Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

A hierarchical Bayesian Belief Network model of household water treatment behaviour in a suburban area: A case study of Palu—Indonesia

  • D. Daniel ,

    Roles Conceptualization, Formal analysis, Methodology, Resources, Visualization, Writing – original draft, Writing – review & editing

    Affiliation Department of Water Management, Delft University of Technology, Delft, The Netherlands

  • Mita Sirait,

    Roles Conceptualization, Project administration, Resources, Writing – review & editing

    Affiliation Department of Ministry Quality and Impact, Health Units, Wahana Visi Indonesia, Tangerang Selatan, Indonesia

  • Saket Pande

    Roles Conceptualization, Data curation, Methodology, Supervision, Validation, Writing – review & editing

    Affiliation Department of Water Management, Delft University of Technology, Delft, The Netherlands


Understanding the determinants of household water treatment (HWT) behavior in developing countries is important to increase the rate of its regular use so that households can have safe water at home. This is especially so when the quality of the water source is not reliable. We present a hierarchical Bayesian Belief Network (BBN) model supported by statistical analysis to explore the influence of household’s socio-economic characteristics (SECs) on the HWT behavior via household’s psychological factors. The model uses eight SECs, such as mother’s and father’s education, wealth, and religion, and five RANAS psychological factors, i.e., risk, attitude, norms, ability, and self-regulation to analyse HWT behavior in a suburban area in Palu, Indonesia. Structured household interviews were conducted among 202 households. We found that mother’s education is the most important SEC that influences the regular use of HWT. An educated mother has more positive attitude towards HWT and is more confident in her ability to perform HWT. Moreover, self-regulation, especially the attempt to deal with any barrier that hinders HWT practice, is the most important psychological factor that can change irregular HWT users to regular HWT users. Hence, this paper recommends to HWT-program implementers to identify potential barriers and discuss potential solutions with the target group in order to increase the probability of the target group being a regular HWT user.


The Sustainable Development Goals (SDGs) are more ambitious than the Millennium Development Goals because safety aspects of drinking water have been included as one of the new targets. Despite significant efforts to achieve this target in the past, three out of ten people worldwide still used contaminated water services in 2017 [1]. Since contaminated drinking water contributes significantly to water-related diseases, especially among the children below the age of five [2, 3], the safety aspect of the drinking water cannot ignored in efforts to achieve SDGs.

The trends of global drinking water service levels from 2000 to 2017 suggest that the target to have 100% safely managed drinking water services by 2030 is hard to achieve [1]. There is also a significant deterioration of water quality during transport and storage of water [47]. Therefore, it is important to have a more thoughtful, “interim,” approach in developing countries so that households can still consume safe drinking water.

Household water treatment (HWT), which means the use of any type of method to treat drinking water at a household level, such as boiling and water filtration, can be considered as an interim solution to improve the water quality at the household level even when the water quality from its source is contaminated [8, 9]. HWT can improve water quality and reduce water-related diseases, such as diarrhea, as long as the users perform it correctly and regularly [10]. However, HWT use has been declining [11] with many households performing HWT irregularly [12, 13].

The purpose of this study is to assess the regular practice of HWT among households in a suburban area of Palu, Province Central Sulawesi, Indonesia. According to the Demographic Health Survey in 2017, 66% of the total households in Central Sulawesi treated their drinking water, which is slightly below the national average 68% [14]. The HWT use in Central Sulawesi decrease slightly compared to the survey in 2012, i.e., 71%, while the national average remained constant [15]. We did not focus on a specific HWT method, but on general HWT behaviour. Therefore, “appropriate” HWT methods, i.e., boiling, water filtration, chlorination, and solar disinfection, were not differentiated.

A Bayesian Belief Network (BBN) model, which combines socio-economic characteristics (SEC) and psychological factors of households, is used to understand HWT behaviour. A three-level hierarchical BBN model is created with household’s socio-economic characteristics in the top layer, the psychological factors as the intermediate nodes, and the HWT behaviour as the output variable based on Daniel et al. [16], who found that the effect of SEC on HWT behaviour is mediated by the psychological factors. Finally, recommendations to increase the regular practice of HWT are also presented based on the obtained results.


