesser importance than other objectives remaining.

The discussion continued and

objectives were removed until only 5 remained (taste, odor, time, ease of use, and
efficacy - see Table 1) that would receive explicit ranks regarding how each system
performed in terms of that objective.

In Naitolia, because pictures were not used, participants discussed the available POU
systems on an attribute-by-attribute basis and pointed out the tradeoffs they were willing
to make and those they were not. This was done until the participants settled on a
preferred option.

Finally, it is important to note that during all workshops we were discouraged from using
recording devices - audio or video - for fear that they would be distracting to
participants. Instead, our research assistant working with a translator took meticulous
notes at all times during the workshop. These notes, as well as all of the materials
gathered from participants at the close of the workshop (e.g., sketches and the
consequence matrix) were cross-referenced with the notes of the two facilitators and the
second translator after the completion of each workshop.

	
  

65	
  

Figure 1. The evaluation of water samples by workshop participants in Naitolia.

3.3 Water Quality
A key element of our research was ensuring that the POU use systems that people in
Milola and Naitolia would choose among were effective in terms of disinfecting water
from commonly used sources. However, as was the case with our workshops, our time
in the field was limited so thorough testing of samples in a European or US-based lab
was not an option8. For this reason, we elected to use IDEXX Colilert-18 test kits
(IDEXX Laboratories, Westbrook, ME) for water samples collected at source in Milola
and Naitolia. These kits are effective for detecting both E. coli and total coliforms and,
importantly, conform to the US EPA’s rule regarding standard methods for the
examination of water and wastewater.

	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
8

More thorough testing of water samples, to include assays for
protozoa and viruses, is planned for a future study.
	
  

66	
  

We collected water at each of the two primary sources in both Milola and Naitolia in the
morning on the day before the SDM workshops began. Enough untreated water from
each source was drawn for pre-treatment testing. The remaining water was divided and
treated using each of the available POU systems. We then used water treated by the
primary source in both site (tapped pipe in Milola and tank in Naitolia) to use in the SDM
workshops when participants evaluated the post-treatment samples (see Section 2.3,
above) because we were certain they were clean. We did not give out samples from
the river (Milola) or pond (Naitolia) because we were less confident in the POU systems’
ability to remove E. coli from those two sources due to the much higher levels of
turbidity. However, enough treated water from each system and source was retained
for analysis with the IDEXX Colilert-18 kits. As part of the water testing process, pre
and post-treatment samples—in replicates of three—were appropriately incubated with
a field incubation unit and assayed prior to and following treatment with each POU
device. Appropriate positive and negative controls, also replicated in triplicate, were
also assayed using the same method.

3.4 Results
	
  
3.4.1 Milola
Because the village water committee is responsible for decisions about bulk acquisitions
of POU treatment methods, we report only the results of the workshop with this group.
Also of note, only four of the five POU methods (boiling, WaterGuard, PUR Sachets,
and SODIS) were evaluated in Milola because, despite repeated attempts over four

	
  

67	
  

weeks leading up to the workshops, we could not get the manufacturers of the clay
filters to deliver them to the village.

Participants across the three workshops in Milola discussed a total of twelve objectives
for POU water treatment methods and the water that they could obtain from them (see
Figure 2 for an example of eight objectives drawn by Milola participants). Objectives
related to the POU methods themselves included ease of use, the price it would cost to
purchase the necessary materials (if necessary), the amount of time required to
disinfect the water, the amount (volume) of water that could be disinfected at a time, the
availability of the POU methods, the number of materials needed to use the systems
(e.g. two buckets are needed for PUR), and storage requirements for each system.
Objectives related to the water obtained from each of the POU methods included color
(i.e., clarity), health and safety (i.e., the efficacy of the POU method), odor (i.e., not
smoky, fetid, or smelling of “medicine”, which is how participants characterized the smell
of chlorine), and taste (i.e., not smoky or tasting like “medicine”).

We discussed with participants the option of collapsing color, odor, and taste (which are
each means objectives) into a single ends objective, which some participants referred to
as the “enjoyment” derived from drinking water that satisfies their thirst.

We also

discussed other ends objectives, such as “improved health”, but we elected to keep
these objectives separate because the majority of participants felt that the ability to
address tradeoffs across them individually was important in terms of being able to more
fully evaluate the different POU methods.

	
  

68	
  

A. Methodological Objectives

B. Post Treatment Objectives

!

!

Figure 2. Examples of sketches by workshop participants depicting objectives for both
POU methods (“methodological objectives”) and the water derived from them (“posttreatment objectives”). The sketches are self-explanatory, except perhaps for the
following: The sketch for “ease of use” is a pot with a lid because participants wished
the ease and simplicity of putting clean water into a pot or cup and using it. The sketch
for “price” is a cashew nut, which many people in Milola harvest and sell at market. The
sketch for “health and safety” depicts a family. The sketch for “taste” is a wrapped piece
of candy.
After further discussion, and mainly for the sake of keeping the number of evaluation
criteria manageable, workshop participants elected to focus their evaluation of the POU
methods on five objectives: Taste, odor, ease of use, the amount of time required to
disinfect the water, and health and safety (i.e., efficacy). For example, the amount of
water that could be treated at a time was excluded because participants felt that a quick
and easy method would allow them to disinfect more than enough water in a reasonable
	
  

69	
  

period of time. The clarity of the water was excluded because participants felt that this
could be easily dealt with, regardless of the POU method used, simply by first filtering
the untreated water through a piece of fabric. Also, price was excluded because the
agreement between the donor-funded development initiative (see Section 2.1) and the
district office was the POU method that was ultimately selected by participants would be
provided at little or no cost to individual households. Decisions about which objectives
to exclude were made prior to the evaluation of the POU methods to minimize bias.

