EG Assignment 3, 4, and 5

Social Science & Medicine 64 (2007) 2427–2442 Behavioral indicators of household decision-making and demand for sanitation and potential gains from social marketing in Ghana Marion W. Jenkins a, , Beth Scott b aDepartment of Civil and Environmental Engineering, University of California Davis, One Shields Avenue, Davis, CA 95616, USA bLondon School of Hygiene and Tropical Medicine, UK Available online 17 April 2007 Abstract Household demand for improved sanitation in developing countries is an important social and behavioral process with implications for public health, sanitation policy and planning, and sanitation design and technology development. This paper develops a behavioral approach to assess household demand for improved sanitation in Ghana. Adoption decision stages of preference, intention, and choice to install a toilet in Ghana are deﬁned, measured in a survey, and used to estimate sanitation demand, identify factors affecting demand at each stage, and classify households by adoption stage to identify targeted demand-stimulation strategies. Results from a representative national sample of 536 households indicate that of 74% of households without any home sanitation, 31% have some likelihood of installing a toilet within the next year, but only 6% are very likely to do so; 62% had not considered the idea. Motivating and constraining factors are compared at each adoption stage and strategies likely to increase toilet installation in Ghana discussed. The approach is useful for assessing behavioral indicators of sanitation demand in developing countries and suggesting where marketing approaches can and cannot work to accelerate adoption of household sanitation improvements.

Keywords:Ghana; Household sanitation; Demand estimation; Adoption decision stages; Policy and planning; Marketing Introduction Good sanitation is a foundation for health that affords protection from a wide range of infections including diarrhea, a leading cause of child deaths, yet 2.6 billion people still do not have a safe means of excreta disposal at home (WHO & UNICEF, 2004).A target to halve this number by 2015 was added to the Millennium Development Goals in 2002. The enormity of the challenge, however, comes with the acknowledgement that public resources alone are unable to solve this global problem and new demand-oriented approaches are needed (Metha & Knapp, 2004;Water supply and Sanitation Colla- borative Council (WSSCC) & World Health Orga- nization (WHO), 2005).

Few large sanitation investment programs have been effective in increasing household sanitation in developing countries, yet people in these countries ARTICLE IN PRESS www.elsevier.com/locate/socscimed 0277-9536/$ - see front matterr2007 Elsevier Ltd. All rights reserved.

doi:10.1016/j.socscimed.2007.03.010 Corresponding author. Tel.: +1 530 754 6427; fax: +1 530 752 0782.

E-mail addresses:[email protected] (M.W. Jenkins), [email protected] (B. Scott). continue to install household toilets on their own without subsidy (Cairncross, 2004;Jenkins & Sugden, 2006). Increasingly, marketing techniques are advocated to stimulate household investment in sanitation at a larger scale, as a way to leverage household and community resources to sustainably close the sanitation gap (Cairncross, 2004; Mehta & Knapp, 2004). While a full discussion of the marketing approach is beyond the scope of this paper, at its core is a consumer focus which allows the development of an integrated mix of strategies designed to facilitate and achieve an exchange process leading to behavior change on the part of the target consumer (Kotler, Armstrong, Saunders, & Wong, 2002;Rothschild, 1999). Social marketing, the application of marketing to achieve behavior change for social good, is used in public health to achieve large-scale changes in health-related perso- nal and household behaviors (e.g., use of bed nets to prevent malaria, condoms to prevent HIV, house- hold disinfectant to treat drinking water) (Donovan & Henley, 2003). It has been argued among sanitation managers that a marketing approach will give households a greater choice to meet their individual sanitation needs, motivate the accompa- nying behavior changes for improved health, and stimulate development of a local sanitation industry to sustain improvements over time (Cairncross, 2004;SDC, 2004).

Within both commercial and social marketing, formative/consumer research, market analysis, audience segmentation, and targeted products, campaigns, and interventions to promote behavior change by reducing environmental constraints and increasing awareness and motivation are funda- mental components of program design (Donovan & Henley, 2003;Kotler et al., 2002). Public funding of the development of social marketing interventions is justiﬁed by the public health beneﬁts gained through the promoted behavior change. In the case of sanitation marketing, public ﬁnancing of promo- tional campaigns and product and supply-chain development would be warranted by the huge potential public health gains and improved environ- mental quality expected from increased safe defeca- tion and excreta management, while household payment for the latrine itself is justiﬁed through substantial private gains (Jenkins & Sugden, 2006).

A marketing approach to sanitation involving public sector and household level ﬁnancing is particularly attractive given the overall failure of supply-side sanitation investments to yield anysigniﬁcant sustainable impact over the past 25years, particularly at scale. However, marketing alone may not be enough to stimulate and support private behaviors to achieve public goals. With this wider perspective in mind,Rothschild (1999)developed a framework for the strategic use of marketing along with two other primary tools for behavior change— education and law—to achieve public social or health goals. The degree to which a target popula- tion is prone, resistant, or unable to adopt a new behavior derives from the presence or absence of three things: self-interest (motivation), opportunity, and ability to voluntarily adopt the new behavior (Rothschild, 1999). Marketing operates to enhance awareness of self-interests and create opportunities to act, and can sometimes overcome lack of ability.

When the target’s self-interests are not served regardless of opportunities or abilities, the law may be needed to gain compliance, particularly when large negative externalities from non-adoption exist as they typically do for sanitation.

Recognizing where and how marketing can affect household sanitation decisions is the ﬁrst of several challenges for sanitation managers wanting to apply commercial marketing techniques and approaches to increase demand for and access to improved sanitation. Where marketing is likely to be effective, a second challenge is to understand existing house- hold sanitation behaviors and adoption decisions in ways that inform development of speciﬁc cost- effective marketing strategies to increase adoption.

In this study, a model of household sanitation adoption decision-making that accounts for moti- vation, opportunity, and ability was developed and applied in Ghana to allow sanitation managers to better measure, understand, and predict behavioral determinants of demand for sanitation and strate- gically plan interventions based on marketing principles. The approach is broadly applicable where household sanitation coverage is low, but can also be adapted to areas where sanitation coverage might be high, but toilets (or latrines) unsafe or in poor condition. The model and survey tool serve three major purposes: Measure baseline household sanitation adoption and demand patterns. Predict changes in demand for sanitation im- provements by mapping the decision-making process into behavioral stages, classifying house- holds within this process, and understanding barriers to adoption at each stage. ARTICLE IN PRESS M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–2442 2428 Identify actions and policies to increase sanita- tion demand among households in different adoption stages.

A study of the decision to install a home toilet among households in Ghana illustrates how this assessment approach achieves these purposes. We label households who have already installed a toilet or latrine in Ghana ‘adopters’, while new demand is approximated by the portion of ‘non-adopter’ (without sanitation) households projected to pay for and build a new home toilet in the next 12 months. We describe our model of sanitation adoption stages and how the survey was conceptua- lized and applied nationally in Ghana. Then we present and discuss the survey results related to estimating new household demand for sanitation in Ghana. Finally, we examine the segmentation of households by adoption stage and how this in- formation can be used to design demand stimula- tion strategies based on marketing principles in the Ghanaian context.