Study setting

We conducted the HWT behavioural study in July 2018 in the district of Palu, Province Central Sulawesi, Indonesia, in collaboration with a national NGO called Wahana Visi Indonesia (WVI). A total of 202 households were visited in three sub-villages within two suburban villages: (1) Wana and (2) Lekatu in village Tipo, and sub-village Salena in village Buluri (Fig 1). The sample size was obtained based in the methodology of [17] (check supporting information S1 for more information). These locations were selected as representatives of suburban villages of the iReach project, initiated and conducted by WVI, that have with high levels of diarrhea occurrence among children under the age of five years. The iReach project itself aims to improve the health of mothers and children in the district of Palu. Considering that many households in this project area still drink unsafe–untreated water, the surveys aimed to assess the practice and perceptions about HWT among the community.

We used a structured household interview which comprised of household’s socio-economic characteristics (SEC) information, such as parent’s education level, religion, and a list of household assets, and also HWT related information on knowledge, perceptions (psychological), and use related behaviour. We used mainly a five-Likert scale answers for the psychological factors and categorical answers for the SECs.

Six locals were trained to conduct the interviews and a pilot test was conducted before the real data collection. We mainly targeted the mother or primary caregiver for the interview. All participants gave written informed consent before being interviewed. The present study conformed to the guidelines of the declaration of Helsinki in human subjects. The study was approved by the human research ethics committee of Delft University of Technology and received government approval at the district level as part of the WVI iReach project. The first and second authors are from Indonesia and, therefore, do not need special permits to conduct the study.

Socio-economic characteristics (SECs)

Eight socio-economic characteristics mentioned by previous studies influencing the HWT or other WASH behaviour were used: (1) water-related health problem [18, 19], (2) information access [20, 21], (3) mother and (4) father’s education [19, 2225] (5) wealth level [26, 27], (6) religion [2830], (7) accessibility [24, 31], and (8) access to water [25]. An answer to the question “how often do you watch TV?” (frequency watching TV) was used to represent information access, especially to mass media [32, 33]. For variable water-related health problem, we used information on the incidence of diarrhea among children below the age of five in the preceding two weeks of the time of the visit.

Psychological factors

We followed the RANAS psychological framework to analyse HWT-related perceptions [34]. RANAS stands for Risk, Attitude, Norms, Ability, and Self-regulation, five psychological factors which are believed to be responsible for individual behaviourial outcome. Risk represents one’s awareness and understanding of the behaviour. Attitude is related to the feeling towards the behaviour. Norms represent social pressure towards the behaviour. Ability indicates one’s confidence in his or her ability to perform the behaviour. Lastly, Self-regulation depicts individual attempts to self-monitor and plan the behaviour and deal with conflicting goals. To cover well all aspects of each factor, RANAS framework uses several questions at a sub-factor level (Table 1). See [34] for detail definition of all the sub-factors. RANAS has been used in many HWT or WASH-related behaviour, see for example [3539]

Outcome variables: Household water treatment (HWT) behaviour

To assess the practice of HWT among the respondents, a self-reported answer of whether they treat their drinking water at the time of visit were combined with respondents’ answers to four questions related to their HWT behaviour. The four questions corresponded to the frequency of drinking raw water daily, percentage of water treated daily, habit of performing HWT, and intention to treat water. The intention behind combining multiple answers is one of the strategies to diminish the bias in self-reported behaviour, which may overestimate the actual behaviour [16, 4042].

Bayesian Belief Network (BBN)

A Bayesian Belief Network (BBN) is a directed acyclic graph showing a hypothetical causal relationship between causal variable (called “parent node” in BBN) and the affected variable (called “child node”) [43]. The causal graph or the BBN structure represents the qualitative aspect of BBN since the structure is often inspired by conceptual theories or frameworks or expert consensus [44]. The quantitative aspect of BBN is reflected by the Conditional Probability Tables (CPT), which measure the strength of relationship between parent and child nodes.

Data analysis

Two main analyses were conducted: (1) statistical analysis: the regression analysis; and (2) the BBN analysis.

