Working Papers in Environmental Social Sciences · disease burden (10.9% of total disability adjusted life years; 2010b), and handwashing with soap is the ... with water and soap

Working Papers in

Environmental Social Sciences

Handwashing behavior change: infrastructural and commitment interventions in the Borena Zone, Ethiopia Nadja Contzen and Hans-Joachim Mosler

Working Paper 2013-06

Working Papers in Environmental Social Sciences

Department of Environmental Social Sciences Eawag: Swiss Federal Institute of Aquatic Science and Technology Überlandstr. 133 8600 Dübendorf Switzerland http://www.eawag.ch/forschung/ess

Recommended Citation

Contzen, N & Mosler H-J. (2013). Handwashing behavior change: infrastructural and commitment interventions in the Borena Zone, Ethiopia. Working Papers in Environmental Social Sciences 2013-06, Department of Environmental Social Sciences, Swiss Federal Institute of Aquatic Science and Technology (Eawag), Dübendorf, Switzerland. http://www.eawag.ch/forschung/ess/workingpapers/.

Author Information

Nadja Contzen is a PhD student at Eawag and University of Zurich. Hans-Joachim Mosler is a senior researcher at Eawag and titular professor at the University of Zurich. As members of the Environmental Health Psychology group at Eawag, their research focuses on health promotions in developing countries and health behavior change.

Author Contact

Eawag, Überlandstr. 133, 8600 Dübendorf, Switzerland [email protected] +41 58 765 5464 http://www.eawag.ch/about/personen/homepages/contzena/

1

Executive summary

Background and objective

In Ethiopia, as well as in many other developing countries, diarrhea is one of the leading causes of

disease burden (10.9% of total disability adjusted life years; 2010b), and handwashing with soap is the

single most effective prevention (Curtis, Cairncross, & Yonli, 2000; Curtis et al., 2011). However,

handwashing rates are considerably low in Ethiopia, with only around 17% of primary caregivers

washing their hands with soap after defecation, and with the majority washing their hands with only

water before eating (Federal Ministry of Health Ethiopia, 2011a, 2011b). Therefore, handwashing

promotions are the focus of many health interventions by local and international relief and

development agencies (Federal Ministry of Health Ethiopia, 2011a, 2011b). These campaigns have

rarely been grounded in theory (Aboud & Singla, 2012). Moreover, they have often been limited to

raising awareness and providing knowledge, while their effectiveness has seldomly been verified.

The present research project, conducted in cooperation with Oxfam America, aimed to select new

theory- and evidence-based promotion activities, and to implement and test them in the Borena zone

of Ethiopia. According to the Risk, Attitudes, Norms, Abilities, and Self-regulation behavioral change

model (Mosler, 2012), it was assumed that, in order to be effective, promotion activities must influence

a set of behavioral factors (i.e., attitudes or norms), which, in turn, change behavior. Thus, in a first

step, a baseline study explored the behavioral factors that are most important in explaining

handwashing with soap. On the basis of this result, in a second step, two handwashing promotions

were selected that were expected to target the key determinants of handwashing: an appeal to

construct tippy taps with maintenance-planning, and primary caregivers’ public commitment (see

below for more details). A participatory f-diagram exercise served as a base intervention. Next, these

promotions were implemented in the field by a local non-governmental organization, Gayo Pastoral

Development Initiative. In a final step, the intervention potential of the promotion activities was tested.

Methods

A longitudinal research design was applied, with two data collection waves; a baseline in early 2012

and a follow-up in early 2013. The selection and implementation of the promotion activities was

conducted in between these two time points.

Four rural Kebeles (the lowest administrative unit in Ethiopia) in the Borena zone served as research

areas: Magole and Kancharo in Dillo district, and Gololcha and Madacho in Dire district. While the

base intervention was implemented in all four Kebeles, the tippy tap construction, and the public

commitment intervention were implemented in a fully crossed design. That is, Kebele 1 received only

the base intervention, Kebele 2 received the public commitment intervention in addition to the base

intervention, Kebele 3 received the tippy tap intervention in addition to the base intervention, and

Kebele 4 received all interventions. Further, half of the households in which a tippy tap was

constructed received a maintenance-planning intervention, while the other half did not.

2

Data were collected by means of structured observations and interviews with primary caregivers in

each household participating in the study. At baseline, primary caregivers of 463 households were

interviewed. Of these, 426 households were maintained at follow-up (dropout rate of 8%). As primary

caregivers of 128 additional households were interviewed at follow-up, the sample size for the

interviews at this time was 554. Observations were conducted in 151 out of the 463 households at

baseline. Of these, 141 were maintained at follow-up (dropout rate of 7%). As all households were

observed at follow-up, the sample size at this time was also 554.

Data were analyzed by means of multiple linear regression and mediation analyses, multivariate

analysis of variance, Kruskal-Wallis and Jonckheere–Terpstra tests.

Results

Baseline results revealed social norms as being an especially crucial determinant of handwashing. In

addition, the prevalence of a rather impractical handwashing practice, involving several vessels and

steps, was observed, while designated places and facilities for handwashing were barely present (only

8% of the respondents had a designated place for handwashing). Further, observations suggested

that forgetting to wash hands was common, and that the persons under study were unaware that they

failed to remember.

On the basis of these results, commitment meetings were organized as a first intervention, whereby

beneficiaries publicly committed to washing their hands at key times and received a scarf as a sign of

their commitment, as well as a commitment certificate. While this intervention was primarily selected to

strengthen social norms and commitment, it was hoped that the scarf and certificate may also serve

as handwashing reminders. As a second intervention, households were motivated to construct tippy

taps to facilitate performance of the behavior, and also to serve as reminders for handwashing. A

supplementary intervention was implemented to facilitate the maintenance of the tippy tap: within the

maintenance-planning intervention, beneficiaries were motivated to schedule and write down when

they planned to re-fill the tippy tap during the day, and with how much water, and when they planned

to re-fill the soap during the week. In the control intervention, beneficiaries learned the f-diagram route

of contamination in a participatory group sorting task.

Due to malfunctioning monitoring, the implementation quality was sometimes questionable. The

quality was least guaranteed in the Kebele receiving the public commitment in addition to the base

intervention.

Longitudinal analyses revealed that the tippy tap intervention alone and in combination with public

commitment had a positive outcome: (1) observed handwashing behavior tended to be higher in the

follow-up than in the baseline for these groups, or at least did not decline; (2) changes in self-reported

handwashing behavior from baseline to follow-up was positive for these groups. Further, almost 100%

of the households that were asked to construct a tippy tap actually did so, and recognized it as their

designated place for handwashing. Moreover, the vast majority of these households had a functioning

tippy tap, with water and soap being present, at follow-up, 2-4 months after completion of

implementation. In addition, the previously mentioned positive outcomes tended to be higher for the

groups that experienced the maintenance-planning, compared to those that did not.

3

In contrast, in the group that experienced only the f-diagram, the behavior had a tendency to decline,

or was carried out at a low level at follow-up. In addition, barely any of these households had a

designated place for handwashing, and soap and water was not easily accessible in more than half of

the households.