Boiling

WaterGuard

PUR Sachet

SODIS

Rate Rank Rate Rank Rate Rank Rate Rank
Taste
Odor
Time
Ease of Use

3.9
4.7
3.3
4.4

3
1
3
1

4.6
4.4
4.2
3.8

1
2
1
2

4.4
4.3
4.0
1.6

Efficacy

Tap

River

Tap

River

Tap

Pre-Treatment
Post-Treatment

Pos
Neg

Pos
Neg

Pos
Neg

Pos
Neg

+
-

2
3
2
4
Rive
r
+
-

—
—
2.6
2.6

—
—
4
3

Tap

River

+
+

+
+

Table 1. Consequence matrix depicting participants’ mean ratings and rankings of POU
methods in Milola. Ratings were provided on a 0 – 5 scale, where 0 = the worst
possible performance and 5 = best possible performance on a given attribute. The
“efficacy” attribute shows the results from the pre and post-treatment assays for the
presence (+) or absence (-) of both E. coli and coliforms.
Table 1 depicts participants’ ratings, and the resulting rankings, for each POU method
on a 0 - 5 scale (where 0 = lowest rating and 5 = highest rating on a given attribute).
The authors used this table to analyze workshops results. Please note that because
SODIS was shown to ineffective in terms of disinfecting water from both sources (tap

	
  

70	
  

and river), we prevented participants from providing evaluations of taste or smell (the
latter out of concern that water may accidentally be ingested).

Moreover, SODIS

performed poorly in terms of both the time required to disinfect water (ranking fourth)
and ease of use (ranking third). As a result, workshop participants had removed SODIS
from further consideration.

The first analysis step taken was to replace the numbers with simple rankings (e.g.
shown as 1-4 in Table 1) and remove objectives with low weights. Odor was the first
objective removed. Thus was due to many Milola participants questioning whether or
not odor should be kept as one of the most important objectives. Few felt there were
great differences between the smell of water treated by each system, making it
relatively uninformative in the overall decision. Moreover, upon our closing discussions,
the great majority of participants had explained to us that if faced with having safe water
with a slight smell or drinking contaminated water with no smell, they would prefer the
safe water. Therefore, odor was removed as an objective on the consequence matrix.

Second, we looked for dominated alternatives.

The table clearly shows that both

WaterGuard and Pur received higher rankings than boiling on all objectives. Therefore,
boiling was removed. This left only WaterGuard and Pur. WaterGuard dominates Pur
on all attributes making it the dominating alternative and thus, the preferred option. This
matches concluding discussions with all participants who most often pinpointed
WaterGuard as the preferred method.

	
  

71	
  

However, it is important to note that WaterGuard was only effective with water from the
communal tap (from the river, only E. coli was neutralized by WaterGuard).

Thus, the

final recommendation by the village water council was for the adoption of WaterGuard
coupled with the implementation of a village-wide risk communication effort urging
people to boil water for domestic use obtained directly from the river.

3.4.2 Naitolia
The process followed in Naitolia was nearly identical to the one followed in Milola. Once
again, we report only the results of the workshop with the village water committee
because they are the body responsible for decisions about bulk acquisitions of POU
treatment methods.

As was the case in Milola, we could not effectively treat water from either source (tank
and pond) using SODIS; i.e., E. coli and coliforms were found in all post-treatment
samples using this method. As a result, SODIS was dropped from our analysis in
Naitolia. However, we were able to obtain the clay filters (by picking them up ourselves
from the manufacturer) and brought these to Naitolia for evaluation by participants.

Participants across the three workshops in Naitolia discussed a total of seven objectives
for POU water treatment methods. Objectives related to the POU methods themselves
included ease of use (which, in Naitolia, encompassed both the time it took to treat
water and the operation of the method itself), the volume of water that could be treated
per unit time (i.e., the amount of water that could be treated with a single use of the

	
  

72	
  

method), and participants’ judgments about the risks associated with using each
method (which was a qualitative attribute that differed by method, see below). For
volume and risk, participants provided qualitative evaluations of either “low” or “high”,
rather than 0 – 5 rankings. Objectives related to the water obtained from each of the
POU methods included taste, color, odor, and health and safety (i.e., the efficacy of the
POU method). Unlike the case in Milola, participants in Naitolia wished to evaluate the
POU methods using all seven objectives.

Ceramic filters performed poorly across most of these objectives, yielding treated water
that performed the poorest in terms of taste, color, and odor (Table 2). Likewise, these
filters performed poorly in terms of the volume of water that could be treated over time
and participants’ judgments about the health risks associated with this method. On this
latter point, participants were concerned that some people, mainly children or men in a
hurry, would simply scoop drinking water out of the top of the unit, where untreated
water had yet to pass through the ceramic filtration stage (rather than using the
relatively slow flowing spigot located at the bottom of the receptacle). So, despite the
fact that the filters were effective in terms of both ease of use (ranked first because all
one needs to do is poor water from a source into the top of the unit) and effectiveness
(in that the filters completely disinfected water from both sources), it was removed from
further consideration by participants.