Before conducting those two main analyses, the PCA was used to create variables corresponding to the nodes of the developed BBN model. These included wealth level, the five RANAS factors Risk, Attitude, Norms, Ability, and Self-regulation, and the output variable HWT behaviour. The PCA for wealth level was performed to estimate a representative value of relative wealth index of a household based on the observations of household assets [45]. A similar approach was used for the five RANAS factors. Since psychological information was available at sub-factor level (Table 1), PCA was used to “reduce” the dimensionality (information or the number of variables in the analysis) and capture the dominant information of the five main RANAS factors. For example, there are three sub-factors of Risk: perceived vulnerability, health knowledge, and perceived severity. PCA was used on these three sub-factors to obtain one representative variable for Risk. The same applies to the other four RANAS factors. PCA was also used to create output variable HWT behaviour using five related questions discussed in the section “outcome variable”.

Forced-entry multivariate regression analysis was performed using all RANAS sub-factors (Table 1) as predictor variables and variable HWT behaviour as the outcome variable. All statistical analysis used IBM SPSS Statistics 25 (IBM Corp., Armonk, NY).

For the BBN analysis, continuous valued variables were discretised since discrete valued BBN model was used. All PCA outputs were discretized into three categories. For the wealth level, the respondents were discretized based on their PCA scores: poor (the lowest 40%), middle (the next 40%), and rich (the last 20%) [45, 46]. Three levels were also assigned for psychological factors: low (lowest one-third of scores, e.g., low Risk), moderate (one-third to two-thirds of the lowest scores, e.g., moderate Risk), and high (the remaining data). Finally, a similar approach for the HWT behaviour was used and three categories were created: “non-user”, “irregular user”, and “regular user”. All the discretised variables were then used in the BBN analysis.

The BBN model was developed using Genie 2.2 ( software package. The software utilizes the expectation maximization (EM) algorithm to estimate the CPTs within the model [47]. The algorithm has proven to be effective in estimating the CPTs in case of incomplete data [48]. The model’s performance was assessed using the same software using a ten-fold cross-validation test. The Area Under the Curve (AUC) value of the Receiver Operating Characteristics (ROC) curve showed model’s performance. A value close to one indicates perfect prediction of the output variable (higher sensitivity and lower false positives) [49]. A sensitivity analysis was also performed to identify sensitive model parameters (entries of CPT). Furthermore, predictive or Bayesian inference was conducted to simulate the effect of specific SECs and psychological nodes on the output node. The most important nodes are the nodes with the highest ΔPHWT behaviour = regular, i.e., highest difference in the probability of HWT behaviour being “regular” between before situation (current situation without any update) and after updating a specific node situation. For example, node accessibility is updated to 100% “easy” and it is observed how it changes the probability of “regular” state of output node HWT behaviour. The same approach was conducted to all categories or levels in all SEC and psychological nodes one at a time and ΔPHWT behaviour = regular is analysed to identify important nodes. In addition, the Chi-square test was also conducted to confirm the statistical relationship between two categorical variables and strengthen the analysis of the BBN.


Socio-demographic characteristics of the respondents

The majority of the respondents had tap connection: 45.5% had access to tap water inside the dwelling, while 41.4% of respondents relied on a public tap. 16.8% of the respondents had no formal education, while 25.7% of the household heads were not attending formal education. Only 13.9% of the respondents had their own toilet, 65.8% used a shared toilet, and 20.3% of the respondents still practiced open defecation. 68.7% of the respondents stated that they had received HWT promotion in the past. The percentage of households with children below the age of five was 55% (range from 1–4 children). All those 55% households also reported diarrhea among their children in the last two weeks at the time of visit. The majority of the respondents said that boiling is the most often HWT method that they practiced (88%), while small portion used other methods (7%), such as solar disinfection or filtration, and 5% stated that they do not use HWT at all. From the self-reported answer, only 38.1% of the respondents said that they are treating water at that moment. Furthermore, based on the PCA results using other pieces information (see section outcome variable), 33.7% of the respondents were categorised as regular HWT users.

Regression analysis

Table 2 shows the results of regression analysis using all RANAS sub-factors as predictors of HWT use. According to the results, barrier planning (a person’s attempts to overcome barriers; a sub-factor of Self-regulation) is the most statistically significant psychological sub-factor, followed by affective belief (taste) (perception about the taste of water; Attitude) and action control (a person’s attempts to self-monitor a behaviour; Self-regulation) (see β value in Table 2). All other sub-factors in Self-regulation factor are also significant, as well as perceived vulnerability (perception on probability to get water-related disease) and health knowledge (knowledge on preventive measures of water-related disease) (both are Risk sub-factors) and confidence in recovering (perception’s on own ability to recover from setbacks; in Ability factor).