Although less pronounced, the same tendencies were found for the public commitment-only

intervention group. Again, these results might express the fact that the public commitment had no

impact. Alternatively, it is credible that the public commitment-only intervention was unsatisfactorily

implemented, and was accordingly unable to fulfill its potential. In both cases, the promotion would not

have been capable of slowing natural behavior reduction (see chapter 4.2 for a discussion of these

findings).

With regard to the mechanism of behavior change, mediation analyses revealed that, when compared

with the base intervention, the tippy tap-only and the combined intervention were associated with the

following behavioral factors: norms, hindrance, action planning and remembering. However, hindrance

and action planning did not explain behavior. Therefore, it emerged that only norms and forgetting

significantly mediated the association between behavior and tippy tap-only, and the combined

intervention. Public commitment-only had no associations with any of the tested behavioral factors,

and thus, neither with behavior, when compared to the f-diagram-only.

Conclusion and recommendations

The present research project tested the effectiveness of a public commitment intervention and the

promotion of tippy taps. The implementation of the public commitment intervention alone had no

influence on handwashing behavior. However, since there is evidence that this specific intervention

was not implemented strictly according to instructions, it is doubtful that public commitment should be

abandoned as a handwashing intervention. Further research is necessary to better assess the

intervention potential of public commitment to promote handwashing behavior, and careful intervention

implementation, according to instructions, and close monitoring are crucial. The latter, of course, is

true of every intervention.

More importantly, the research project revealed the promising potential of a tippy tap promotion:

almost all study households were successfully motivated to construct a tippy tap, and nearly all of

these households recognized it as their designated place for handwashing. Moreover, in a substantial

number of households, the tippy tap was functioning, with water and soap being present, 2-4 months

after intervention termination. Maintenance-planning seemed to aid the positive outcome. Therefore, it

can be highly recommended that the tippy tap intervention be further applied to promote handwashing

in the Borena zone; in combination with maintenance-planning, if possible.

To conclude, the tippy tap intervention successfully induced a crucial prerequisite for handwashing: a

designated place and facility for handwashing, in which soap and water is present.

4

Acknowledgements

This research was supported by funding from Oxfam America.

The authors especially thank Myra Foster, public health specialist for Oxfam America, for initiating,

supporting, and advising the research project.

Further thanks go to Kebede Eticha, public health specialist for Oxfam America, Horn of Africa

Regional Office and the Oxfam support office in Yabello, especially to Abarufa Jatani, former head of

the Oxfam support office.

Thanks are due to Gayo Pastoral Development Initiative for implementing the handwashing

promotions. Belay Aschalew, WASH coordinator for GDPI during the drought response, is especially

gratefully acknowledged for his continuous help and advice.

Special thanks go to Iara Meili, research assistant, for her continuous support, her critical thoughts

and her hard work during the follow-up.

The fieldwork for this project was only made possible by the great effort of numerous people: Sarah

Zgraggen, research assistant during the baseline; Wario Dima Godana and Chaka Yohannes Chaka,

Field Research Coordinators; the data collectors, especially Ibsa Arero, Abdi Wako and Jarso Dokata;

the Kebele leaders of the study Kebeles; and the community members who participated in the study

and generously shared with us their time, thoughts, and experience.

5

Table of Contents

Executive summary ................................................................................................................................. 1 

Acknowledgements ................................................................................................................................. 4 

Abbreviations ........................................................................................................................................... 7 

1  Introduction ...................................................................................................................................... 8 

1.1  Diarrheal and respiratory diseases in Ethiopia, malnutrition, the importance of handwashing and its prevalence ....................................................................................................... 8 

1.2  The Borena zone in southern Ethiopia: pastoral living, drought responses and handwashing campaigns ............................................................................................. 8 

1.3  Results from a preliminary study: evaluating Oxfam handwashing promotions in the 2011-2012 drought response in Borena, Ethiopia ................................................................. 10 

1.4  Theoretical background and research questions .......................................................... 13 

2  Method ........................................................................................................................................... 14 

2.1  Research design ....................................................................................................... 14 

2.2  Selection of the study areas and study households ...................................................... 14 

2.3  Data collection method and sample size ..................................................................... 14 

2.3.1  2-3-hour household observations ........................................................................ 16 

2.3.2  Face-to-face interviews – questionnaire and measures.......................................... 18 

2.4  Data analysis procedure ............................................................................................ 19 

3  Results ........................................................................................................................................... 21 

3.1  Socio-demographic variables ..................................................................................... 21 

3.2  Results from the baseline survey ................................................................................ 21 

3.2.1  Frequency of handwashing at key times and their factor structure .......................... 21 

3.2.2  Behavioral factors explaining handwashing behavior ............................................. 23 

3.2.3  Prevalence of behavioral factors ......................................................................... 25 

3.2.4  Qualitative results: some observations ................................................................. 26 

3.3  Selection of handwashing interventions and their implementation process ..................... 27 

3.3.1  Public commitment intervention ........................................................................... 28 

3.3.2  Promotion of constructing tippy taps .................................................................... 29 

3.3.3  Intervention implementation plan and base intervention ......................................... 31 

3.3.4  Implementation process, implementation personnel, and their training .................... 33 

3.4  The implementation process: some qualitative results .................................................. 33 

3.4.1  Some incidents and inconsistencies in the implementation process ........................ 33 

3.4.2  A specific shortcoming in the implementation process: the public commitment sign .. 34 

3.4.3  Interference in the data collection ........................................................................ 35 

3.5  Reach of the promotion activities ................................................................................ 35 

3.6  Influence of the handwashing interventions on selected outcome variables .................... 36 

3.6.1  Influence on observed behavior ........................................................................... 36 

3.6.2  Influence on self-reported behavior ...................................................................... 38 

6

3.6.3  Proxy measures: tippy taps in function ................................................................. 40 

3.6.4  Influence of maintenance-planning on self-reported behavior and presence of soap and water .......................................................................................................... 42 

3.7  Mechanism of change: Psychological variables explaining the influence of the handwashing interventions on change in self-reported handwashing ............................. 43 

4  Discussion ..................................................................................................................................... 45 

4.1  Baseline results: behavioral factors to be addressed and derived promotion activities ..... 45 

4.2  Longitudinal study: summary of the interventions’ effect ............................................... 45 

4.3  Limitations ............................................................................................................... 48 

5  Conclusions and recommendations .............................................................................................. 50 

References ............................................................................................................................................ 51 

Annex ..................................................................................................................................................... 54

7

Abbreviations

BCA Behavior Change Agent

CHV Community Health Volunteer

DLHS District Level Health Supervisor

FDO F-diagram intervention only

FRC Field Research Coordinator

GHD Global Handwashing Day

GPDI Gayo Pastoral Development Initiative

HEW Health Extension Worker

NGO Non-governmental organization

OA Oxfam America

PCO Public commitment intervention only

PC+TT Public commitment and tippy tap intervention

PC+TTMP Public commitment and tippy tap with maintenance-planning intervention

TTO Tippy tap intervention only

TTOMPO Tippy tap intervention with maintenance-planning only

UN United Nations

UNICEF United Nations International Children's Emergency Fund

WASH Water, Sanitation and Hygiene

WC WASH Coordinator

WHO World Health Organization

8

1 Introduction

1.1 Diarrheal and respiratory diseases in Ethiopia, malnutrition, the importance of handwashing and its prevalence

Ethiopia has a mortality rate of 188 per 1000 live births in those aged under 5 years (WHO Regional