	
  

73	
  

Boiling

WaterGuard

PUR Sachet

Ceramic Filter

Rate
Taste
Color
Odor
Ease of Use
Volume·Time-1
Perceived
Risk
Efficacy
Pre-Treatment
Post-Treatment

Rank

Rate

Rank

Rate

Rank

Rate

Rank

4.7
4.2
4.6
2.1
Low

1
2
1
4
2

3.9
4.1
4.0
2.6
High

2
3
2
2
1

3.6
4.5
3.8
2.4
High

3
1
3
3
1

3.1
4
3
2.7
Low

4
4
4
1
2

Low

1

Low

1

High

2

High

2

Tank
+
-

Pond
+
-

Tank
+
-

Pond
+
+

Tank
+
-

Pond
+
-

Tank
+
-

Pond
+
-

Table 2. Consequence matrix depicting participants’ mean ratings and rankings of POU
methods in Naitolia. Ratings were provided on a 0 – 5 scale, where 0 = the worst
possible performance and 5 = best possible performance on a given attribute. The
“efficacy” attribute shows the results from the pre and post-treatment assays for the
presence (+) or absence (-) of both E. coli and coliforms.
Boiling was effective in terms of completely disinfecting water from both sources. And,
unlike Milola, participants in Naitolia were pleased with both the taste and odor of
treated water (both ranked first). The risks associated with boiling were judged to be
negligible because all participants were accustomed to working with fire. However,
boiling performed poorly in terms of both the volume of water that could be treated (i.e.,
one small pot at a time) and ease of use. On this latter point, during the workshops
participants had pointed to the hours required to collect firewood and water, and the
added time of bringing water to a boil. As a result, boiling was also dropped from further
consideration.

	
  

74	
  

As was the case in Milola, this left only WaterGuard and the PUR sachet. Again, of
these two POU methods, only the PUR Sachet was 100% effective with water from both
the tank and the pond. However, participants felt that treated water derived from the
use of WaterGuard tasted and smelled better.

In addition, WaterGuard had low

perceived risk (due mainly to widespread familiarity with this method in Naitolia), was
judged to be easier to use, and could treat a large amount of water in a single, short
session. In terms of the PUR sachet, in workshops participants were once again leery
of the method by which it rendered even the most turbid water clear (PUR sachets were
ranked first in terms of water clarity). One participant had in fact expressed concern
that it might do something similar inside the body (referring to the flocculating properties
of ferric sulphate) when people drank water treated with this method. Despite our (and
the manufacturer’s) assurances to the contrary, the high perceived risk associated with
the PUR sachet remained in workshops.

Thus, it was removed from further

consideration at this time.

As a result, there was consensus among representatives of the village water committee
that WaterGuard should be adopted as the POU method of choice. We pointed out that
WaterGuard was not effective, according to our tests, at treating water obtained from
the pond; only E. coli was neutralized while coliforms remained. Nevertheless, the
judgment of the committee was to select WaterGuard and recommend that
development funds be used to pipe water from the holding tank to tap that would be
built closer to the village center. If this water was more readily available, community
members would be less likely to use pond water. In addition, we recommended a

	
  

75	
  

village-wide risk communication effort warning people of the need to boil water obtained
from the pond specifically.

3.5 Discussion
Two of the guiding objectives of our research were to help people identify a water
treatment system (or suite of systems) that stood the best chance of seeing daily use in
rural households while also ensuring that the potential water systems proved effective
with the local water supply.

At both of our study sites, Milola and Naitolia, the

application of the SDM process led to WaterGuard being selected as the method of
choice. Our tests (Tables 1 and 2) showed that this method was indeed effective with
water obtained from the primary source in both areas. However, WaterGuard was not
shown to be effective with water obtained from either secondary source (i.e., the river
near Milola or pond in near Naitolia).

As a result, we grappled with our own decision as facilitators of an international
development effort to recommend to the private donors a POU method that was not
100% effective considering all of the local water sources. In the end, however, we must
side with the participants in our Tanzanian workshops for three main reasons. First,
WaterGuard was shown to be effective with the primary water source at both study sites
during our study period. Second, our job as facilitators was aimed at helping local
participants to identify a POU method that was consistent with their own objectives and
the associated tradeoffs that were most appropriate to them (vs. the objectives and
associated tradeoffs as might be determined by outside researchers or development

	
  

76	
  

practitioners). The fact that participants could also point to their own experiences—e.g.,
as it relates to the low uptake rate for boiling water despite widespread
acknowledgement of previous recommendations by human health and development
practitioners for people to use this method—as justification for not selecting a method
that was 100% effective also served to further ease our minds. And third, workshop
participants left us with the strongest impression that WaterGuard would actually be
used. As we note above, very few people in Milola and Naitolia undertake POU water
treatment of any kind. As a result, a recommendation that leads to a POU method that
people are likely to actually adopt on a sustained basis despite some imperfections is, in
our view, an important step forward.

We stress, however, that the recommendation to adopt WaterGuard is far from set in
stone.

In both Milola and Naitolia, the decision by participants to recommend

WaterGuard was driven in large part by concerns about the risks associated with
alternative POU methods, namely the PUR Sachet. Based on our observations, the
reluctance to adopt the PUR Sachet was based on a general lack of familiarity with—
and, as a result, trust in—this particular method. However, workshop participants were
nearly unanimous in their agreement that the PUR sachet addressed a majority of
concerns associated with its use and produced water of very high quality (as defined by
their own objectives).