Table 2. Regression analysis of all RANAS sub-factors of psychosocial factors on HWT practice.

The Bayesian Belief Network (BBN) model

The BBN model is presented in Fig 2. It also shows the predicted probabilities of various states of the nodes after the model was calibrated (estimation of CPTs) on the household survey data. The model is “highly accurate”, according to Greiner et al., (2000), as shown by the AUC value of 0.90. The average model accuracy in predicting the output node is 79%.

Fig 2. The hierarchical BBN model shows the hypothetical causal relationships between socio-economic characteristics (SEC), RANAS psychosocial factors, and HWT behaviour.

The percentages in each node show the probability that a node is in a certain state. All 202 households were considered in calibrating the BBN model.

The sensitivity analysis shows that the node mother’s education is the most sensitive SEC and the node self-regulation is the most sensitive psychological factors (Fig 3). Node Water-related health problem, i.e., whether there was a diarrhea case in a household in the last two weeks, is far less important in the analysis. This implies that the occurrence of diarrhea among children below the age of five has no effect in influencing household’s psychology.

Fig 3. Sensitivity analysis of individual nodes on the probability of output node HWT behaviour being “regular”.

The predictive inference shows quite similar results with the sensitivity analysis, i.e., mother’s education and self-regulation are the most important SEC and psychological factor, respectively (Table 3). The influence of the node mother’s education is far bigger than any other SEC, i.e., ΔPHWT behaviour = regular = 10%. When we looked at the influence of mother’s education on each of the five RANAS psychological factors in more detail, we found that the level of attitude and ability change quite significantly in response to a change in mother’s education compared to other psychological factors. Furthermore, the influence of each SEC node on the output variable was a “mixed” effect: better SEC does not always result in a higher probability of HWT behaviour being “regular”. For example, the higher the level of parent’s education and relatively easy access result in a higher probability of being “regular”, while access to water did not lead to regular use of HWT. There is a small effect of religion on the behaviour, even though not statistically significant (X2 (2) = 5.40, p = 0.07). In addition, there is small negative effect of wealth on the HWT behaviour, but this effect is low in BBN and far from significant in the statistical analysis (X2 (4) = 5.32, p = 0.26). Amongst the psychological nodes, ability comes up as the second most important node and risk is the least important node.

Table 3. Predictive inference that measures the effect of each state in each node on HWT practice.

The value under each category corresponding to a node as displayed in the first column is the updated probability of the output node being “regular” given that all households maintain this state. The baseline probability was 31% (Fig 3).

Since self-regulation is the most important psychological factor, how the output node HWT behaviour reacts to changes in probabilities in this node was investigated. Fig 4 shows that when the level of self-regulation is changed from 100% “low” to 100% “moderate” (compare Fig 4A and 4B), “non-user” group switches to “irregular user” group. The difference between “non-user” and “irregular user” probabilities is dramatic. The difference in probability of being a “non-user” and a “irregular user” in the situation of when the probability of self-regulation being “low” is 100% is 51–26% = 25% (Fig 4A). In comparison, the difference in probability of being a “non-user” and “irregular” user is 34–35% = 1% in the situation of “moderate” self-regulation (Fig 4B), when). The difference between “irregular user” and “regular user” probabilities was not high in the respective situations (26–23% = 3% in Fig 4A and 35–31% = 4% in the Fig 4B).

Fig 4. The variation of the probability at the output node HWT behaviour in response to the node self-regulation being in a certain state while holding the probabilities in all other nodes constant.

In contrast, when the state of self-regulation was changed from 100% “moderate” to 100% “high” (compare Fig 4B and 4C), “irregular user” group switched to “regular user.”The difference of probability of being a “irregular user” and a “regular user” in the situation of “moderate” self-regulation is 35–31% = 4% (Fig 4B). In comparison, the difference of probability of being a “irregular user” and a “regular user” in the situation of “high” self-regulation is 42–29% = 13% (Fig 4C). The difference in probability of being a “non-user” and a “irregular user” is not high in the respective situations.


Socio-economic characteristics of households are often considered as the root cause of any health-related behaviour [50, 51]. Moreover, since the influence of household’s SECs on the behaviour was found to be mediated by psychological factors [16], it is important to analyse them in one such causal system, wherein SECs of households can “influence” the psychology of households to use HWT regularly.