Office for Africa, n.d.). Diarrhea is the leading cause of these deaths (23% of all deaths in children

aged under 5), followed by pneumonia (15%; WHO Regional Office for Africa, 2010b). Moreover,

diarrheal disease is greatly interlinked with child malnutrition, which is more prevalent in Ethiopia than

in its surrounding countries (34.6% and 22%, respectively of underweight children under five; WHO

regional Office for Africa, 2010a). Diarrhea is not only particularly lethal in malnourished children, it is

also a major cause of malnutrition, which makes children more vulnerable to other diseases (Bartram

& Cairncross, 2010; Prüss-Üstün, Bos, Gore, & Bartram, 2008; UNICEF & WHO, 2009). With respect

to the entire Ethiopian population, diarrhea and pneumonia are the leading causes of disease burden

(10.9% and 14.3% of total disability adjusted life years, respectively; WHO Regional Office for Africa,

2010b). In recent years, several outbreaks of acute watery diarrhea have hit the country (WHO, n.d.),

and, in addition to the element of human suffering, it is estimated that Ethiopia loses up to US$ 650

million each year, as a result of diarrheal and other waterborne diseases (Federal Ministry of Health

Ethiopia, 2011b). With a reduction rate of 40%, the single most effective preventive measure against

diarrheal disease is handwashing with soap at key times (Curtis et al., 2000; Curtis et al., 2011), which

are after defecation or after wiping a child’s bottom (feces-related handwashing), and before eating,

preparing food, or feeding or breastfeeding a child (food-related handwashing). Furthermore,

handwashing lowers rates of childhood pneumonia and other respiratory illnesses by up to 50%

(Curtis et al., 2011; Luby et al., 2005; Luby & Halder, 2008). Malnutrition is also reduced, not only

indirectly through lower diarrhea rates: there is growing evidence to indicate that tropical enteropathy

caused by fecal bacteria is a major cause of malnutrition (Humphrey, 2009), and the best protection

against fecal bacteria is safe feces disposal and handwashing with soap after contact with feces

(Curtis et al., 2000; Humphrey, 2009).

In spite of its preventive power, handwashing is uncommon in Ethiopia, although comparable to rates

in most developing countries (Iyer, Sara, Curtis, Scott, & Cardosi, 2005; Scott, Curtis, Rabie, &

Garbrah-Aidoo, 2007): only around 17% of primary caregivers wash their hands with soap after

defecation, and the majority wash their hands with only water before eating (Federal Ministry of Health

Ethiopia, 2011a, 2011b). Accordingly, handwashing campaigns are high on the agendas of health,

relief, and development agencies (Federal Ministry of Health Ethiopia, 2011a, 2011b).

1.2 The Borena zone in southern Ethiopia: pastoral living, drought responses and handwashing campaigns

More specifically, in the Borena zone in Ethiopia, handwashing promotions started in 2006, when

CARE Ethiopia launched a first handwashing intervention (Aschalew, 2013). Since then, the Ethiopian

government and many NGOs, in addition to CARE, such as GOAL Ethiopia, Action For Development

(AFD), or Oxfam America (OA) through Gayo Pastoral Development Initiative (GPDI), have repeatedly

implemented handwashing promotions.

9

The Borena zone is located in the southern part of Ethiopia at an altitude of 1,000 to 1,600 meters

above sea level.1 The region is semi-arid, with 70% of the area being sparsely wooded grassland. The

primary ethnic group in the region is the Borena, which is part of the larger Oromo ethnic group. Their

main livelihood is pastoralism. The Borenas are semi-nomadic, with men seasonally migrating with

their cattle, looking for pasture.2 The water supply coverage is low (20% to 35%; ITAB CONSULT

PLC, 2011), meaning long distances must be walked for water collection, which is the responsibility of

women and girls. The work burden of Borena women is high, encompassing (in addition to water

collection) cooking, collecting fodder for calves and other small animals, and collecting fire wood. The

Borena culture embraces an old and complex political system, the Gada system. This institution

manages the rangeland utilization and natural resources, and contains important social support

mechanisms. These are especially important, since the region is regularly hit by droughts threatening

the survival of livestock and causing food insecurity and famine, such as in 1984–85 (region-wide);

1999–92 (region-wide); 1993 (Moyale); 1996–97 (Teltele, Dire, Moyale, Yabello, Arero, Shakiso, and

Liben); 1999–2000 (lowland Woredas); and 2005–06, 2008–09, and 2011 (region-wide). The

government and NGOs have responded to these droughts with relief food distributions, the

establishment of early warning systems, and development projects, with the first food aid being given

in 1974.

The last drought in 2010-2011 was caused by the failure of two consecutive rain seasons and the

effect of La Nina, leading to a critical emergency situation from October 2010 up to November 2011.

This included a water and food shortage, as well as increased malnutrition and incidence of diseases,

and escalated the number of people affected, and requiring humanitarian assistance. Reports by

NGOs and UN agencies from July 2011 described alarming situations, particularly in the Oromia and

Somali regions of Ethiopia, where child malnutrition was in the tens of thousands, and the overall

number of people in need of food assistance was much higher than indicated in the reports by the

government; 4.5 million people were in need in the country. In collaboration with GPDI, OA

implemented an emergency WASH3 response in four Woredas (districts) in the Borena zone of

Southern Oromia, Ethiopia: Moyale, Teltele, Dillo, and Dire. Part of the WASH response was the

promotion of handwashing with an agent4 at key times. A preliminary research project conducted in

2012 evaluated these handwashing promotions.

1 This paragraph is based on the following work: Debsu, D. N.(2013) . Local institutions, external interventions, and adaptations to climate variability: The case of the Borena pastoralists in southern Ethiopia. In K. Pfeifer (Ed.), Oxfam America Research Backgrounder series. 2 In addition, during droughts men tend to migrate in search of jobs, whereas women stay to care for the family and the farm. 3 WASH stands for Water, Sanitation and Hygiene. The Global WASH Cluster, led by UNICEF, was established as part of the international humanitarian reform programme, and provides an open, formal platform for all emergency WASH actors to work together (http://www.washcluster.info/). The cluster made available a variety of resources and training materials to develop hygiene promotion activities and capacities at the onset of an emergency. 4 GPDI promoted the washing of hands with soap and to use ash or sand if no soap is available. For simplicity, in the following text “handwashing” stands for handwashing with an agent.

10

1.3 Results from a preliminary study: evaluating Oxfam handwashing promotions in the 2011-2012 drought response in Borena, Ethiopia

The promotion activities implemented by GPDI were akin to the promotions compiled by the Global

WASH cluster (UNICEF, 2007), and primarily focused on knowledge provision. It was explained how

diarrhea is transmitted (by means of the f-diagram) and what its effect is on the body. Further,

prevention and treatment methods were also explained. With regard to prevention, four key messages

were disseminated around handwashing: Wash your hands with soap (1) …after using the latrine, (2)

…before food preparation, (3) …before eating, and (4) …before breastfeeding and feeding children.