Few studies have assessed consumer demand for sanitation in developing countries, and most of these have applied economic contingent valuation methods to approximate demand with willingness- to-pay bids (Altaf, 1994;Altaf & Hughes, 1994; Wittington et al., 1993). WTP estimates provide a measure of the hypothetical monetary value people place on what is often a narrowly prescribed sanitation change (crucial for bid accuracy) but are unable to provide time-bound predictions of demand. WTP studies typically ignore the transac- tion costs and constraints households encounter in real life and the tradeoffs they face when deciding to adopt and purchase new sanitation systems for the ﬁrst time in developing countries. Yet these features emerge as important determinants of household demand for sanitation in this and other studies (Jenkins 1999, 2004). Contingent valuation studies also offer limited insight into weak demand, apart from price and income, and minimal guidance on ways to stimulate demand and change behavior.

Applying decision-making models to estimate new sanitation demand draws from cognitive psychology and consumer purchase decision beha- vior to explain and predict changes in individual sanitation behavior over time by observing past and future sanitation adoption decisions, measuring their behavioral determinants, and mapping popu- lation into categories useful for understanding behavior change. Of particular interest are theindividual attitudinal and structural determinants of preferring and choosing different competing behavioral outcomes, rather than socio-economic characteristics, to explain and predict household demand. Although commonly applied in WTP and other surveys, socio-economic characteristics typi- cally lack explanatory power and provide poor predictors of individual changes in behavior for strategic planning and policy evaluation.

Sanitation change adoption in developing countries A model is developed of the household decision to adopt a sanitation change, focusing on the process, observable stages, and households’ logic behind each decision stage.

Adoption decision process For a household without adequate sanitation in a developing country, deciding to improve sanitation by installing a toilet (e.g., pit latrine, bucket latrine, ﬂush toilet, water closet, or another excreta disposal facility) for the ﬁrst time, changing to a new toilet system, or connecting to a sewer, can be a complicated and lengthy process. To ﬁrst contem- plate this decision, a household must be aware of the personal beneﬁts of the sanitation change and the availability of products and services. Consumer theory and empirical evidence suggest that with sufﬁciently strong interest, a household will actively seek information about options, perhaps discuss with family members and technical specialists how and when to make a change. An adoption plan might mean choosing a sanitation technology (or service level) to ﬁt the household’s budget and lifestyle, picking a site, ﬁnding a mason and supplier of construction materials, negotiating costs, saving money, and acquiring a building permit in some settings. Viewing the adoption decision as a progression of evolving attitudes, knowledge, and actions provides a more detailed causal under- standing of what generates demand for sanitation, and where sanitation coverage is low, factors that may constrain it.

Broadly speaking, non-adopter households can be categorized by whether or not they have thought about making a change to home sanitation, and if so, how far they have taken such thinking. While some may have considered a change, for example, installing a latrine, others will have little awareness of options or meaningful beneﬁts of having a latrine ARTICLE IN PRESS M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–24422429 and therefore never considered adopting. Among those who have considered installing a latrine, intention to actually build it will vary with priority given to the outcome and with the time frame and level of planning and preparations. Some may want a latrine but have ruled it out as unobtainable or ‘wishful’ thinking. Perceived lack of ability to control arises from personal context or resources, or from absence of local information and opportu- nities, making the choice to build improved sanita- tion unavailable or beyond reach.

Preference, intention, and choice stages Drawing from behavior change models and cognitive theories of individual decision-making, particularly the theories of reasoned action (Fishbein &Ajzen,1975), and planned behavior (Ajzen, 1985) and empirical study of household sanitation choices in Benin (Jenkins, 1999, 2004;Jenkins & Curtis, 2005), we develop a simpliﬁed model of three progressive stages of the decision to adopt a sanitation change called preference, intention, and choice.Fig. 1illustrates the key theorized determi- nants of progression at each stage.

Preference The adoption decision starts with development of ‘preference’ for a sanitation improvement over one’s present defecation practice. Preference captures purchase motivation and the expected relative advantages, beneﬁts, and reasons (perceived utility gain) for wanting a sanitation improvement. Moti- vation to change sanitation arises from dissatisfac- tion with current household defecation or excreta management practices coupled with sufﬁcientawareness of advantages of new options (Jenkins & Curtis, 2005). In this stage, households are interested in and have considered a sanitation change but have not necessarily begun to plan it.

Intention Households in the intention stage have begun to plan a sanitation change, but vary in their degree of plan development. For many behavior changes, once awareness and motivation are sufﬁcient, change is within the voluntary control of the individual or household. This is the case for example with hand washing, when soap, water, and technical knowledge usually exist within the household. However, changing sanitation infra- structure for the ﬁrst time requires more than awareness and motivation, including new and unfamiliar materials, products/services, technical knowledge, and skills, much of which must be acquired outside the home. These less controllable inputs often entail uncertainty and risks for house- holds without prior sanitation experience. Even do- it-yourself latrine construction usually requires special materials and new technical knowledge.

Starting a plan to change home sanitation technol- ogy or practice (‘intention’ inFig. 1) depends directly on the availability, quality, and cost of opportunities to acquire materials, products, con- struction services, ﬁnancing, skills, and knowledge, and on the personal resources, experience, and abilities of individual households to take advantage of these opportunities. Commitment to changing sanitation reﬂects the priority given to this change compared to expected outcomes from alter- native uses of limited savings, time, effort, and other household resources. If required abilities or ARTICLE IN PRESS Preference (Motivation) Intention Choice Awareness of sanitation options Dissatisfaction with current defecation practice Absence of temporary constraints to acquiring sanitation Absence of permanent constraints to acquiring sanitation Priority of change among competing goals No Intention No Preference Delay DETERMINANTS OF SANITATION ADOPTIONSTAGE 1:

PREFERENCESTAGE 2:

INTENTIONSTAGE 3:

CHOICE Fig. 1. Adoption decision stages and determinants of new sanitation demand.M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–2442 2430 opportunities are perceived to be lacking, or priority low, a household is unlikely to form an intention to adopt a sanitation change (Ajzen & Madden, 1986).

Choice Choice, the last stage of a successful adoption process, involves the individual’sactualability to use and control opportunities to carry out their intention to adopt (Ajzen, 1985). This stage carries a very high likelihood of adopting a sanitation change within a short time. Choice requires a well-devel- oped intention and concrete actions consistent with a strong imminent intention to change sanitation practice, including acquisition of relevant knowl- edge (e.g., cost), saving money, and site and toilet technology or service provider selection. Observed sanitation choices are the outcomes of this decision process and collectively generate new demand for a sanitation change (in general and for speciﬁc technologies, services, and products).