The sensitivity analysis and predictive inference suggest that the level of mother’s education is critical in identifying HWT users, i.e., whether they are non-user, irregular, or regular users. The influence of mother’s education on psychological factors attitude and ability are quite dominant. This implies that more educated mothers have more positive attitude towards HWT and have more confidence in their ability to perform HWT. Another interpretation is that targeting and educating mother, with regards to HWT or WASH issues, is an important step to change the community behaviour, especially because they are often responsible for managing water in the household [52]. In contrast to mother’s education, father’s education does not influence the HWT behaviour much, in contrast to Figueroa & Kincaid [25] who indicated that father’s education may influence the household’s norm.

The influence of other SECs is far less influential. For example, diarrhea occurrence among the children and access to water do not influence the psychology of households to adopt HWT much. Access to mass media, such as TV, and type of religion that households follow also have little influence on the behaviour.

Self-regulation appears to be the most important psychological factor. This in line with the results of statistical analysis in which all sub-factors of self-regulation are statistically significant (Table 2). There was significant decline in the percentage of non-users when the probability of self-regulation was changed from 100% “low” to 100% “moderate” and significant increase in the probability of regular users was observed when the probability of self-regulation was changed from 100% “moderate” to 100% “high”. This finding shows how psychosocial factor self-regulation changes the non-users to irregular users and also from irregular users to regular users. This suggests that self-regulation, i.e., self-monitoring and evaluating their own current behaviour, is critical to convert irregular users to regular users, as also has been suggested by [34]. Moreover, the sub-factor barrier planning of self-regulation comes up as the most significant sub-factor according to the regression analysis. This means that households that have strategies to overcome possible barriers that hinder the behaviour are more likely to practice HWT regularly.

Hence, suggested strategies to change the behaviour are, first, to discuss and make a list of possible barriers with the respondents and help them to come up with potential solutions or strategies to overcome those barriers. Afterward, the counsellor or implementer should encourage the respondents to apply those strategies, i.e., eliminating physical and social interferences that may bar them from adopting the behaviour and to anticipate other barriers [53].

The important limitation in this study is that since the study was conducted in the intervention locations of the ongoing project of the NGO WVI, the responses may suffer from social desirability bias. However, we tried to minimize it by explaining the anonymity and confidentiality of their responses and mentioning that the study is conducted by independent university which is not related to the NGO. Furthermore, even though the sample size was sufficient to identify important factors associated with the HWT behaviour [54], larger sample size may provide more solid interpretations. Finally, most of the respondents were familiar with boiling and we are aware that there are some concerns related to the practice of boiling, such as time or cost spent, type of fuel used, and the issue of household air pollution [55, 56]. These concerns are out of the scope of this study but need to be taken into account by the project officer.


The socio-economic characteristics and psychological determinants of household water treatment behaviour in a suburban area of Indonesia were investigated using a Bayesian Belief Network model. The mother’s education level was the most important socioeconomic characteristics, while self-regulation was the most important psychological factor. Mother’s education influences the level of attitude and ability of households with regard to HWT behaviour. Self-regulation was found to be critical for the continuation of the HWT behaviour, i.e. change the irregular users to regular users. The self-regulation’s sub-factor barrier planning was found very significant on the behaviour. This suggests that households that are able to overcome potential barriers to perform HWT have a higher chance to perform HWT regularly.


We thank the respondents for their participation in the study, all interviewers, WVI officers in Palu (Agustince Simatupang, Grace Nasya Masela, dr. Yosellina, Rosnita), and WVI officers in the national headquarter in Jakarta (Yayan).