It’s our view, therefore, that an opportunity exists to slowly

familiarize communities with the PUR Sachet as a means of building future trust in, and
support for, this method.

	
  

77	
  

For example, alongside making WaterGuard widely available, it is our view that
community exposure to PUR Sachets should also be increased; this may be achieved
through a variety of means, including risk communications that focus upon this
particular POU method, offering samples to individuals and families who would like to
test the method, and facilitated demonstrations of how it may be used (including a
discussion of how the various elements of the method—particularly the flocculant, ferric
sulphate—work). Indeed, these strategies could be implemented with a focus on both
PUR Sachets and WaterGuard as part of an adaptive management (Gregory et al.
2006; McDaniels & Gregory 2004) framework aimed at providing additional insights to
potential users about the pros and cons of both methods.

The third objective of our research focused on the development of an SDM approach
that could be implemented in a developing country context with the added constraint of
time pressure. Though the amount of time available to conduct each workshop was
short—3.5 hours—our goal was to follow as closely as possible the same basic SDM
framework that has been the focus of other risk management initiatives (Arvai et al.
2001b; Gregory & Long 2009; Wilson & Arvai 2006a); this included time devoted to
defining the decision problem that was the focus of the SDM effort, eliciting and
clarifying objectives, discussing measurement criteria, evaluating alternatives, and
confronting the tradeoffs that choosing among the alternatives entailed.

We were very impressed by how effective the time-compressed SDM model turned out
to be; this was, in large measure, a result of the level of sophistication displayed by the

	
  

78	
  

participants in all of our workshops. When we arrived in Tanzania, we were warned by
our in-country partners that working with people in Milola and Naitolia would be difficult
and potentially tedious due to their low level of education and literacy. Likewise, we
were told to expect a largely passive group of participants that were more likely to tell us
what they thought we wanted to hear, rather than what mattered most to them. But,
even though the levels of literacy and education among participants was low, the level
of sophistication on display in all of our workshops was remarkably high; in point of fact,
we would characterize it as being on par with what we have experienced when working
in North America and Europe. With minimal prompting from us, workshop participants
in both Milola and Naitolia were able to articulate in amazing detail their views as they
related to each step in the SDM approach (e.g., about objectives, measurement criteria,
tradeoffs, etc.).

Moreover, all of the SDM workshops were incredibly lively with

participants openly discussing their thoughts and feelings and, at times, challenging one
another (and us as facilitators) on the basis of certain claims and arguments.

Critical in terms of ensuring that our workshops would unfold in this way was the
interactive nature of the SDM workshops. In the weeks prior to our arrival in Tanzania,
we discussed the need to contextualize both the POU water treatment and decision
support initiatives for workshop participants so that they could contribute effectively and,
importantly, make their values clear to our project leaders. The analogy we drew upon
in this regard was a combination of a western structured decision making process and
Julia Child’s The French Chef.

On the one hand, we wanted to be faithful to the

required decision support elements identified in previous SDM efforts and then

	
  

79	
  

implement these in a workshop setting. On the other, we wanted to give participants
the opportunity to observe and interact with each of the POU methods in real time to the
point of being able to see—and taste and smell—the water provided by each method.
As we note above, this required us to prepare many of the water samples in advance
(using local water treated by each of the POU methods) so that they would be on hand
for the evaluation phase of the workshop.

Indeed, the interactive nature of the SDM workshops was, in our view, essential in
terms of providing people with the necessary context for our discussion of objectives
and, later, the exploration of tradeoffs. For example, the ability to interact with each of
the POU water treatment methods allowed participants to reflect upon objectives that,
otherwise, may have been difficult to contextualize. An illustration of this point were the
objectives related to time, which is typically not conceptualized by looking at a clock;
instead, time is judged relative to other activities (e.g., allowing water to boil for the
amount of time it takes to collect an additional armful of wood for a fire). Another
example relates to participants being able to see each of the methods in action; e.g.,
concerns about the PUR Sachet would likely not have come to light if participants could
not observe in real time the effect of ferric sulphate on a turbid water sample. And,
clearly, being able to relate their observations to objectives—while having this
information fresh in their minds—helped during the discussion of tradeoffs when
comparing the different POU methods.

	
  

80	
  

From an international development standpoint, decades of work by researchers and
practitioners have focused on improving quality of life for people living in developing
regions of the world. Much of this work has encompassed projects that hold as core
objectives the need to facilitate more democratic and participatory models of decision
making and governance, while also enhancing human health through the provision of
sustainable infrastructure. Along these lines, several participants reported to us during
the workshops that our visit to Naitolia was the first time that western researchers had
taken the time to discuss with them in detail their objectives and concerns, and how
these inform their preferences, in any community development context. It was reported
to us that the norm is for researchers to visit, speak briefly with village elders or highranking officials from the District Office, and then return with recommendations that to
many people seem disconnected from local realities. This was a very positive result in
terms of the ability of our work to address core development goals.

In sum, it’s

emblematic of how the nature of SDM can be married with previously applied
approaches in international development to better understand the preferences of local
citizens not only for water quality initiatives, but for many more international
development efforts.