Role of constraints Perceived inabilities, inadequate resources, and lack of opportunities are different kinds of ‘con- straints’ to adoption (Fig. 1). In Benin, 13 con- straints were found related to construction problems, individual situations, and psychosocial factors that blocked the choice to build a home toilet (Jenkins, 1999, 2004). Similar constraints havebeen reported elsewhere (Jenkins & Sugden, 2006).

Increasing absence of perceived constraints, or ‘perceived behavioral control’ in the theory of planned behavior, similar in function and concept to self-efﬁcacy in the Health Beliefs Model (Ajzen, 2002), directly strengthens intention to change sanitation practice while actual behavioral control determines whether the intended behavioral choice is achieved (Ajzen & Madden, 1986).

Constraints vary in their effects on sanitation decisions. Those perceived as unchangeable (perma- nent) are thought to act early in the decision process to block progression from preference to intention, while constraints perceived as removable (tempor- ary) are expected to act later in the decision process to delay preparations and ﬁnal choice (Jenkins, 1999). The differential effects of constraints on adoption are explored in the Ghana study.

Adoption stage indicators, determinants and question formats Indicators were developed and questions con- structed in a survey format to measure the three decision stages and their determinants (Table 1).

Sanitation demand in Ghana was assessed by the number of non-adopter households at each stage of preference, intention, and choice, and new demand estimated by households in the last stage of ‘choice’ who expressed a ‘high’ likelihood of building a home toilet within 12 months. ARTICLE IN PRESS Table 1 Question formats and decision stage indicators and determinants for measuring sanitation demand Question formatsDecision stage indicatorDecision stage determinant 1 Have you considered installing a household toilet? ‘‘YES’’/’’NO’’ P 2 Reason stated for not considered: ‘‘SATISFIED WITH CURRENT PLACE’’ P 3 How satisﬁed are you with your current place of defection? ‘‘VERY SATISFIED, SATISFIED, UNSATISFIED, VERY UNSATISFIED’’P 4 What are the top three reasons for building a household toilet/latrine? P 5 For those who have considered, what is the likelihood that if I come back in a year you will have a latrine built? ‘‘HIGH, MEDIUM, LOW vs. NONE’’I 6 What are the three biggest constraints to your installing a household toilet/latrine? (absence of most permanent constraints, e.g., don’t have space, tenancy issues, poverty)I 7 Mention in answer to Q6: ‘‘COMPETING PRIORITIES’’ I 8 Mention in answer to Q6: ‘‘SATISFIED WITH CURRENT DEFECTION PLACE’’ P 9 Of those who have considered, likelihood in Q5 HIGH that if I come back in a year you will have a latrine builtC 10 What are the three biggest constraints to your installing a household toilet/latrine? (absence of permanent and most temporary constraints)C Note:P, preference; I, intention; and C, choice.M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–24422431 While the household is our unit of analysis, dynamics of intra-household decision-making are beyond this research. To capture a household’s adoption decision behavior in an interview format, the household member most responsible for making decisions about changes in home sanitation infra- structure should be interviewed. In our experience, this has consistently been the head of household in non-tenant households but is less clear who this should be in tenant households in developing countries where tenancy is often informal and takes many different forms under complex occu- pancy patterns (seeGilbert, 1983;Rakodi, 1995; UNCHS, 1996).

Materials and methods A questionnaire based on the indicators in Table 1was developed to achieve the following objectives:

1. measure baseline household sanitation coverage levels, 2. estimate new demand for household toilets among those without adequate sanitation, mea- sured by a high likelihood of installing a toilet within 12 months, 3. classify households by adoption stage to assist in designing marketing strategies, 4. examine predictors at each stage, related to awareness, dissatisfaction, motivations, and con- straints, and 5. be quick and easy to implement in developing countries.

Ghana questionnaire and survey execution Table 1shows the question formats used in the Ghana survey to measure the three adoption stages and their determinants. The questionnaire was divided into ﬁve sections:

1. Description of current defecation places and technologies, satisfaction with current situation, and most and least appreciated features of present defecation place.

2. Information related to ownership, age, decision- making, cost, and trigger reason for installed toilet asked of adopter households, identiﬁed from Section 1.

3. Expected beneﬁts of installing home sanitation, asked to all households.4. Perception of constraints and awareness of toilet technologies asked to households without home sanitation, consideration of home toilet installa- tion, and strength of intention within 12 months.

5. Socio-economic and demographic characteristics.

Thirty-four structured questions (15 on socio- demographics) were included in a larger baseline survey for the National Handwashing Promotion Program (www.globalhandwashing.org) under the Ghana Community Water and Sanitation Agency.

Coded responses were initially informed by a small qualitative study of reasons for and barriers to home toilet installation conducting using in-depth individual interviews in one town in the Ashanti Region (Obika, Jenkins, Curtis, Howard, & TREND, 2002). Question formats and coding were reﬁned following pre-testing of the survey in late July 2003 and ﬁeld work conducted from 29 August through 22 September 2003. Trained enumerators conducted one-on-one interviews under the man- agement of Research International, an international market research agency with ofﬁces in Accra.

Study site Ghana has a population of about 21 million.

Most urban and peri-urban households use public latrines while a majority of rural households defecate in the open. The wide-spread traditional use of public toilets and historic dependence on bucket (pan) latrine technologies for sanitation in Ghana is rather unique among developing countries in Africa and reﬂects a historical government policy of actively constructing, operating, and managing public toilets for household use and a conservancy labor system for bucket emptying. As a conse- quence, private investment in sanitation and house- hold coverage is relatively low and stagnant (less than 30% in urban areas and lower in rural areas) (Ayee & Crook, 2003;WHO & Unicef, 2006).

Public toilets are run mostly by local governments, charging a fee of about 1.25 cents US. They are often in very poor condition and lack privacy.

Public toilets are commonly non-water based and include cesspit, ventilated improved pits (VIP) and Kumasi VIP latrines (KVIP) designed as a double vault composting toilet, but rarely operated this way in public settings. Bucket latrines in public and private use, while previously common, are being phased out with the end of the conservancy labor system. In some cases, ﬂush toilets (water closets or ARTICLE IN PRESS M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–2442 2432 WCs) have been installed in public toilets but this is rare. Much confusion exists around the terms KVIP and VIP in Ghana. While technically distinct, they are commonly misunderstood as the same techno- logy in Ghana, with the term KVIP used for public toilets and VIP in domestic settings.

Sampling and sample description A sample of 536 rural and peri-urban households was selected following a sampling approach used by professional market researchers in Ghana for a representative national sample. A subset of regions, including Greater Accra, Ashanti, Eastern, Wes- tern, and Northern, was selected to represent the three socio-ecological zones of Ghana. Within each region, ten Census Enumeration Areas (EAs) were randomly selected from the list of all EAs. Within each EA, 10–11 households with children under 5 years were selected using the random route walk technique, for a total sample of 536 households.