  1. 1. UNICEF, WHO. Progress on household drinking water, sanitation and hygiene 2000–2017. Special focus on inequalities. Progress on Drinking Water, Sanitation and Hygiene 2000–2017. 2019. 140 p.
  2. 2. Collaborators GDD. Estimates of global, regional, and national morbidity, mortality, and aetiologies of diarrhoeal diseases: a systematic analysis for the Global Burden of Disease Study 2015. Lancet Infect Dis. 2017;17(9):909–48. pmid:28579426
  3. 3. Black RE, Victora CG, Walker SP, Bhutta ZA, Christian P, Onis M De, et al. Maternal and Child Nutrition 1 Maternal and child undernutrition and overweight in low-income and middle-income countries. 2011;
  4. 4. Wright J, Gundry S, Conroy R. Household drinking water in developing countries: A systematic review of microbiological contamination between source and point-of-use. Trop Med Int Heal. 2004;9(1):106–17. pmid:14728614
  5. 5. Rufener S, Mäusezahl D, Mosler HJ, Weingartner R. Quality of drinking-water at source and point-of consumption-Drinking cup as a high potential recontamination risk: A field study in Bolivia. J Heal Popul Nutr. 2010;28(1):34–41. pmid:20214084
  6. 6. Meierhofer R, Bänziger C, Deppeler S, Kunwar B, Bhatta M. From Water Source to Tap of Ceramic Filters—Factors That Influence Water Quality Between Collection and Consumption in Rural Households in Nepal. Int J Environ Res Public Health [Internet]. 2018;15(11):2439. Available from: pmid:30388837
  7. 7. Elala D, Labhasetwar P, Tyrrel SF. Deterioration in water quality from supply chain to household and appropriate storage in the context of intermittent water supplies. Water Sci Technol Water Supply. 2011;11(4):400.
  8. 8. Sobsey MD, Stauber CE, Casanova LM, Brown JM, Elliott MA. Point of use household drinking water filtration: a practical, effective solution for providing sustained access to safe drinking water in the developing world. Environ Sci Technol. 2008;42(12):4261–7. pmid:18605542
  9. 9. Ren D, Colosi LM, Smith JA. Evaluating the sustainability of ceramic filters for point-of-use drinking water treatment. Environ Sci Technol. 2013;47(19):11206–13. pmid:23991752
  10. 10. Wolf J, Hunter PR, Freeman MC, Cumming O, Clasen T, Bartram J, et al. Impact of drinking water, sanitation and handwashing with soap on childhood diarrhoeal disease: updated meta-analysis and meta-regression. Trop Med Int Heal. 2018;23(5):508–25. pmid:29537671
  11. 11. Murray AL, Napotnik JA, Rayner JS, Mendoza A, Mitro B, Norville J, et al. Evaluation of consistent use, barriers to use, and microbiological effectiveness of three prototype household water treatment technologies in Haiti, Kenya, and Nicaragua. Sci Total Environ. 2019;
  12. 12. Hunter PR, Zmirou-Navier D, Hartemann P. Estimating the impact on health of poor reliability of drinking water interventions in developing countries. Sci Total Env [Internet]. 2009;407(8):2621–4. Available from:
  13. 13. Rosa G, Clasen T. Estimating the scope of household water treatment in low- and medium-income countries. Am J Trop Med Hyg [Internet]. 2010;82(2):289–300. Available from: pmid:20134007
  14. 14. National Population and Family Planning Board, Statistical Indonesia, Ministry of Health—Kemenkes, ICF. Indonesia Demographic and Health Survey 2017 [Internet]. Jakarta, Indonesia; 2018. Available from:
  15. 15. Ministry of Health. Indonesia Demographic and Health Survey 2012. Natl Popul Fam Plan Board [Internet]. 2013; Available from:
  16. 16. Daniel D, Pande S, Rietveld L. The effect of socio-economic characteristics on the use of household water treatment via psychosocial factors: a mediation analysis. Hydrol Sci J. 2020;
  17. 17. Krejcie R V., Morgan DW. Determining Sample Size for Research Activities. Educ Psychol Meas. 1970;30(3):607–10.
  18. 18. Christen A, Duran Pacheco G, Hattendorf J, Arnold BF, Cevallos M, Indergand S, et al. Factors associated with compliance among users of solar water disinfection in rural Bolivia. BMC Public Health [Internet]. 2011;11(1):210. Available from: pmid:21463508