In addition to knowledge provision, coping with barriers was fostered by instructing the beneficiaries to

use ash or sand if no soap is available. Action knowledge was promoted by handwashing

demonstrations, handwashing behavioral trials, and by observing and correcting handwashing

behavior.

The promotion content was disseminated via nine different promotion activities or channels, which are

summarized in Table 1.

Table 1

Overview of promotion activities

Promotion activity Description

Home visit Main goal: disseminate the key messages and to observe and correct behavior. Material: F-diagram and pictures. Conducted by HEW and/or CHV.

Health education at gatherings

Health education at community meetings, cash-for-work sites, women’s cooperative meetings, and coffee ceremonies. Main goal: disseminate the key messages, show how to wash hands and mobilize the community by encouraging the participants to talk with their neighbors about health issues. Material: F-diagram, posters, and leaflets. Conducted by HEW, CHV, and GPDI.

Health information at the market, spread by megaphone

Mass health education by means of megaphone. Main goal: disseminate the key messages. Conducted by GPDI.

GHD activities Activities at the GHD. Main goal: disseminate the key messages, mobilize the community and demonstrate behavior. Activities: drama, poems, handwashing demonstration and joint handwashing, distribution of soap and t-shirts. Conducted by GPDI, with support from the HEW and CHV.

Material distribution with handwashing demonstration

Mass distribution of PUR, soap, and buckets. Main goal: provide material, explain how it is used, and why it should be used, dissemination of the key messages. Conducted by GPDI.

Handwashing demonstration

Demonstration of how to wash hands with soap. Main goal: disseminate action knowledge, and the key messages. Conducted by GPDI.

F-Diagram Used at home visits and during health education. Main goal: explain how diarrhea is transmitted.

Picture Used at home visits, and during health education. Main goal: show when and how to wash hands with soap.

Leaflet Distributed at community meetings. Main goal: explain diarrhea prevention methods, including washing hands with soap.

11

While GPDI workers implemented some of the promotion activities themselves, other activities were

carried out by the Health Extension Workers (HEW) and the Community Health Volunteers (CHV) in

each Kebele5. For these purposes, GPDI held trainings with the HEW and CHV in each Woreda to

pass on the required health knowledge and explain the different promotional approaches.

Although all the activities disseminated the same key messages, and all were aimed at changing

handwashing behavior, their effectiveness differed.

The overall associations between the promotion activities and self-reported feces- and food-related

handwashing (the total effects) are displayed in Figure 1 and 2. Only three promotion activities had a

significant total association with feces- and food-related handwashing. Significant positive associations

were found for receiving a leaflet and for taking part in a material distribution: beneficiaries that

received a diarrhea leaflet or participated in a material distribution tended to wash their hands more

often than those who did not receive a diarrhea leaflet. Global Handwashing Day (GHD) was also

significantly associated with handwashing frequency, however, negatively. Those who participated in

the GHD activities tended to wash their hands less often than those who did not take part. The

remaining six promotion activities were not significantly associated with feces- and food-related

handwashing.

Figure 1. Total unstandardized effects of promotion activities on self-reported feces-related handwashing. N = 443. Dark blue bars represent significant unstandardized regression coefficients with p < .05.

Therefore, only diarrhea leaflets and material distributions were positively associated with

handwashing, and thus have the potential to effectively promote handwashing. One reason for the

effectiveness of diarrhea leaflets, amongst others, might be that, when hung up in the house, such a

leaflet may serve as a prompt and deliver information continuously. In terms of material distributions,

the result parallels the findings from a study conducted in Haiti, where material distributions were also

positively associated with handwashing (Contzen & Mosler, 2013). This promotion activity has the

advantage of providing the beneficiaries with the required materials and, probably more importantly,

fosters the beneficiaries’ behavioral competence by demonstrating correct behavior.

5 Lowest administrative unit in Ethiopia.

12

In contrast, GHD activities were negatively associated with handwashing and might even have

hindered this behavior. Respondents’ evaluation of the promotion activities corresponded to the

negative association: while the beneficiaries rated the applied promotion activities very favorably

overall6, they liked the GHD activities slightly less than the other activities. Although the observed

association is contrary to expectations and contradicts the purpose of the promotion activity, the

finding is in line with results from Haiti, where GHD activities also had a negative association with self-

reported handwashing (Contzen & Mosler, 2013).

Figure 2. Total unstandardized effects of promotion activities on self-reported food-related handwashing. N = 443. Dark blue bars represent significant unstandardized regression coefficients with p < .05.

None of the remaining promotion activities; home visits, health education provided at gatherings,

health information provided at the market, handwashing demonstrations, an f-diagram, and pictures,

were significantly associated with handwashing; in their current form they do not appear to have the

potential to effectively change behavior.

Given the importance of handwashing in preventing diarrhea, these mixed results are disquieting and

call for alternative handwashing promotions beyond knowledge provision. Therefore, the goal of the

present research project, which is part of the recovery phase of Oxfam’s drought response of 2010-

2011, was to select new, evidence-based, and potentially more effective promotion activities, and

implement and test these in the Borena zone.

6 Mean levels of liking, conviction, and trustworthiness were M = 3.20 and higher on a scale ranging from 0 to 4.

13

1.4 Theoretical background and research questions

Effective hygiene promotions change unhealthy, unhygienic behavior into healthy, hygienic behavior,

by altering behavioral factors within the individuals concerned. Therefore, a constellation of behavioral

factors must be present in the individual for a target behavior to be sustainably adopted.

Consequently, those hygiene promotions that affect these critical behavioral factors are most effective

in changing behavior. Behavioral factors potentially relevant in changing behavior were specified in a

recent model of behavior change; the RANAS model (Mosler, 2012)7, which integrates the behavioral

factors proposed by major theories of behavior change, such as the Health Belief Model (Rosenstock,

1974), Protection Motivation Theory (Floyd, Prentice-Dunn, & Rogers, 2000), or the Theory of Planned

Behaviour (Ajzen, 1991), into a comprehensive model. These RANAS factors can be classified into

five factor blocks: risk factors, attitude factors, norm factors, ability factors, and self-regulation factors.

A description of the five factor blocks is shown in Table 2.

Table 2

Overview of the RANAS model

Factor block Description

Risk factors Perceived vulnerability and perceived severity of contracting a disease, and factual knowledge about the possibility of being affected by a potential contamination.

Attitude factors Instrumental beliefs about costs and benefits of the targeted behavior, as well as affective beliefs, i.e., feelings arising when thinking about the behavior.

Norm factors Include different social influences: descriptive norms (behaviors typically performed by others), injunctive norms (behaviors typically approved or disapproved of by others) and personal norms (personal standards, what should be done).

Ability factors People’s perceptions of performing a behavior (perceived behavioral control) and the confidence in one’s ability to organize and manage the targeted behavior (self-efficacy).

Self-regulation factors

Help to manage conflicting goals and distracting cues when intending to implement and maintain a certain behavior. Important determinants are action control, coping planning, remembering the behavior, and commitment.

To accomplish the above specified research goal in the present study, the critical behavioral factors

first had to be specified; the behavioral factors explaining handwashing behavior were determined. On

the basis of this result, in a next step, handwashing promotions that were expected to target exactly

these behavioral factors were selected. The promotions were then implemented in four Kebeles in two

out of the four Woredas where GPDI was working during the drought response. The promotions’

effectiveness was tested longitudinally in a final step.