	
  

81	
  

REFERENCES

	
  

82	
  

REFERENCES
Adamowicz, W., Jordan Louviere, and Joffre Swait. 1998. Introduction to AttributeBased Stated Choice Methods. Advanis, Edmonton, Alberta.
Agrawal, A., and C. C. Gibson. 1999. Enchantment and Disenchantment: The Role of
Community in Natural Resource Conservation. World Development 27:629-649.
Albert, J., Jill Luoto, and David Levine. 2010a. End-User Preferences for and
Performance of Competing POU Water Treatment Technologies among the
Rural Poor of Kenya. Environmental Science & Technology 44:4426-4432.
Albert, J., J. Luoto, and D. Levine. 2010b. End-User Preferences for and Performance
of Competing POU Water Treatment Technologies among the Rural Poor of
Kenya. Environmental Science & Technology 44:4426-4432.
Anderson, B., J. Romani, H. Phillips, M. Wentzel, and K. Tlabela. 2007. Exploring
environmental perceptions, behaviors and awareness: water and water pollution
in South Africa. Population & Environment 28:133-161.
Arnold, B., and J. Colford. 2007. Treating water with chlorine at point- of-use to improve
water quality and reduce child diarrhea in developing countries: A systematic
review and meta-analysis. American Journal of Tropical Medicine and Hygiene
76:354.
Arnstein, S. R. 1969. A Ladder of Citizen Participation. JAIP 35:216-224.

	
  

83	
  

Arvai, J., R. Gregory, and T. L. McDaniels. 2001a. Testing a Structured Decision
Approach: Value-Focused Thinking for Deliberative Risk Communication. Risk
Analysis 21:221-231.
Arvai, J., R. Gregory, D. Ohlson, B. Blackwell, and R. Gray. 2006. Letdowns, Wake-Up
Calls, and Constructed Preferences: People's Responses to Fuel and Wildfire
Risks. Journal of Forestry 104:173-182.
Arvai, J. L. 2007. Rethinking risk communication: Lessons from the decision sciences.
Tree Genetics and Genomes 3:173-185.
Arvai, J. L., and R. Gregory. 2003. Testing alternative decision approaches for
identifying cleanup priorities at contaminated sites. Environmental Science &
Technology 37:1469-1476.
Arvai, J. L., R. Gregory, and T. McDaniels. 2001b. Testing a structured decision
approach: Value-focused thinking for deliberative risk communication. Risk
Analysis 21:1065-1076.
Ashraf, N., J. Berry, and J. Shapiro. 2011. Can higher prices stimulate product use?
Evidence from a field experiment in Zambia. American Economic Review In
press.
Asian Forest Network. 2002. Participatory Rural Appraisal for Community Forest
Management: Tools and Techniques. Pages 1-24. Asian Forest Network, Santa
Barbara, California.
Beebe, J. 1995. Basic Concepts and Techniques of Rapid Appraisal. Human
Organization 54:42-51.
Bloom, D. E., and D. Canning. 2001. Demographic change and economic growth: The
role of cumulative causality. in N. Birdsall, A. C. Kelley, and S. Sinding, editors.
Population Does Matter: Demography, Growth, and Poverty in the Developing
World. Oxford University Press, New York, NY.
Bohnenblust, H., and P. Slovic. 1998. Integrating technical analysis and public values in
risk-based decision making. Reliability Engineering and System Safety 59:151159.
Bradlow, D. D. 2001. The World Commission on Dams' Contribution to the Broader
Debate on Development Decision-Making. American University International Law
Review 16:1531-1572.
Brown, T. C. 2003. Introduction to Stated Preference Methods. Pages 99-110 in K. J. B.
a. T. C. B. Patricia A. Champ, editor. The Economics of Non-Market Goods and

	
  

84	
  

Resources, A Primer on Non-Market Valuation. Kluwer Academic Publishers,
Dorddrecht/Boston/London.
Chambers, R. 1981. Rapid rural appraisal: rationale and repertoire. Public
Administration and Development 1:95-106.
Chambers, R. 1993. Project Selection for Poverty-focused Rural Development: Simple
is Optimal. Pages 27-39. Challenging the Professions: Frontiers for rural
development
Chambers, R. 1994a. Participatory Rural Appraisal (PRA): Analysis of Experience.
World Development 22:1253-1268.
Chambers, R. 1994b. The Origins and Practice of Participatory Rural Appraisal. World
Development 22:953-969.
Chambers, R. 1997. Whose Reality Counts?: putting the first last. Intermediate
Technology Publications (ITP) London (United Kingdom).
Chambers, R. 2003. Notes for Participants in PRA-PLA Familiarisation Workshops in
2003 in P. R. Center, editor. Institute of Development Studies, Brighton, United
Kingdom.
Chess, C., and K. Purcell. 1999. Public Participation and the Environment: Do We Know
What Works? Environmental Science & Technology 33:2685-2692.
Clasen, T. 2009. Scaling up Household Water Treatment among Low-Income
Populations. World Health Organization, Geneva, Switzerland.
Clasen, T., W. P. Schmidt, T. Rabie, I. Roberts, and S. Cairncross. 2007. Interventions
to improve water quality for preventing diarrhoea: Systematic review and metaanalysis. BMJ 334:782.
Cleaver, F. 2001. Institutions, Agency and the Limitations of Participatory Approaches to
Development. Pages 36-55 in B. C. a. U. Kothari, editor. Participation: The New
Tyrrany? Zed Books, London.
Clemen, R. T. 1996. Making Hard Decisions: An Introduction to Decision Analysis.
PWS-Kent Publishing Co., Boston, MA.
Cooke, B., and U. Kothari 2001. Participation: The New Tyranny? Zed Books, London,
New York.
Cornwall, A. 1992. Body mapping in health RRA/PRA. Pages 69-76. RRA Notes. IIED,
London.