Screening ensured that only households with a mother of young children were interviewed as this was the primary target audience of the National Handwashing Promotion Campaign.

Descriptive characteristics of the sample are reported inTable 2. Mothers had a mean age of 30, were mostly Christian, predominantly Akan, living in compound houses, and more than 50% were not educated beyond primary school. In most cases (86.4%), the father of the child lived with the mother. Less than one fourth of these men were educated beyond junior secondary school and a quarter of them worked in agriculture. Almost three quarters of all households earned 500,000 cedis/ month (about $55 US) or less, with one third earning less than 250,000 cedis/month ($39).

Data analysis Descriptive statistics were computed. Chi-squared values for the likelihood ratio were used to test for signiﬁcance differences in satisfaction levels, moti- vations, constraints, and other hypothesized deter- minants at each stage of the adoption model.

Results Existing household sanitation Adults in over half (58.2%) of the sample households used public toilets while 14% practicedopen defecation. Only one quarter (25.6%) can be described as household toiletadopters, possessing a private toilet either in their individual household (11.0%) or shared compound (14.6%). Those with compound toilets are included here as household adoptersas compound houses composed of multiple households represent a common living arrangement in Ghana; half of this sample lived in compound houses (Table 2). These ﬁgures are comparable to other estimates of sanitation coverage in Ghana ranging from 27% for urban to 11% for rural households with a private toilet (WHO & Unicef, ARTICLE IN PRESS Table 2 Sample socio-economic description Category CharacteristicN¼536 Sample description Respondent’s age Range 17–68 years Mean 30 years Region Greater Accra 21.9% Ashanti 27.3% Eastern 15.7% Western 16.3% Northern 18.7% Ethnicity Akan 53% Ga/Adangbe 17% Ewe 6% Mole-Dagbani 16% Other 8% Respondent’s educationNone 27.1% Primary 28.4% Junior secondary 31.9% Senior secondary 11.8% University 0.9% Occupancy Tenancy 34.7% Dwelling type Compound 50.8% One family house 26.2% Flat/apartment 3.4% Room(s) 20.2% Father’s occupationProfessional (white collar employment)20.1% Sales/commerce 12.5% Services 17.7% Trader 4.1% New trade 4.1% Agriculture 24.8% Unemployed 1.9% Father absent 14.6% Income /montho250k Cedis 34.7% 251–500 K 35.1% 501–900 K 20.0% 901 K+ 11.2% M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–24422433 2006). Children under 5 years predominantly used potties (82.0%), while the toilet habits of children above this age broadly reﬂected those of their parents, particularly in adopter households.

Among non-adopters, a higher percentage of older children (25.2%) than adults (14.0%) practiced open defecation.

Five household toilet technologies presently exist in Ghana—the ﬂush toilet (WC) (15%), bucket latrine (20%), traditional pit latrine (21%), and KVIP/VIP latrine (44%). Among the 137 household toilet adopters, two thirds were able to recall when their current toilets had been constructed. Of those recalling, less than 5% had built their toilet prior to 1990, with adoption accelerating in the late 1990s.

In over half of cases (56.2%), landlords (general- ized term for compound owner) were reported to have decided to build the household latrine (consistent with shared compound living), while household heads were the decision-maker in one in four adopter households. It was extremely rare (1.5%; 2 out of 137) for tenants to decide to install a toilet. Where household heads had taken the decision, the perception of ownership was some- times more broadly deﬁned. In over a third of these cases, women stated that it was the whole house- hold, not just the head, who ‘owned’ the toilet.

Although rarely making the decision themselves, nearly 80% of respondents were able to cite motives for constructing the facility. The most common reasons given were:

for sick or old relatives (23.2%), to offer safety at night (18.8%), for convenience (12.5%), to make it easier to keep the facility clean (9.8%).

Non-adopter households—satisfaction levels Non-adopters were 74.4% of sample households, reﬂecting the general population in Ghana. Among these, 65.2% were dissatisﬁed with their current place of defecation. The two most disliked attributes of current defecation places were that they were smelly (27.1%) and dirty (26.6%). Other cited dislikes included the distance to toilet facilities (8.3%), lack of comfort (7.0%), having to pay to use them (6.0%), and having to share with others (5.8%).

A third of non-adopters could cite no positive attributes for their place of defecation. Among those that could, the most liked attributes were that toilets were convenient (26.6%) and clean (17.8%).Dissatisfaction with one’s defecation facility was signiﬁcantly associated with being unable to cite any positive attributes, with citing privacy as the most positive attribute, and with disliking its dirty state (Table 3). Conversely, satisfaction was signiﬁcantly associated with valuing its cleanliness and safety.

Non-adopter households—motivating reasons for toilet installation All non-adopters were asked to give three top reasons for building a household toilet. The most heavily cited reasons were convenience (51.4%), that they are easy to keep clean (43.1%), good health (41.9%), and general cleanliness (27.8%).

Presently non-adopters travel to open defecation sites and public latrines, and in the latter case, join long queues in the morning or evening. Widely used in Ghana, public latrines in particular tend to be ARTICLE IN PRESS Table 3 Most and least liked attributes of current defecation place (non- adopter households) Satisﬁed N¼139 (34.8%) (% citing)Unsatisﬁed N¼260 (65.2%) (% citing)p-value Most valued attribute Nothing 7.2 47.7o0.01 Cleanliness 41.0 5.4o0.01 Privacy 2.2 6.9 0.03 Safety 5.0 1.5 0.05 Good health 7.9 3.8 0.10 Comfort 4.3 1.5 0.10 Convenience 28.8 25.4 ns Get to work/ school on time3.6 5.0 ns Least liked attribute Dirty 14.4 33.1o0.01 Pay to use 12.9 2.3o0.01 Distance 12.9 5.8 0.02 Share with others9.4 3.8 0.03 Hard to maintain0.7 3.1 0.10 Lack of comfort4.3 8.5 0.11 None 7.9 5.0 0.25 Smell 25.9 27.7 ns Queuing 5.0 3.5 ns Fills up quickly/difﬁcult to empty2.2 2.7 ns chi-squaredp-value for log likelihood ratio comparing ‘Satisﬁed’ with ‘Unsatisﬁed’. M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–2442 2434 dirty and squalid, with feces lying around squat holes which emit heat, gases, and bad odors, believed to cause ill health (Obika et al., 2002).