To summarize, the following research questions were addressed:

1. Which behavioral factors best explain handwashing with soap, and thus should be targeted by

handwashing promotions?

2. By means of which promotional activities is it expected that these behavioral factors will be

changed?

3. What is the intervention potential of the selected handwashing interventions on handwashing

behavior?

7 See Supplemental material 1 for the paper by Mosler (2012).

14

2 Method

2.1 Research design

To answer the above research questions, a longitudinal research design was applied. That is, data

were collected at two points in time, with a promotion phase in between. Figure 3 displays the different

study phases. Between the end of the implementation period and the beginning of the follow-up data

collection, there was a time lag of 2–4 months. A longitudinal design provides evidence for the effects

of an intervention.

Figure 3. Study phases.

In addition, the selected handwashing promotions were tested in the field in an experimental design,

with four different Kebeles receiving different promotions, or combinations of these, respectively (see

further below).

2.2 Selection of the study areas and study households

Two out of the four Woredas where GPDI was working were chosen as study areas, according to their

geographic location near Yabello; Dillo and Dire. Within each of the two Woredas two Kebeles were

selected, according to the following criteria: limited NGO operation, apart from GPDI, reasonable

access, and type of the foregoing OA drought response. For Dire, Madacho and Gololcha were

selected, and Magole and Kancharo were chosen in Dillo. The selection of the four Kebeles was

approved by the zonal health office and the two Woreda health offices after they had received detailed

information about the research project; all three offices promised their support for the project. Kebele

leaders and elders were also visited to inform them of the planned research project and to ask for their

support, which was readily provided.

Within the Kebeles, Ollas (rural villages) were chosen according to size and accessibility. Only

households with at least one child under the age of 5 years, and where the primary caregiver was

present during data collection, were included in the study. Although the plan had been to select every

third household within an Olla, in the majority of Ollas the number of households was too small to

apply this procedure. Thus, every available household was included.

2.3 Data collection method and sample size

Data were collected in February and March 2012 (T1) and in February and March 2013 (T2), by

means of 2- to 3-hour household observations and structured face-to-face interviews. At T1, primary

caregivers of 463 households were interviewed. Of these, 426 households were maintained.

Therefore, with only 37 dropouts, the dropout rate was very low (8%). Half of the 37 dropouts (i.e., 18

Phase 1 – 1/1/12 to 4/30/12

Baseline data collection and baseline study to determine relevant behavioral factors

Phase 4: 1/1/13 to 6/30/13

Follow-up data collection and longitudinal study to evaluate the implemented

Phase 2 – 1/5/12 to 30/6/12

Selection of handwashing promotions, based on the results of phase 1

Phase 3 – 1/7/12 to 1/16/13

Implementation of the selected handwashing promotions

15

households) had moved to other Kebeles, a quarter was out of the village during the data collection

days (i.e., nine households), and six households were unexpectedly not at home at the appointed

time. Only three primary caregivers explicitly refused to take part in the second data collection (< 1%

refusal rate). The low dropout rate was achieved by the following measures:

1. The support of the Kebele leaders and elders was gained prior to data collection.

2. Each household was visited one day prior to data collection to convince the primary caregiver

to participate. It was explained in detail how important it was to maintain the households

observed and members interviewed in the baseline study for the success of the project.

3. The main reason for primary caregivers being reluctant to re-participate was that they had

planned to go to fetch water during the appointed observation and interview time. Therefore,

the research team started to go to fetch water for the study households, whenever indicated.

This dramatically boosted the willingness to participate. However, it simultaneously led to

many additional logistical and physical efforts by the drivers and the research team, especially

the Field Research Coordinator (FRC), and was very time-intensive.

In one Kebele, Kancharo, in 2012, data collection was terminated earlier than scheduled; fewer data

were collected than planned, leading to a relatively small sample size within this Kebele. The earlier

termination was caused by major exhaustion8, which affected the data collection team, the driver and

the research team. In 2013, new households (i.e., 128) were recruited in Kancharo to increase the

sample size. Thus, the sample size at T2 was 554 for the interviews. Observations were conducted in

151 out of the 463 households at T1. Of these, 141 were maintained at T2 (a dropout rate of 7%). As

all households were observed at T2, the sample size for observations at this time point was also 554

(see Table 3 regarding the distribution of the sample over the Woredas and Kebeles). The new

households recruited at T2 were not included in the primary longitudinal analyses.

The primary caregiver in each household was interviewed, since this is the person who is responsible

for childcare and food preparation, and thus has the highest chance of passing on diarrheal diseases

to other family members. In addition, primary caregivers are thought to be capable of influencing the

hygiene behaviors of other members of the household, especially children, and other members of the

community, by acting as role models and passing on their knowledge and habits. Only households

with a child under the age of 5 years were considered for observation/interview. All participants

provided informed consent.

The interviews and observations were carried out by a team of 10 (T1) and 14 (T2) local students and

social workers. Although female applicants were explicitly encouraged to apply for the job, the

research teams consisted of only two females, but 8 and 12 males, respectively. In Ethiopia, women

are still disadvantaged in terms of education. Therefore, we were not successful in finding a greater

number of women who possessed the skills and qualifications to accomplish the job. Analyses testing

for data collector effects showed that the gender of the data collectors did not influence the answers or

observation pattern.

8 The exhaustion was due to early departure times (as early as 4 am), long travel times (up to 2.5 hours’ drive), hard working conditions, and an enormous work-overload, caused by a prescribed limited timeline. This experience meant that data collection was reorganized at T2, insofar as a greater number of days were appointed for data collection, and the entire team stayed in the field to avoid the early departure times and long travel distances.

16

In both years, the team was trained in interviewing and observation techniques in a workshop, prior to

data collection, and supervised during data collection by researchers and the FRC. Observations and

interviews are both described in the following sub-sections.

Table 3

Distribution of the sample over Woredas and Kebeles

Woredas

Dillo Dire

Kebeles Madacho Gololcha Total Magole Kancharo Total

Dat

a co

llect

ion

met

ho

d

T1 Interviews 135 167 302 132 29 161

Observations 48 49 97 46 8 54

T2 Revisited Interviews & Observations

127 153 280 123 23 146

New HH Interviews & Observations

- - - - 128 128

Total Interviews & Observations

127 153 280 123 151 274

Note. HH = household.