  19. 19. Freeman MC, Trinies V, Boisson S, Mak G, Clasen T. Promoting Household Water Treatment through Women’s Self Help Groups in Rural India: Assessing Impact on Drinking Water Quality and Equity. PLoS One. 2012;7(9). pmid:22957043
  20. 20. Tidwell JB, Gopalakrishnan A, Lovelady S, Sheth E, Unni A, Wright R, et al. Effect of Two Complementary Mass-Scale Media Interventions on Handwashing with Soap among Mothers. J Health Commun. 2019;24(2):203–15. pmid:30912707
  21. 21. Arriola KRJ, Ellis A, Webb-Girard A, Ogutu EA, McClintic E, Caruso B, et al. Designing integrated interventions to improve nutrition and WASH behaviors in Kenya. Pilot Feasibility Stud. 2020;6(1).
  22. 22. Fotue Totouom AL, Sikod F, Abba I. Household Choice of Purifying Drinking Water in Cameroon. Environ Manag Sustain Dev [Internet]. 2012;1(2):101–15. Available from:
  23. 23. Nauges C, Van Den Berg C. Perception of Health Risk and Averting Behavior: An Analysis of Household Water Consumption in Southwest Sri Lanka. TSEWP working paper series. Toulouse; 2009.
  24. 24. Dubois AE, Crump JA, Keswick BH, Slutsker L, Quick RE, Vulule JM, et al. Determinants of use of household-level water chlorination products in rural Kenya, 2003–2005. Int J Env Res Public Heal [Internet]. 2010;7(10):3842–52. Available from:
  25. 25. Figueroa M, Kincaid D. Social, cultural and behavioral correlates of household water treatment and storage. Cent Publ HCI 2010–1 Heal Commun Insights, Balt Johns Hopkins Bloom Sch Public Heal Cent Commun Programs [Internet]. 2010;0–55. Available from:
  26. 26. Roma E, Bond T, Jeffrey P. Factors involved in sustained use of point-of-use water disinfection methods: A field study from Flores Island, Indonesia. J Water Health. 2014;12(3):573–83. pmid:25252361
  27. 27. Opryszko MC, Majeed SW, Hansen PM, Myers JA, Baba D, Thompson RE, et al. Water and hygiene interventions to reduce diarrhoea in rural Afghanistan: A randomized controlled study. J Water Health. 2010;8(4):687–702. pmid:20705980
  28. 28. Kley LS- van der, Reijerkerk L. Water a way of life. CRC Press/Balkema. Leiden: CRC Press/Balkema; 2009. 150 p.
  29. 29. Waterworth P, Pescud M, Braham R, Dimmock J, Rosenberg M. Factors influencing the health behaviour of indigenous Australians: Perspectives from support people. PLoS One. 2015;10(11):1–17. pmid:26599437
  30. 30. Behailu BM, Pietilä PE, Katko TS. Indigenous Practices of Water Management for Sustainable Services. SAGE Open. 2016;6(4):215824401668229.
  31. 31. Goldman N, Pebley AR, Beckett M. Diffusion of ideas about personal hygiene and contamination in poor countries: Evidence from Guatemala. Soc Sci Med. 2001;52(1):53–69. pmid:11144917
  32. 32. Alexander CC, Shrestha S, Tounkara MD, Cooper S, Hunt L, Hoj TH, et al. Media access is associated with knowledge of Optimalwater, sanitation and hygiene practices in Tanzania. Int J Environ Res Public Health. 2019;16(11).
  33. 33. Rabbi SE, Dey NC. Exploring the gap between hand washing knowledge and practices in Bangladesh: a cross-sectional comparative study. BMC Public Health. 2013;13:89. pmid:23363772
  34. 34. Mosler H-J. A systematic approach to behavior change interventions for the water and sanitation sector in developing countries: A conceptual model, a review, and a guideline. Int J Environ Health Res. 2012;22(5):431–49. pmid:22292899
  35. 35. Lilje J, Kessely H, Mosler HJ. Factors determining water treatment behavior for the prevention of cholera in Chad. Am J Trop Med Hyg. 2015;93(1):57–65. pmid:25918206
  36. 36. Stocker A, Mosler HJ. Contextual and sociopsychological factors in predicting habitual cleaning of water storage containers in rural Benin. Water Resour Res. 2015;51(4):2000–8.
  37. 37. Seimetz E, Boyayo AM, Mosler HJ. The influence of contextual and psychosocial factors on handwashing. Am J Trop Med Hyg. 2016;94(6):1407–17. pmid:27139449