Indeed, while 41.9% of respondents cited good health as a key reason to build a household toilet, only one third said that germs were the cause of ill health, two thirds believing illness to be caused by heat, smell, feces or dirt. Thus toilets need to be clean to protect health, but further, in Ghana, people have a particular need or desire to be neat, clean, and not smell, reﬂecting not just physical but mental and moral purity (van der Geest, 1998, 2002). In his anthropological studies on this topic among the dominant Ghanaian ethnic group—the Akan,van der Geest (1998)reports a strong association between bowel elimination and personal moral cleanliness. The Akan language and culture uses a different much more negative word for ‘dirt’ that is inside the body, including feces, vomit, and other such matter, while concepts of cleanliness are tied closely to notions of beauty, respect, civility, pride, goodness, and purity (van der Geest, 2002). Our qualitative study conﬁrmed the impor- tance of eliminating ones bowels ﬁrst thing each morning so as to be ‘clean’ inside before starting the day. Informants explained how they greet friends in the morning only on the way back from the public toilets and not on the way there because of their full bowels, and identiﬁed numerous other ways that people hold themselves differently before and after their morning ablutions (Obika et al., 2002).

Non-adopter households—constraining factors blocking adoption Respondents also were asked about the con- straints to constructing household toilets. The major constraints cited by non-adopters were limited space(48.4%), high costs (33.6%), no one to build (32.3%), competing priorities (31.8%), and savings and credit issues (30.1%).

These constraints were also commonly cited in in- depth qualitative research conducted in a small Ghanaian town broadly representative of peri-urban Ghana (Obika et al., 2002). Households are densely packed into tight areas, compound housing common, and spare space highly limited, existing toilet technologies are expensive to install apart from the bucket latrine, and there is limited knowledge of their operation and of masons to construct them. School fees are a priority for limited savings and few if any formal credit mechanisms exist for home improve- ments (Obika et al., 2002). The inﬂuence of these motivations and constraints on sanitation adoption decisions and new demand is examined next.

Determinants of new demand in Ghana Among non-adopter households, most (61.7%) had never considered installing a household toilet and therefore not yet entered the adoption process (Fig. 2). Among those with preference for installing a toilet (38.3%), the proportion at the intention stage was high (81.7%), as measured by some likelihood of building a toilet in the next year. However, of the sample of 399 non-adopter households, only 5.8% expressed a high likelihood of building within the next 12 months, resulting in a low rate of new demand. The next sections explore factors contribut- ing to preference for toilet installation and to the likelihood of toilet building in Ghana.

Households with preference Households who had considered installing a home toilet were less satisﬁed with their current defecation place and stated signiﬁcantly more ARTICLE IN PRESS No Preference Yes 81.7% Yes 38.3%Yes 18.5% Stage 1: Preference for Sanitation (100% of non-adopters)Stage 2: Intention to Install (38.3 % of non-adopters with preference)Stage 3: Choice (31.3 % of non-adopters with preference & intention) New Demand 5.8 % of non-adopters = Fig. 2. New sanitation demand and adoption stage rates in Ghana 2003. M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–24422435 reasons for building household toilets than those who had not considered installing one (Table 4).

Such households were more likely to mention six reasons: good health, ease of cleaning, cleanliness, privacy/dignity, safety/security, and avoid sharing with others. Differences in dissatisfaction and their tendency to cite reasons related to good health, privacy/dignity, and safety/security were signiﬁcant(pp0.05), while the others were nearly so (Table 4).

Those who had considered installing sanitationand stated good health as a reason were signiﬁcantly (pp0.05) more likely to say that feces was the cause of ill health than those stating good health who had not considered installation. Relationships between health and use of public and poor condition latrines or open defecation in Ghana were probed in the ARTICLE IN PRESS Table 4 Non-adopter households differences at preference, intention, and choice stages Variable No preference (N¼246 of 399)Preference (N¼153 of 399)Preference but No intention (N¼28 of 153)Preference and intention (N¼122 of 153)Intention but No choice (N¼102 of 122)Intention and choice (N¼20 of 122) Satisfaction with current defecation place Dissatisﬁed 61.4% a 71.2% 67.9% 72.1% 71.6% 75% Very dissatisﬁed 15.4% 16.3% 10.7% 18% 14.7% 35% Top three reasons for building Convenience 50.4% 52.9% 50% 54.9% 52.9% 65% Good health 35.8% 51.6% 53.6% 50.0% 53.9% 30% Cause germs c 29.5% 36.7% Heat c 20.5% 19.0% Smell c 26.1% 16.5% Dirt c 5.7%* b 13.5% Feces c 1.1% 7.6% Pests c 6.8% 3.8% Easy to keep clean 39.8%* 48.4% 42.9% 50.8% 52% 45% Cleanliness 24.4%* 33.3% 39.3% 32.8% 32.4% 35% Privacy/dignity 17.1% 27.5% 32.1% 26.2% 28.4% 15% Safety/security 17.1% 26.8% 32.1% 25.4% 27.5% 15% Visitors/guests 23.6% 24.2% 25% 24.6% 23.5% 30% Avoid sharing with others/strangers13.8% 18.3% 28.6% 16.4% 17.6% 10% Comfort 11.4% 13.1% 14.3% 11.5% 13.7% 0% Prestige/pride; don’t have to pay to use; old age/illness; for children to usep4.1%p2.6%p3.6%p3.3%p2.9%p5% Number of reasons cited 2.4 3.0 3.2 3.0 3.0* 2.5 Constraints High costs 26.4% 45.1% 28% 50% 53.9% 30% No one to build 27.6% 39.9% 21% 45% 49% 25% Water table/soil conditions10.6% 22.9% 14% 25% 27.5% 15% Technical complexity 4.1% 11.1% 3.6%* 13% 15.7% 0% Savings, credit issues 29.7% 30.7% 50% 27% 26.4% 30% Competing priorities 32.8% 30.1% 55% 25% 25% 30% Tenancy issues 29.3% 25.2% 50% 20% 18.6% 30% Limited space 54.1% 39.2% 71% 33% 33% 30% Permit problems 11.8%* 6.5% 7.2% 6.6% 6.9% 5% Satisﬁed with toilet 8.5% 5.9% 3.6% 6.6% 6.9% 5% Lack decision making 4.9% 2.6% 0% 3.1% 2.9% 5% Poor options 0.8% 1.3% 0% 1.6% 2% 0% a,b po0.05, *po0.10 chi-square value likelihood ratio for difference in prevalence between two groups. cAsked only of those mentioning ‘good health’.M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–2442 2436 initial qualitative study and revealed a web of beliefs linking feces to ill health that hardly featured germs.

Respondents felt risks to health arose when latrine users were exposed to heat and foul gases from decomposing feces which transmitted diseases and that vomiting, stomach ache, loss of appetite, constipation, headache, and even infectious ill- nesses, among other health problems, could be induced through the sight or smell of fresh feces.

The miasma theory rather than the germ theory appears to be the more relevant operating principle in these beliefs about avoidance of feces and good health, echoing previous ﬁndings from the neigh- boring country of Benin (Jenkins & Curtis, 2005).

Households with intention The likelihood of building a latrine, rated as ‘no chance, low, medium, or high’, was measured for the 153 non-adopter households who had consid- ered installation. Of these, nearly 4 of 5 expressed some positive intention to build within 12 months.