2.3.1 2-3-hour household observations

Potential households were visited 1 day before planned observations, and asked for permission to

conduct the observation. The specific objectives of the study were undisclosed to avoid reactivity. In

addition, observers were instructed to keep interaction with household members to a minimum during

observation. At T1 and T2, observations began immediately after the family woke up in the early

morning (around 6 am). At T2, in order to conduct observations in all households, additional

observations were conducted at around 11 am, when the primary caregivers returned from fetching

water, collecting firewood, etc., and started to prepare lunch for their families. The observers seated

themselves in such a way that they could view handwashing behavior at key times. The following

critical behaviors engaged in by the primary caregiver were observed: cleaning a child’s bottom after

its defecation, clearing up a child’s feces, primary caregiver going to the bush or latrine for defecation9,

food preparation, feeding or breastfeeding10 a child, primary caregiver eating, drinking, and water-

handling11. In the case of a critical event occurring, it was marked down in a structured format, along

with the information regarding whether one or two hands were washed with water before or after the

event, respectively, and whether soap, ash, or sand was used. During data preparation, observed

handwashing was calculated for each critical event (e.g., food preparation) separately, as the

percentage of critical events (e.g., food preparation), in which both hands were washed with soap,

9 This could be observed quite straightforwardly, insofar as beneficiaries most often go for defecation right outside the Olla’s fence. Thus, the proxy for observing primary caregiver defecation was the latter leaving the house and going to a latrine, or passing through the fence and squatting behind a nearby bush. 10 In the Borena zone, breastfeeding is not a sensitive behavior that is done only in private or in front of family members or females. Thus, it was possible for male data collectors to observe breastfeeding as easily as did female data collectors. 11 Handling drinking water means taking water out of the jerry can stored in the house. This is not to be confused with water-fetching, whereby women go to a water source to collect water. The latter behavior was not observed; it happened before or after observation (depending on the time of observation), as the water sources are rather distant from the households.

17

ash, or sand, out of all critical events (e.g. food preparation) that happened, in which the person

should have washed her hands.

In each household, the primary caregiver was also interviewed immediately after the observation was

completed.

At T1, observations were conducted for 3 hours in each household, followed by a 1-hour interview.

However, it emerged that 3-hour observations did not fit with the usual schedule of Borena women,

who finish their household work around 2 to 3 hours after rising. They then leave the house to fetch

water at water sources distant from the house, look after the small livestock or collect firewood or

fodder. Therefore, they leave the house for 1 to several hours. This often caused earlier termination of

observations at T1, and enormous efforts of persuasion were required to convince participants to stay

at least for the interview. Furthermore, when participants were willing to stay for the remaining

observation time, they had no work to do, and were often just sitting around, chatting with the data

collector, without engaging in any critical behavior. Consequently, the duration of observations was

reduced to 2 hours at T2.12

Due to the reduction in observation time and the additional observations at 11 am13, somewhat fewer

critical events were observed at T2 than at T1. Since observed handwashing was calculated as a

percentage (see above), the reduction in critical events observed should not have influenced the

observation measure.

The sample size of the longitudinal observation data was too small for the purpose of drawing

statistically rigorous conclusions.14 The reasons for the small sample size are as follows:

1. Due to budget and time constraints, it was decided that not all study households, but only 200

out of 600, would be observed.15

2. At T1, only 151 observations could be conducted due to (a) difficulties in obtaining the

required number of households in each Kebele because of low population density and high

absence of households during data collection days, and (b) the earlier termination of the data

collection in one Kebele, in Kancharo.

3. During the observations, critical events happened with low frequency. Therefore, at T1 and T2

feces-related events were observed in only around 50% of the households; feces-related

events were observed at both T1 and at T2 in only 38 households. Furthermore, the average

12 The reduction was necessary, although water was fetched by the research team for the majority of the study households. 13 While we could observe behavior at both points in time, primary caregiver defecation, for example, happens more often in the early morning than around midday. The frequency of other behaviors, such as cleaning a child’s bottom, is not dependent on the time of day. 14 When sample sizes are too small, the power of a test is also small. This means that the test cannot detect associations or differences that are present in reality. 15 At T2, a greater number of observations could be conducted because (1) more days were available for data collection, (2) observations were also conducted at midday, and (3) additional money that was not part of the research grant was allocated for the data collection.

18

event frequency was below one for feces-related events and between three and four for food-

related events, meaning that the data drew on a limited number of observed events.16

As a consequence, observation data could not be used for the primary longitudinal analyses. Rather,

these had to be realized by using self-reported data collected by structured interviews.

2.3.2 Face-to-face interviews – questionnaire and measures

At T1 and T2, structured interviews were conducted in Afan Oromo. Each interview took around 1

hour, and was based on a questionnaire that covered socio-demographic characteristics, self-reported

handwashing frequencies, and behavioral factors, such as attitudes and beliefs. In addition, previously

experienced promotion activities were measured. The questionnaire was prepared in English and was

translated into Afan Oromo at T1 by an Ethiopian social worker who had been employed by Eawag in

earlier projects, and who was very familiar with the type of research and the associated questionnaire.

The translation of each question was subsequently checked in terms of its adequacy by the FRC and

the project leader together; the exact meaning of each question in English was discussed and

compared to the meaning in Afan Oromo. For T2, some additional questions were added. These were

translated by one of the data collectors at T1, a linguist that was also familiar with the research

background and the type of questionnaire. A countercheck was conducted at T2 in the same way as at

T1.

A pretest verified the applicability of the questionnaire (N = 20).

Figure 4. One of the data collectors conducting an interview with a primary caregiver.

Handwashing at key times: At T1 and T2, handwashing behavior was measured by means of self-

reports, using 5-point Likert-scales, ranging from “0 – almost a no time” to “4 – almost at each time”.

For the primary longitudinal analyses, the measure was recoded, ranging from 0% to 100%

handwashing, in order to make it easier to understand the results. The questions were introduced by

the sentence “In general, how often do you wash your hands with soap and water when the following

situations occur?” Surveyed key times were handwashing after defecation, after wiping a child’s

16 It has been suggested that household observations should be conducted for at least 3, or better, 5 hours. However, the limited timeline and the usual schedule of Borena women precluded longer observation times.

19

bottom, after other kinds of contact with feces, before eating, before preparing food, before feeding a

child, before breastfeeding a child, and before handling water. A sample question would be: “Before

you eat, how often do you wash your hands with soap and water?”

At T2, imaginative-covered handwashing recall was also used. In this, interviewees were presented

with short sequences of daily routines representing handwashing key times and asked to explain how

they usually carry out these routines, in as much detail as possible. An example would be:

Imagine you have just finished feeding the goats. Now your child is hungry and you have to feed it.

Please describe exactly what you do from leaving the goats’ house until you feed the child.

Data collectors recorded whether the interviewee mentioned handwashing with water, handwashing

with soap, or no handwashing behavior during the description of their routine.

Proxy measures were used at T1 and T2 as approximations for handwashing behavior, such as

presence of a designated place for handwashing, or presence of soap and water.

Behavioral factors were measured at T1 and T2, according to suggestions in the RANAS approach

(Mosler, 2012), and supplemented with items from previous handwashing studies (Aunger et al., 2010

see Annex 1 for example items, means and standard deviations). For each factor, one or several

questions were included in the questionnaire. Principal factor analysis with promax rotation and

reliability analysis were executed with SPSS 19 to test the factor structure. If possible, questions

measuring the same factor were combined in summary variables to facilitate the analyses (see Annex

1 for question numbers and Cronbach’s alpha). 9-point Likert-scales were used to measure bipolar

variables and 5-point Likert-scales were used to measure unipolar variables. With regard to risk

factors, the questionnaire included questions concerning perceived vulnerability and perceived

severity of a diarrhea contagion, and health knowledge. Attitude factors were covered by questions

measuring instrumental beliefs (i.e., efforts and costs, return, attractiveness, and nurture) and affective

beliefs (i.e., liking, the smell of soap and disgust). Furthermore, the questionnaire contained questions

capturing descriptive, injunctive, and personal norms, and compliance with community mobilizers.