	
  

85	
  

Damasio, A. R. 1994. Descartes' Error: Emotion, Reason, and the Human Brain. G.P.
Putnam's Sons, New York.
Dogbe, T. 1996. The one who rides the donkey does not know the ground is hot. Pages
34-36. PLA Notes. IIED, London.
Drinkwater, M. 1993 Sorting fact from opinion: the use of a direct matrix to evaluate
finger millet varieties. Pages 24-28. PLA Notes 1988-2001. IIED, London.
Ebrahim, A., and S. Herz. 2005. A Call for Participatory Decision-Making: Discussion
Paper on World Bank-Civil Society Engagement. Pages 1-204. World Bank-Civil
Society Joint Facilitation Committee.
Ecoagriculture Partners. 2009. Holistic Management of Rangelands in Dimbangombe,
Zimbabwe. Ecoagriculture Snapshots. Ecoagriculture Partners.
Ellis, F., and S. Biggs. 2001. Evolving Themes in Rural Development 1950s-2000s.
Development Policy Review 19:437-448.
Escobar, A. 1995. Encountering Development: The Making and Unmaking of the Third
World. Princeton University Press, Ewing, NJ.
Finucane, M., A. Alhakami, P. Slovic, and S. M. Johnson. 2000. The Affect Heuristic in
Judgments of Risks and Benefits. Journal of Behavioral Decision Making 13:117.
Francis, P. 2001. Participatory Development at the World Bank: The Primacy of
Process. Pages 72-87 in B. C. a. U. Kothari, editor. Participation: The New
Tyranny? Zed Books, London.
Frisch, D., and S. K. Jones. 1993. Assessing the Accuracy of Decisions. Theory and
Psychology 3:115-135.
Gintis, H. 2000. Beyond homo-economicus: evidence from experimental economics.
Ecological Economis 35:311-322.
Gregory, R. 2002b. Incorporatin Value Trade-offs into Community Based Environmental
Risk Decisions. Environmental Values 11:461-488.
Gregory, R., J. L. Arvai, and T. McDaniels. 2001. Value-focused thinking for
environmental risk consultations. Research in Social Problems and Public Policy
9:249-275.
Gregory, R., and R. L. Keeney. 1994. Creating Policy Alternatives Using Stakeholder
Values. Management Science 40:1035-1048.

	
  

86	
  

Gregory, R., S. Lichtenstein, and D. Macgregor. 1993. The Role of Past States in
Determining Reference Points for Policy Decisions. Organizational Behavior and
Human Decision Processes 55:195-206.
Gregory, R., and G. Long. 2009. Using structured decision making to help implement a
precautionary approach to endangered species management. Risk Analysis
29:518-532.
Gregory, R., T. McDaniels, and D. Fields. 2001b. Decision Aiding, Not Dispute
Resolution: Creating Insights through Structure Environmental Decisions. Journal
of Policy Analysis and Management 20:415-432.
Gregory, R., D. Ohlson, and J. L. Arvai. 2006. Deconstructing adaptive management:
Criteria for applications to environmental management. Ecological Applications
16:2411-2425.
Gregory, R. S., and R. L. Keeney. 2002. Making Smarter Environmental Management
Decisions. Journal of the American Water Resources Association 38:1601-1612.
Grimble, R., M.-K. Chan, J. Aglionby, and J. Quan. 1995. Trees and trade-offs: a
stakholder approach to natural resource management. Gatekeeper Series. IIED,
London.
Grimble, R., and K. Wellard. 1997. Stakeholder Methodologies in Natural Resource
Management: A Review of Principles, Context, Experiences and Opportunities.
Agricultural Systems 55:173-193.
Guggenheim, S. 2006. Crises and Contraditions: Understanding the Origins of a
Community Development Project in Indonesia in A. a. S. G. Bebbington, E. Olson
and M Woolcock. , editor. The Search for Empowerment: Social Capital as
Theory and Practice at the World Bank. Kumarian Press, West Hartford.
Hammond, J., R. L. Keeney, and H. Raiffa 1999. Smart Choices: A Practical Guide to
Making Better Decisions. Harvard Business School Press, Cambridge, MA.
Hickey, S., and G. Mohan. 2004. Towards Participation as Transformation: Critical
Themes and Challenges. Pages 3-24 in S. H. a. G. Mohan, editor. Participation:
From Tyrrany to Transformation? Exploring new approaches to participation in
development Zed Books, London.
Kahneman, D., P. Slovic, and A. Tversky 1982. Judgment Under Uncertainty: Heuristics
and Biases. Cambridge University Press, Cambridge, UK.
Kapoor, I. 2002. The Devil's in the Theory: A Critical Assessment of Robert Chambers'
Work on Participatory Development. Third World Quarterly 23:101-117.