No signiﬁcant differences in either satisfaction levels or motivation were found between those with and without intention to build (Table 4). As hypothe- sized in the adoption stages model and consistent with planned behavior theory, signiﬁcant differences between the two groups were only found for constraints and priority. Those with no intention to build within 12 months were 2–2.5 times more likely to mention limited space (71% to 33%), competing priorities (55% to 25%), savings/credit issues (50% to 27%), and tenancy issues (50% to 20%) than those with positive intention, showing these constraints block the formation of intention.

Households who have chosen to build a toilet Only 1 in 6 of the 122 households with some intention to build in the next year said they had a high likelihood of completing construction. The remaining 5 of 6 intender households expressed medium or low likelihoods of completion. Satisfac- tion levels, motivating reasons, and stated con- straints of these two groups were compared (Table 4). Those with a high likelihood of building were twice as likely to be very dissatisﬁed with their current defecation place as those with medium or low likelihood. Choosers were also statistically less likely to give good health and comfort as top reasons for installation, but were otherwise motivationally similar to those with medium or lowintentions, although convenience, cleanliness, and visitors/guests were more commonly mentioned as top reasons. Constraints blocking ﬁnal choice to build included high costs, no one to build, water table/soil conditions, and technical complexity.

These constraints were positively correlated with progression from preference to intention stages (Table 4). The statistical signiﬁcance or near signiﬁcance of their relative absence among choo- sers suggests these four factors are important barriers to new demand that operate late in the decision process after intentions take shape, to block implementation by postponing or delaying construction.

Three of the four factors blocking early intention relate primarily to structural factors associated with the individual household’s situation, abilities, and resources (i.e., tenancy, savings/credit, and limited space). Difﬁcult to change in the near to medium term and perceived early in the decision process, these would make it futile to explore plans to build.

Conversely, all four factors that later block choice relate to the nature, quality, and availability of opportunities to build a toilet which are external to the household, but essential for adoption (i.e., high cost, no one to build, water table/soil conditions, and technical complexity).

Preferred toilet types No statistically signiﬁcant preferences for toilet type emerged for any adoption stage. Slightly more households with positive intention chose ventilated improved pit latrines (KVIP/VIP) than those with preference but no intention, while slightly more households with a high likelihood of building within the year picked ﬂush toilet than those stating medium and low likelihoods, but these differences were not signiﬁcant.

Discussion The survey results, summarized inTable 5, support a model of three adoption stages of preference, intention, and choice as a practical tool for assessing demand for sanitation among house- holds in Ghana.Preferencefor changing sanitation is largely created by dissatisfaction with current practices and good awareness of the beneﬁts of home toilets in Ghana.Intentionto build is determined by positive preference, prioritization, and the absence of structural constraints related to ARTICLE IN PRESS M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–24422437 individual situational factors or abilities that may appear insurmountable to the household. The ﬁnal choiceto install a toilet depends on the additional access to appropriate opportunities to build, related to product choices, cost, building services, soil conditions, and access to good technical informa- tion and support.

High levels of dissatisfaction with one’s current defecation place in the choice stage suggest added urgency to put an existing plan into action. Trigger events such as an embarrassing accident or missing an important school or work event due to queues at the public toilets (Obika et al., 2002), or sudden sickness or deteriorated health of an aged parent, may increase dissatisfaction with the household’s present defecation situation and raise priority for a home toilet above other demands on household resources. Other trigger events identiﬁed in our qualitative research included building a new house, hosting a important social event with many invited guests for such affairs as funerals in Ghana, or an extended visit from a relative from the city or abroad.

Strategies to increase sanitation demand in Ghana Strategically designed and targeted marketing and market-based interventions could remove or reduce some barriers to adoption identiﬁed in Ghana. We discuss strategies to increase demand for household toilets in Ghana at each adoptionstage and examine where marketing approaches are unlikely to change decision behavior. In such cases, approaches involving legal mechanisms and educa- tional/informational campaigns may be needed as complementary tools (Rothschild, 1999).

Increasing preference Signiﬁcant determinants at each stage of toilet adoption in Ghana show that awareness, dissatis- faction, and motivation are needed to start the decision process. Large-scale marketing communi- cations using advertising and consumer information dissemination methods to enhance awareness of options, highlight beneﬁts, and arouse motivation offers a promising strategy to stimulate non-adopter households in Ghana who have not considered toilet installation (460%) to do so. The campaign should focus on the salient beneﬁts of installing household toilets in the Ghanaian context—conve- nience, safety, and cleanliness—and increase aware- ness of negative aspects of current defecation practices associated with dissatisfaction, speciﬁcally their dirty and smelly state. Motivations for adopting sanitation in Ghana appear largely un- related to the fecal–oral transmission of disease, conﬁrming similar ﬁndings elsewhere (Cairncross, 2004;Jenkins, 2004). Reasons to change sanitation have been shown to vary considerably across households as a function of lifestyles, local environ- ment, and socio-cultural aspects of excreta handling and defecation practices, but typically have little to do with preventing fecal-oral diseases (Frais & Muhkerjee, 2005;Jenkins & Curtis, 2005;Mukher- jee, 2001;Obika et al., 2002). In Ghana, cleanliness and neatness are particularly salient motivations for a wide range of hygiene behaviors. Neatness is culturally tied to notions of moral and social purity, while diseases associated with feces are believed to be transmitted via sighting feces and by fecal heat and odor produced in open latrines (Obika et al., 2002;Scott, Curtis, Rabie, Garbrah-Aidoo, & Research International, 2002;van der Geest, 1998).

Increasing intention to build In Ghana, household situational constraints related to limited space, tenancy, and savings were the main factors preventing households with posi- tive preference for a home toilet from forming an intention to build one. Tenant households in Ghana have little or no control over the sanitation ARTICLE IN PRESS Table 5 Households without home sanitation by adoption stage in Ghana Decision stage Factors blocking decision to adopt improved sanitation Preference Lack of awareness of beneﬁts of household toilet Weak or few motivations Satisfaction with existing defecation practice Intention Lack of preference Lack of priority or competing priorities Permanent constraints related to individual situation, including: limited space, tenancy issues, credit and savings difﬁculties Choice Lack of preference and intention Satisfaction with existing place Temporary constraints related to opportunities: high cost, no one to build, water/soil conditions, and technical complexityM.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–2442 2438 infrastructure where they live. Only two of the 76 tenant adopter households said they had made the decision to build and only one indicated some claim to ownership of the latrine. This explains the disproportionately higher fraction of non-adopter households stalled at the preference stage who are tenants (55%) compared to the intention stage (20%). Interestingly 100% of tenant households stated competing priorities as a main reason for not building a toilet compared to just 5.6% of other non-adopter households. Investigation of the vari- ety of tenant occupancy patterns, tenant priorities, and how different kinds of landlords make sanita- tion installation decisions in Ghana is warranted to understand the unique sanitation access problems of non-owner households and identify strategies for this distinct population segment. Tenancy in tenant- only houses is likely to create a housing situation where marketing is ineffective in achieving sanita- tion improvements without legal action to encou- rage landlords to add sanitation facilities to their properties.