  38. 38. Daniel D, Diener A, Pande S, Jansen S, Marks S, Meierhofer R, et al. Understanding the effect of socio-economic characteristics and psychosocial factors on household water treatment practices in rural Nepal using Bayesian Belief Networks. Int J Hyg Environ Health. 2019 Jun 1;222(5):847–55. pmid:31047815
  39. 39. Dockx K, Van Remoortel H, De Buck E, Schelstraete C, Vanderheyden A, Lievens T, et al. Effect of contextualized versus non-contextualized interventions for improving hand washing, sanitation, and health in rural tanzania: Study design of a cluster randomized controlled trial. Int J Environ Res Public Health. 2019;16(14).
  40. 40. Reygadas F, Gruber JS, Dreizler L, Nelson KL, Ray I. Measuring user compliance and cost effectiveness of safe drinking water programs: A cluster-randomized study of household ultraviolet disinfection in Rural Mexico. Am J Trop Med Hyg. 2018;98(3):824–34. pmid:29363448
  41. 41. Schmidt W, Cairncross S. Household Water Treatment in Poor Populations: Is There Enough Evidence for Scaling up Now? Household Water Treatment in Poor Populations: Is There Enough Evidence for Scaling up Now? Environ Sci Technol. 2009;43(4):986–92. pmid:19320147
  42. 42. Parvez SM, Azad R, Pickering AJ, Kwong LH, Arnold BF, Rahman MJ, et al. Microbiological contamination of young children’s hands in rural Bangladesh: Associations with child age and observed hand cleanliness as proxy. PLoS One. 2019;14(9). pmid:31504064
  43. 43. Pearl J. Probabilistic reasoning in intelligent systems: networks of plausible inference. San Francisco, CA, USA: Morgan Kaufmann Publishers Inc.; 1988. 552 p.
  44. 44. Nadkarni S, Shenoy PP. A causal mapping approach to constructing Bayesian networks. Decis Support Syst. 2004;38(2):259–81.
  45. 45. Vyas S, Kumaranayake L. Constructing socio-economic status indices: How to use principal components analysis. Health Policy Plan. 2006;21(6):459–68. pmid:17030551
  46. 46. Houweling TAJ, Kunst AE, Mackenbach JP. Measuring health inequality among children in developing countries: Does the choice of the indicator of economic status matter? Int J Equity Health. 2003;2:1–12. pmid:12605720
  47. 47. Druzdzel MJ, Sowinski T. GeNIe Modeler [Internet]. Pittsburgh; 1995. Available from:
  48. 48. Do CB, Batzoglou S. What is the expectation maximization algorithm? Nat Biotechnol. 2008;26(8):897–9.
  49. 49. Greiner M, Pfeiffer D, Smith RD. Principles and practical application of the receiver-operating characteristic analysis for diagnostic tests. Prev Vet Med. 2000;45(1–2):23–41. pmid:10802332
  50. 50. Manstead ASR. The psychology of social class: How socioeconomic status impacts thought, feelings, and behaviour. Br J Soc Psychol. 2018;57:267–91. pmid:29492984
  51. 51. Braveman P, Gottlieb L. The Social Determinants of Health: It ‘ s Time to Consider the Causes of the Causes. Public Health Rep. 2002;129:19–31.
  52. 52. Allen E, Morazan IM, Witt E. Actively engaging women is helping solve the global water crisis. J Water, Sanit Hyg Dev [Internet]. 2018;8(4):632–9. Available from:
  53. 53. Contzen N, Mosler H. RANAS (Risks, Attitudes, Norms, Abilities, and Self‐regulation) methodological fact sheets—6 methodological fact sheets on behavior change. Dübendorf, Switzerland: Eawag, Swiss Federal Institute of Aquatic Science and Technology; 2015.
  54. 54. Wilson Van Voorhis CR, Morgan BL. Understanding Power and Rules of Thumb for Determining Sample Sizes. Tutor Quant Methods Psychol. 2007;3(2):43–50.
  55. 55. Clasen T, Mclaughlin C, Nayaar N, Boisson S, Gupta R, Desai D, et al. Microbiological Effectiveness and Cost of Disinfecting Water by Boiling in Semi-urban India. Am J Trop Med Hyg. 2008;79(3):407–13. pmid:18784234
  56. 56. Cohen A, Tao Y, Luo Q, Zhong G, Romm J, Colford JM, et al. Microbiological evaluation of household drinking water treatment in rural China shows benefits of electric kettles: A cross-sectional study. PLoS One. 2015;10(9):1–16. pmid:26421716