	
  

87	
  

Keeney, R. 1992. Value-Focused Thinking: A Path to Creative Decisionmaking. Harvard
University Press, Cambridge.
Keeney, R., and H. Raiffa 1976. Decisions with Multiple Objectives. John Wiley and
Sons, Inc, New York.
Kellon, D. S., and J. Arvai. 2011. Five propositions for improving decision making about
the environment in developing communities: Insights from the decision sciences.
Journal of Environmental Management In Press.
Kerr, J., and N. K. Sanghi. 1992. Indigenous Soil and Water Conservation in India's
Semi-arid Tropics. Gatekeeper Series. International Institute for Environment and
Development, Sustainable Agrictulre Programme, London.
Kirsopp-Reed, K. 1994. A Review of PRA methods for livestock research and
development. Pages 11-36. RRA Notes. IIED, London.
Kleindorfer, P. R., H. C. Kunreuther, and P. J. H. Shoemaker 1993. Decision Sciences:
An Integrative Perspective. Cambridge University Press, New York, NY.
Leeuwis, C. 2000. Reconceptualizing Participation for Sustainable Rural Development:
Towards a Negotiation Approach. Development and Change 31:931-959.
Lewellen, T. C. 2003. The Anthropology of Globalization: Cultural Anthropology Enters
the 21st Century. Bergin and Garvey, Westport, CT.
Louviere, J. J., D. A. Hensher, and J. D. Swait 2000. Stated Choice Methods: Analysis
and Application. Cambridge University Press, Cambridge, UK.
Madajewicz, M., A. Pfaff, A. van Geen, J. Graziano, Hussein, H. Momotaj, R. Sylvi, and
H. Ahsan. 2007. Can information alone change behavior? Response to arsenic
contamination of groundwater in Bangladesh. Journal of Development
Economics 84:731–754.
Mansuri, G., and V. Rao. 2004. Community-Based and Driven Development: A Critical
Review. World Bank Research Observer 19:1-39.
Maquire, L. A., and C. Servheen. 1992. Integrating Biological and Sociological
Concerns in Endangered Species Management: Augmentation of Grizzly Bear
Populations. Conservation Biology 6:426-434.
March, J. 1978. Bounded rationality, ambiguity, and the engineering of choice. The Bell
Journal of Economics 9:587-608.

	
  

88	
  

Martinussen, J. 1997. Society, State and Market: A Guide to Competing Theories of
Development. Zed Books Ltd, London.
Mascarenhas, J., and P. D. P. Kumar. 1991. Participatory mapping and modelling users'
notes. Pages 9-20. RRA Notes. IIED, London.
Maxwell, S., and C. Bart. 1995. Beyond Ranking: exploring relative preferences in
P/RRA. Pages 28-35. PLA Notes. IIED, London.
Mayoux, L., and R. Chambers. 2005. Reversing the Paradigm: Quantification,
Participatory Methods and Pro-Poor Impact Assessment. Journal of International
Development 17:271-298.
McDaniels, T., and R. Gregory. 2004. Learning as an objective within a structured risk
management decision process. Environmental Science & Technology 38:19211926.
McDaniels, T., R. Gregory, and D. Fields. 1999. Democratizing risk management:
Successful public involvement in local water management decisions. Risk
Analysis 19:497-510.
Mintz, E., J. Bartram, and M. Wegelin. 2001. Not Just a Drop in the Bucket: Expanding
Access to Point-of-Use Water Treatment Systems. American Journal of Public
Health 91:1565-1570.
Mosse, D. 2001. 'People's Knowledge', Participation and Patronage: Operations and
Representations in Rural Development. Pages 16-35 in B. C. a. U. Kothari,
editor. Participation: The New Tyranny? Zed Books, London.
Motteux, N., T. Binns, E. Nel, and K. Rowentree. 1999. Empowerment for development:
Taking participatory appraisal further in rural South Africa. Development in
Practice 9:261-273.
National Research Council. 1996. Understanding Risk: Informing Decisions in a
Democratic Society. National Academy Press, Washington, D.C.
Neely, C. L., and J. Butterfield. 2004. Holistic Management of African Rangelands.
LEISA Magazine.
Payne, J. W., J. R. Bettman, and E. J. Johnson. 1992. Behavioral Decision Research: A
Constructive Processeing Perspective. Annual Review of Psychology 43:87-131.
Plous, S. 1993. The Psychology of Judgement and Decision Making. McGraw Hill, New
York.

	
  

89	
  

Porras, I. T., and R. A. Hope. 2005. Using Stated Choice Menthods in the design of
Payments for Environmental Services Schemes. IIED and University of New
Castle.
Pretty, J. N., I. Guijt, J. Thompson, and I. Scoones 1995. Participatory Learning and
Action: A Trainers Guide. IIED, London.
Quinn, C. H., M. Huby, H. Kiwasila, and J. C. Lovett. 2003. Local perceptions of risk to
livelihood in semi-arid Tanzania. Journal of Environmental Management 68:111119.
Reed, M. S. 2008. Stakeholder participation for environmental management: A literature
review. Biological Conservation 141:2417-2431.
Ribot, J. C. 2002. Democratic Decentralization of Natural Resources: Institutionalizing
Popular Participation. WRI, Washington.
Richards, C., K. L. Blackstock, and C. E. Carter. 2004. Practical Approaches to
Participation. Serg Policy Brief No. 1 Macauley Land Use Research Institute,
Aberdeen.