Regarding ability factors, perceived behavioral control, maintenance and recovery self-efficacy,

response efficacy, and hindrance were included. Self-regulation factors were covered by questions

measuring action control, action planning, action knowledge, forgetting, control planning, and

commitment.

Promotion activities: Experience of promotion activities was surveyed by self-reports using a

dichotomous format; experienced versus not experienced. In addition, attitudes towards experienced

promotion activities and self-reported promotional effects were measured using Likert-scales.

2.4 Data analysis procedure

In phase 1, regression analyses were conducted to test which behavioral factors best explained self-

reported handwashing, on the basis of which handwashing promotions were selected. Regression

analysis is used to assess the amount of variance in a dependent variable, here handwashing, which

can be explained by one or more independent variable(s), here behavioral factors. An independent

variable explains the dependent variable if a change in the independent variable is accompanied by a

change in the dependent variable. For example, we could expect that ease of handwashing would

explain actual handwashing behavior. If the former actually explains the latter, than a change in the

perceived ease of handwashing should be accompanied by a change in the amount of handwashing

20

behavior. More specifically, if one person thinks that it is very easy to always wash hands, while

another thinks that it is very difficult to always wash hands, and if ease of handwashing does indeed

explain actual handwashing behavior, then the first person should show more frequent handwashing

behavior than the second person. The coefficients received by regression analysis indicate how much

the magnitude of handwashing changes when the magnitude of any of the relevant behavioral factors

changes, while all other variables are held fixed. Those behavioral factors that substantially explained

handwashing (B 0.10 and ≤ .05) should be targeted by promotion activities.

In phase 4, the effectiveness of the applied handwashing promotions was tested. A behavior increase

from T1 to T2 suggests that a particular intervention was successful in achieving its goal. Between-

subjects multivariate analysis of covariance (MANCOVA) was conducted to test for group differences

between the intervention groups, with regard to change in self-reported behavior. MANCOVA

evaluates whether the population means on a set of dependent variables vary across groups. In the

present study, MANCOVA tested whether groups (approximately Kebeles) experiencing different

promotions, or combinations thereof, differed in their means of change in self-reported feces- or food-

related handwashing. Therefore, the changes from T1 to T2 in self-reported feces- or food-related

handwashing served as the dependent variables in the analysis, while the handwashing promotions

were the independent variable. Change from T1 to T2 was calculated by subtracting the T1 values in

behavior from the values obtained at T2. Age and self-reported handwashing frequency at T1 were

entered as covariates.

In addition, Kruskal-Wallis tests, a nonparametric test equivalent to ANOVA and Jonckheere-Terpstra

tests, checking for linear trends in data, were conducted to test for group differences in terms of the

imaginative-covered handwashing recall at T2.

Further, mediation analyses were conducted to test whether the change in self-reported handwashing

behavior, due to the interventions, was mediated by the hypothesized behavioral factors. That is, we

assessed whether the tippy tap intervention and the public commitment intervention, or their

combination, in contrast to the control group (f-diagram-only), influenced the hypothesized behavioral

factors, and whether these influenced the handwashing behavior. Mediation analyses test for these

indirect influences. Indirect effects are obtained by multiplying the effect that an intervention has on a

single behavioral factor by the effect that this single behavioral factor has on handwashing. The total

indirect effect is the sum of all indirect effects per intervention and represents the association that an

intervention has with handwashing, via all behavioral factors tested. Finally, the total effect

summarizes the total indirect effect of an intervention and the direct effect that the intervention has on

handwashing. It is the overall influence of the intervention on the behavior.

For all analyses, all assumptions for conducting the analyses were tested and met and the a priori

level of significance was set at = .05.

In addition, the frequencies of observed handwashing behavior and some proxy measures were

examined between groups to improve understanding of the behavior change potential of the applied

promotion activities.

21

3 Results

3.1 Socio-demographic variables

The age of the respondents ranged between 15 and 96 years, with a mean age of 34.26 years (SD =

13.79). All respondents were female, and 90% were married. A total of 97% of the sample had never

attended school, and 98% could neither read nor write. Of the sample, 94% had traditional beliefs,

while the remaining 6% consisted of Muslims, Protestants, and Catholics. All the respondents were

Oromo. The main type of livelihood was pastoralism and livestock for 98% of respondents. On

average, study households comprised one child under the age of 5 years.

3.2 Results from the baseline survey

3.2.1 Frequency of handwashing at key times and their factor structure

To assess the prevalence of handwashing, two different measures were accounted for; self-reports

and observations.

For self-reports, the mean level of handwashing at key times was rather high (see Table 4). On

average, respondents stated that they washed their hands slightly more than most of the times after

wiping a child’s bottom (M = 3.34), after other kinds of contact with feces (M = 3.15), before eating (M

= 3.07), before preparing food (M = 3.05), and before feeding a child (M = 3.07). Hands were washed

slightly less than most of the times after defecation (M = 2.95), before breastfeeding (M = 2.76), and

before handling drinking water (M = 2.75).

Table 4

Means (M) and standard deviations (SD) of handwashing for self-reports and observations

Self-report Observations

Key time N M SD N M SD

After defecation 461 2.95 1.16 40 1.05 1.75

After wiping a child’s bottom 444 3.34 1.08 40 2.30 2.00

After other kinds of contacts with feces 461 3.15 0.99 44 2.09 2.02

Before eating 461 3.07 0.99 131 1.38 1.82

Before breastfeeding 352 2.76 1.26 54 0.49 1.15

Before feeding a child 443 3.07 1.09 79 1.21 1.82

Before preparing food 461 3.05 1.03 98 2.27 1.85

Before handling water 458 2.75 1.22 56 1.64 1.95

Note: Introductory question for self-reports: In general, how often do you wash your hands with soap and water when the following situations occur? All measures ranged from 0 – (almost) at no time /0 out of 10 times to 4 – (almost) each time / 10 out of 10 times. The same range applied to observations.

Observation data revealed that self-reported handwashing frequencies were overestimated (see Table

4). Observed behavior ranged between washing hands one out of 10 times (M = 0.49; before

breastfeeding) and washing hands approximately half of the times (M = 2.30; after wiping a child’s

bottom). While the mean levels of the observation data were markedly lower than the self-reported

results, the ranking order of the handwashing frequencies for the different key times was almost

22

identical (see Figure 5). The ranking differed between self-report and observation for only two key

times: washing hands before preparing food and washing hands before handling drinking water were

observed more frequently than would have been expected from the self-report ranking order.

Figure 5. Mean level of handwashing at key times – comparison between self-report and observational data.

All the analyses in the baseline study presented in the following were conducted using the self-

reported data. However, additional analyses using observation data would be valuable.

Confirmatory factor analysis with Mplus showed that self-reported handwashing can be divided into

feces- and food/water-related handwashing. To facilitate further analysis, two mean scores were

computed to represent the two factors (see Table 5). The first mean score incorporates handwashing

before eating, before preparing food, before feeding and before breastfeeding a child, and before

handling drinking water, while the second subsumes handwashing after defecation, after wiping a

child’s bottom, and after other kinds of contact with feces. Subsequent analyses were separately

conducted for feces- and food/water-related handwashing.