Limited space is also a more complex constraint to overcome—it may require development of new sanitation technologies and services before market- ing can be applied. Expression of this constraint is symptomatic of lack of pit emptying services in poor urbanizing areas and of sanitation technologies that require excessive amounts of space, relative to the cost and opportunity value of space in poorer dense neighborhoods and over-crowded slums. The buck- et latrine with frequent and regular emptying service was a product-service package that worked well for decades for hundreds of thousands of Ghanaian households who lacked space and capital, until national policy called for fazing them out and public conservancy labor arrangements ended in many towns. Indeed many respondents in the qualitative work spoke positively of the bucket latrine, its major limitation being that it was now difﬁcult to ﬁnd people, and often expensive, to empty them.

People valued their low cost, their limited smell due to regular emptying and the removal of feces from their home. In Tanzania, the importance of pit emptying services for sustaining on-site urban sanitation has emerged as a critical factor for adoption and maintenance of household toilets (Jenkins & Sugden, 2006).

The savings constraint reﬂects two structural problems: real poverty and a lack of ﬁnancing and credit options for home improvement. Marketing is unlikely to be able to fully address either of theseand laws, public policies, and other mechanisms are required. However, work in Vietnam is ﬁnding that development of ﬂexible payment schemes can help reduce savings difﬁculties related to the high initial cash cost of sanitation installation (Frais & Muhkerjee, 2005).

Increasing nal choice Approximately 30% of non-adopter households in Ghana have begun planning to install sanitation as shown by their positive intention to build.

However 5 of 6 are stalled in the process, failing to carry through their intention. Major reasons for failure to progress are: perceived high costs of toilet options, no one to build, the complexity of building related to lack of information, and water table and soil problems. These constraints to new demand could be addressed by actions to improve the quality, range, and costs of toilet technologies offered in the market place, innovative ways and incentives to extend the private sector supply chain of toilet products and related services (e.g.

vault or pit emptying) needed to build, operate, and maintain toilets closer to these households, and sales promotion and product education and market- ing to reduce households’ transaction time and effort costs involved in searching for good informa- tion about technologies and how to get them built; product, place, price, and sales promotion compris- ing the four basic ‘Ps’ of a marketing plan (Kotler et al., 2002).

Sanitation gains using marketing strategies Projections to 2015 were made for Ghana to examine sanitation access scenarios in light of development goals and the potential gains from marketing strategies emerging from this national demand assessment. Assuming a uniform annual rate of new household toilet adoption at 5.8% of non-adopter households, 100% maintenance of installed household facilities, phasing out bucket latrines by 2010, current housing patterns, and a population growth rate of 1.9%, calculations indicate home toilet access in Ghana would rise from 25.6% of households in 2003 to 54.3% by 2015 (Fig. 3) without additional action. This projection approaches 57.5%, the coverage required for Ghana to meet its 2015 millennium development goal for sanitation, assuming coverage in 1990 was 15% as recently re-estimated (WHO & Unicef, ARTICLE IN PRESS M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–24422439 2006). How much would proposed marketing strategies to reduce blockages at the preference and choice stages be expected to increase adoption rates and coverage?

In total, 2015 projections for two marketing scenarios were compared to the ‘‘no action’’ base case. Scenario 1 assumes a national advertising and communications campaign raises household aware- ness and interest in home toilet installation and increases the rate of preference (those who have considered installing a toilet) by 50% (from 38.5% to 57.8% of non-adopter households). No actions to change the base rates of intention and choice are assumed. Scenario 1 increases adoption rates to 8.7% per year and achieves 65% projected sanitation access by 2015. Scenario 2 adds a mix of marketing strategies to scenario 1 to reduce barriers and increase the rate of choice in the last stage of adoption by 50% (from 18.5% to 27.8% of intenders). Again, no change is made to the intention rate, determined largely by constraints which market- ing appears ineffective at addressing. Combining the changed rates of preference and choice, Scenario 2 yields a 13.1% annual rate of adoption and projected 77% sanitation access by 2015, achieving an estimated 1.2 million more households or 5.8 million more people with home sanitation by 2015 over the baseline in Ghana. Although these projections reﬂect the simpliﬁed assumptions of a static adoption rate and instantaneous effects, they illustrate causalpathways by which marketing strategies would be expected to work to accelerate adoption rates above baseline trends.

Segmenting households This analysis reveals how population subgroups blocked at the intention stage would remain without access to a home toilet unless other non-market- based actions and policies were developed to address constraints of extreme poverty and tenancy that block home toilet installation in Ghana.

Consumer segmentation is a fundamental planning tool in marketing, whereby the target population is divided into more manageable homogenous seg- ments for which a speciﬁc mix of marketing activities can be developed. A core implication of our sanitation adoption decision model and this analysis is the need for different interventions at different stages in the adoption process. Programs may choose to design a set of strategies that target only one adoption stage, or all of them, but one blanket intervention for everyone is unlikely to work. Using a behavioral approach to assess demand allows managers to work out where most of their target population lies in the adoption process and how vulnerable subgroups compare to the majority, enabling development of actions that target barriers at each adoption stage for any population segment of interest. ARTICLE IN PRESS 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 20032004 2005 20062007 2008 2009 2010 2011 2012 2013 2014 201 5 Ye a r Percent Households with Toilet Base (5.8% per year) Scenario 1 (8.7% per year) Scenario 2 (13.1% per year) Fig. 3. Projected household toilet coverage rates in Ghana. M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–2442 2440 Limitations and recommendations Response rates and reliability for some questions in the Ghana survey would have been improved by interviewing the person in each household respon- sible for decisions about building or improving housing facilities. One third of the respondents in this survey were unable to say when their toilets were built and over 20% unable to say why they had been built because they were not the decision- makers. Nonetheless, the results indicate that mothers in Ghana in most households participate in infrastructure decisions sufﬁciently to be able to answer many of the questions related to future toilet acquisition and demand. Using the tool, it was possible to categorize all non-adopter households in terms of their preference, intention, and choice to install household sanitation. Three non-adopters with preference had missing intention and choice data because a section of non-adopter questions was mistakenly skipped.

Prior to adapting the survey for another setting or context, a small qualitative study using in-depth interviews or focus group discussions, as mentioned earlier is required to establish sensible codes for questions related to motivations, constraints, toilet types, and preferred attributes. We drew from a recent qualitative study of these issues and supple- ment it with pilot testing.

Overall, the survey was relatively straightforward to conduct and took about 20 min to administer. It required no additional training or skills beyond those of personnel trained and experienced in quantitative household survey methods and sam- pling techniques and familiar with the project population. An experienced market research agency was able to conduct the ﬁeld work with oversight developing the questionnaire and coding schemes, and pilot testing.