Rietbergen-McCraken, J., and D. Narayan. 1998. Participation and Social Assessment:
Tools and Techniques. The World Bank, Washington, D.C. .
Salmen, L. F. 1989. Beneficiary Assessment: Improving the Design and Implementation
of Development Projects. Evaluation Review 13:273-291.
Salmen, L. F. 1999. Beneficiary Assessment Manual For Social Funds. Social
Development Department, World Bank.
Salmen, L. F. 2002. Beneficiary Assessment, An Approach Described. Social
Development, World Bank.
Savory, A. 1991. Holistic resource management: a conceptual framework for ecologially
sound economic modelling. Ecological Economics 3:181-191.
Simon, H. 1987. Bounded Rationality. Pages 15-18 in J. Eatwell, Murray Milgate and
Peter Newman, editor. The New Palgrave: Utility and Probability. The Macmillan
Press Limited, New York.
Simon, H. 1990. Invariants of Human Behavior. Annual Review of Psychology 41:1-19.
Slovic, P. 1995. The Construction of Preference. American Psychologist 50:364-371.

	
  

90	
  

Slovic, P. 2000. Rational actors and rational fools: The influence of affect on judgement
and decision making. Roger Williams University Law Review 6:1630212.
Slovic, P., Ellen Peters, Melissa L. Finucane, and Donald G. MacGregor. 2005. Affect,
Risk, and Decision Making. Health Psychology 24:S35-S40.
Slovic, P., M. Finucance, E. Peters, and D. G. MacGregor. 2002. The Affect Heuristic.
Pages 397-420 in D. G. a. D. K. T. Gilovich, editor. Heuristics and Biases: The
Psychology of Intuitive Judgement. Cambridge University Press, New York.
Slovic, P., M. Finucane, E. Peters, and D. G. MacGregor. 2004. Risk as Analysis and
Risk as Feelings: Some Thoughts about Affect, Reason, Risk and Rationality.
Risk Analysis 24:1-12.
Slovic, P., B. Fischoff, and S. Lichtenstein. 1977. Behavioral Decision Theory. Annual
Review of Psychology 28:1-39.
Slovic, P., D. Griffin, and A. Tversky. 1990. Compatability Effects in Judgement and
Choice. Pages 5-27 in R. M. Hogarth, editor. Insights in Decision Makin: A
Tribute to Hillel J. Einhorn. University of Chicago Press, Chicago.

Sobsey, M. D., C. E. Stauber, L. M. Casanova, J. M. Brown, and M. A. Elliott. 2008.
Point of use household drinking water filtration: A practical, effective solution for
providing sustained access to safe drinking water in the developing world.
Environmental Science & Technology 42:4261-4267.
Stiglitz, J. E. 1989. Markets, Market Failures, and Development. The American
Economic Review 79:197-203.
Terluin, I. J. 2003. Differences in economic development in rural regoins of advanced
countires: an overview and critical analysis of theories. . Journal of Rural Studies
19:327-344.
Tippett, J., J. F. Handley, and J. Joe Ravetz. 2007. Meeting the challenges of
sustainable development - A conceptual appraisal for a new methodology for
participatory ecological planning. Progress in Planning 67:9-98.
Tversky, A., and D. Khaneman. 1981. The Framing of Decisions and the Psychology of
Choice. Science 211:453-458.
Unicef/Who. 2008. Progress on Drinking Water and Sanitation: Special Focus on
Sanitaiton in U. a. WHO, editor. World Health Organization and United Nations
Children's Fund Joint Monitoring Programme for Water Supply and Sanitation
(JMP), New York; Geneva.

	
  

91	
  

UNICEF/WHO. 2009. Diarrhoea: Why children are still dying and what can be done.
The United Nations Children's Fund and The World Health Organization, New
York, NY.
Van Tatenhove, J. P. M., and P. Leroy. 2003. Environment and Participation in a
Context of Political Modernisation. Environmental Values 12:155-174.
von Winterfeldt, D., and W. Edwards 1986. Decision Analysis and Behavioral Research.
Cambridge University Press, Cambridge.
Wilkins, L., P. Swatman, and T. Castleman. 2004. Faster, Richer, Better: Rapid
Appraisal Techniques for the Study of IS Implementation in Virtual Communities.
The Qualitative Report 9:161-175.
Wilson, C., and T. McDaniels. 2007. Structured decsion-making to link climate change
and sustainable development. Climate Policy 7:353-370.
Wilson, R. S., and J. L. Arvai. 2006a. Evaluating the quality of structured environmental
management decisions. Environmental Science and Technology 40:4831-4837.
Wilson, R. S., and J. L. Arvai. 2006b. When Less if More: How Affect Influences
Preferences When Comparing Low and High-risk Options. Journal of Risk
Research 9:165-178.
World Health Organization. 1999. World Health Report 1999: Making a Difference,
Geneva, Switzerland.
World Health Organization. 2002. World Health Report 2002: Reducing Risks,
Promoting Healthy Life. World Health Organization, Geneva, Switzerland.
World Health Organization. 2004. Meeting the MDG Drinking Water & Sanitation Target:
A Mid-Term Assessment of Progress. A Report of the WHO-UNICEF Joint
Monitoring Programme for Water Supply and Sanitation. World Health
Organization, Geneva Switzerland.
Wright, J., S. Gundry, and R. Conroy. 2003. Household drinking water in developing
countries: A systematic review of microbiological contamination between source
and point-of-use. Tropical Medicine and International Health 9:106–117.

	
  

92