Conclusions Findings from a national survey to assess sanita- tion demand in Ghana ﬁt a preference–inten- tion–choice behavioral decision model of household sanitation adoption. Results show how satisfaction with existing defecation practices, mo- tivations for improving sanitation, priority over competing household concerns, and situational and implementation-related constraints affect preference for and likelihood of household toilet installation and can lead to new demand for sanitation inGhana. This survey-based behavioral approach provides a quick and effective method to assess and understand what drives household demand for improved sanitation, segment households by adop- tion stage, and pinpoint focused strategies to stimulate increased rates of preference, intention, and choice to improve sanitation. In the Ghana case, categorizing the target population in terms of the adoption stages that generate new demand for home toilets provides useful information to identify policies and design interventions to stimulate higher rates of demand. In particular, marketing strategies aimed at the preference and choice stages appear to be promising ways to increase household sanitation demand and coverage in Ghana.

Acknowledgments Nana Garbrah-Aidoo and Research International, Ghana, managed the survey ﬁeld work and data entry and designed the sampling approach. Steven Sugden, Sandy Cairncross, and Jay Lund provided suggestions and comments on earlier drafts.

References Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In J. Kuhl, & J. Beckman (Eds.),Action control:

From cognition to behavior. Heidlburg: Springer.

Ajzen, I. (2002). Perceived behavioral control, self-efﬁcacy, locus of control, and the theory of planned behavior.Journal of Applied Social Psychology,32, 1–20.

Ajzen, I., & Madden, T. J. (1986). Predicting of goal-directed behavior: Attitudes, intentions, and perceived behavioral control.Journal of of Experimental Social Psychology,22, 453–474.

Altaf, M. A. (1994). Household demand for improved water and sanitation in a large secondary city.Habitat International, 18(1), 45–55.

Altaf, M. A., & Hughes, J. A. (1994). Measuring the demand for improved urban sanitation services: Results of a contingent valuation study in Ouagadougou, Burkina Faso.Urban Studies,31, 1763–1776.

Ayee, J., & Crook, R. (2003).Toilet wars: Urban sanitation services and the politics of public-private partnerships in Ghana.

IDS Working Paper No. 213, Institute of Development Studies, Brighton, Sussex, UK, December 2003.

Cairncross, S. (2004).The case for marketing sanitation. Field Note, Sanitation and Hygiene Series. Water and Sanitation Program—Africa, The World Bank, Nairobi, Kenya, August 2004.

Donovan, R., & Henley, N. (2003).Social marketing: Principles and practice. Melbourne, Australia: IP Communications.

Fishbein, M., & Ajzen, I. (1975).Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley. ARTICLE IN PRESS M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–24422441 Frais, J., & Muhkerjee, N. (2005).Harnessing market power for rural sanitation. Field Note. Water and Sanitation Program— East Asia and Paciﬁc, The World Bank, Jakarta, Indonesia.

Gilbert, A. (1983). The tenants of self-help housing: Choice and constraint in the housing markets of less developed countries.

Development and Change,14, 447–449.

Jenkins, M. W. (1999).Sanitation promotion in developing countries: Why the latrines of Benin are few and far between.

Ph.D. dissertation, Department of Civil and Environment Engineering, University of California, Davis, California.

Jenkins, M. W. (2004).Who buys latrines, where and why?Field Note, Sanitation and Hygiene Series. Water and Sanitation Program—Africa, The World Bank, Nairobi, Kenya, Sep- tember 2004.

Jenkins, M. W., & Curtis, V. (2005). Achieving the ‘good life’:

Why some people want latrines in rural Benin.Social Science & Medicine,61(11), 2446–2459.

Jenkins, M. W., & Sugden, S. (2006).Rethinking sanitation:

lessons and innovation for sustainability and success in the New millennium. Sanitation Thematic Paper, UNDP Human Development Report 2006, UNDP HDRO, New York, January 2006.

Kotler, P., Armstrong, G., Saunders, J., & Wong, V. (2002).

Principles of marketing(3rd ed.). New York: Prentice Hall Press.

Metha, M., & Knapp, A. (2004)The challenge of nancing sanitation for meeting the Millennium Development goals.

Commissioned paper for the Commission on Sustainable Development, Norwegian Ministry of the Environment, Water and Sanitation Program—Africa, The World Bank, Nairobi, Kenya, March 5, 2004.

Mukherjee, N. (2001)Achieving sustained sanitation for the poor:

Policy lessons from participatory assessments in Cambodia, Indonesia and Vietnam. Field Note, Water and Sanitation Program—East Asia and the Paciﬁc, Jakarta, Indonesia.

Obika, A., Jenkins, M., Curtis, V., Howard, G. & & TREND.

(2002).Social marketing for urban sanitation: Review ofevidence and inception report. Loughborough, UK: Water, Engineering and Development Centre, Loughborough Uni- versity.

Rakodi, C. (1995). Rental tenure in the cities of developing countries.Urban Studies,32(4–5), 791–811.

Rothschild, M. L. (1999). Carrots, sticks, and promises: A conceptual framework for the management of public health and social issue behaviors.Journal of Marketing,63, 24–37.

Scott, B., Curtis, V., Rabie, T., Garbrah-Aidoo, N., & Research International. (2002).What motivates handwashing in Ghana?

A re-analysis of the results of the formative research. Available at:/www.globalhandwashing.orgS.

Swiss Agency for Development and Cooperation (SDC). (2004).

Sanitation is a Business: Approaches for demand oriented policies. Water and Sanitation Program, The World Bank and Swiss Agency for Development and Cooperation, Bern, Switzerland.

UNCHS. (1996).An urbanizing world: Global report on human settlements 1996. Oxford: Oxford University Press.

van der Geest, S. J. (1998). Akan shit: Getting rid of dirt in Ghana.Anthropology Today,14(3), 8–12.

van der Geest, S. J. (2002). The night-soil collector: Bucket latrines in Ghana.Postcolonial Studies,5(2), 197–206. Water Supply and Sanitation Collaborative Council (WSSCC) and World Health Organization (WHO). (2005).Sanitation and hygiene promotion programming guidance. Geneva, Switzerland: WHO Press, The World Health Organization.

WHO and UNICEF. (2004).Meeting the MDG drinking water and sanitation target: A mid-term assessment of progress, 2004.

WHO/UNICEF JMP for Water Supply and Sanitation, The World Health Organization, Geneva, Switzerland.

WHO and UNICEF. (2006).MDG Assessment Report 2006.

Ghana coverage data accessed January 3, 2007 at/http:// www.wssinfo.org/pdf/country/GHA_san.pdfS.

Wittington, D., Lauria, D. T., Wright, A. M., Choe, K., Huges, J.

A., & Swarma, V. (1993). Household demand for improved sanitation services in Kumasi, Ghana: A contingent valuation study.Water Resources Research,29(6), 1539–1560. ARTICLE IN PRESS M.W. Jenkins, B. Scott / Social Science & Medicine 64 (2007) 2427–2442 2442