Table 5

Means (M) and standard deviations (SD) of feces- and food-related handwashing

Self-report Observations

Handwashing type N M SD N M SD

Feces-related hand washing (mean score) 463 3.15 0.97 77 1.76 1.84

Food-related hand washing (mean score) 463 3.00 0.92 146 1.63 1.54

Note: Range: 0 – (almost) at no time /0 out of 10 times to 4 – (almost) each time / 10 out of 10 times.

In addition to the prevalence of handwashing, the reasons given for handwashing were investigated

using an open questions format; that is, no answer categories were given (see Figure 6). The majority

of respondents stated that they wash their hands due to health reasons (52%), namely to protect their

health (14%) and protect against diarrhea (11%), sickness (18%), or bacteria (9%). Nevertheless, 30%

of the respondents said that they washed their hands to remove dirt or smell.

23

Figure 6. Reasons for washing hands with soap. Open question: “In general, why do you wash your hands with soap, ash, or sand?” N (stated reasons) = 1083.

3.2.2 Behavioral factors explaining handwashing behavior

To successfully change behavior, promotion activities must address the factors that influence

behavior. Therefore, in phase 1, the behavioral factors explaining self-reported handwashing had to be

identified to enable an evidence-based selection of promotion activities. Multiple linear regression

analyses were used to identify relevant behavioral factors for feces- and food-related handwashing

(see Table 6). Four behavioral factors were significantly and substantially positively associated with

feces- and food-related handwashing:

respondents who thought that diarrhea is severe (severity),

who washed hands to set a good example to the children (nurture – teaching),

who thought it is disgusting not to wash hands after defecation or before handling food

(disgust),

whose family members washed hands and expect the respondent to wash hands (norms),

tended to wash hands more often after any contact with feces and before handling food than other

respondents.

In addition, in terms of feces-related handwashing, respondents who felt committed to washing hands

tended to do so more frequently after any contact with feces than other respondents (commitment).

Further, one significant and substantial negative association was elicited for feces- and food-related

handwashing: respondents who thought they were often hindered in washing hands (hindrance) did so

less frequently than other respondents after any contact with feces and before handling food.

Moreover, and contrary to expectations, those who were certain that handwashing is effective in

prevention against diarrhea (response efficacy) washed their hands with soap less often before

handling food than did other respondents. However, in contrast with the regression results, the

correlation between response efficacy and food-related handwashing was positive. This indicates that

the negative association within regression could have emerged due to suppressor effects.

24

Table 6

Linear regression analysis for behavioral factors explaining self-reported handwashing

Handwashing type

Feces-related Food-related

Behavioral factor group

Behavioral factor

Constant -0.97 -0.48

Risk factors Perceived vulnerability -0.06 0.00

Conditional vulnerability: not washing hands 0.02 0.02

Conditional vulnerability: washing hands 0.05 0.01

Perceived severity 0.34*** 0.42***

Health knowledge -0.10 0.02

Attitude factors Instrumental – attractiveness 0.10 0.04

Instrumental – nurture: Health of children 0.09 0.07

Instrumental – nurture: Teaching 0.40*** 0.15*

Instrumental – return -0.09 0.03

Response efficacy -0.07 -0.13*

Affective – liking and pleasantness -0.03 0.00

Affective – cleanliness -0.04 -0.03

Affective – disgust 0.11* 0.15**

Norm factors Norms – descriptive and injunctive 0.34*** 0.50***

Personal norm – obligation 0.02 -0.10

Personal norm – guilt -0.06 -0.06

Compliance with HEW/CHV -0.02 -0.01

Ability factors Perceived behavioral control –

difficulties

-0.08 -0.07

Maintenance and recovery self-efficacy – handwashing in general

-0.03 0.04

Self-efficacy – feces 0.07 0.00

Hindrance -0.20** -0.19**

Self-regulation factors Action control 0.05 0.09

Action planning – handwashing facility -0.08 -0.02

Action knowledge – handwashing -0.02 -0.05

Coping planning – no water 0.04 -0.01

Coping planning – no agent 0.00 0.05

Coping planning – forgetting 0.03 -0.02

Commitment 0.16** 0.00

Adjusted R2 .51 .49

Note: *p ≤ .05, **p ≤.01, ***p ≤.001. Factors in bold should be considered for targeting by promotion activities, due to significance level ≤ .05 and unstandardized regression coefficient 0.10.

25

To summarize, five behavioral factors were revealed as those that should be targeted in handwashing

promotions to effectively change handwashing behavior: severity, the teaching aspect of nurture,

disgust, norms, and (lowering of) hindrance. Commitment is also relevant with regard to feces-related

handwashing.

The behavioral factors together explained 51% of the variance in feces- and 49% of the variance in

food-related handwashing. Therefore, 51% and 49%, respectively, of the differences between

respondents in the frequency of feces- and food-related handwashing were explained due to

differences in behavioral factors. An explained variance of around 50% is average within psychological

studies, while an explained variance of under 30% would be small, and an explained variance of over

60% would be high. The remaining 49% and 51%, respectively, of the variance is explained by factors

not captured within the analysis. These could be environmental facilitators or barriers (Curtis,

Danquah, & Aunger, 2009; Schmidt et al., 2009).

3.2.3 Prevalence of behavioral factors

The RANAS model (Mosler, 2012) states that a behavior is sustainably adopted when certain attitudes

and beliefs, the behavioral factors, are present at an optimal level in an individual. The nearer the

actual level is to this optimal level, the more likely it is that a behavior is executed (given that the

behavioral factor actually influences the behavior). Promotion activities should focus especially on

those behavioral factors for which the mean level is suboptimal, since they have the highest potential

to change. Annex 1 presents the behavioral factors included in the present study and their mean level

in the sample. Mean levels that were suboptimal are in bold lettering.

With regard to risk factors, the level of perceived vulnerability, conditional vulnerability, and health

knowledge was not optimal: the respondents were inadequately aware of the risk of contracting

diarrhea, and their knowledge of its causes and effects and prevention methods was insufficient.

The mean levels of the attitude factors were close to optimal. In particular, the levels of the

instrumental beliefs effort, expensiveness, distance of the handwashing facility, and time consumption

were almost perfect: respondents thought that it was not at all an effort, not at all expensive and not at

all time-consuming to wash hands, and that the handwashing facility was not at all distant from their

usual area of activity. In addition, the variance of these behavioral factors was near zero.

The mean levels of the following norm factors were only mediocre: descriptive and injunctive norms,

personal obligation, and compliance with the HEW/CHV, meaning that the belief that personally

important people approve handwashing and wash their hands themselves, was not markedly present,

nor was the feeling of being personally obliged to wash hands, or the tendency to wash hands to

comply with the HEW or CHV.

The ability factors were close to optimal in the sample. Only maintenance and recovery self-efficacy

had some progression potential, meaning, that respondents were not entirely sure whether they would

be capable of maintaining handwashing behavior, even when problems arise, or that they would

restart washing hands after being hindered.

Of the self-regulation factors, the level of action knowledge was only mediocre: people were not

completely aware of how to wash hands correctly. The level of coping planning was insufficient: most

of the respondents had no specific and detailed plan of how to cope with the problems of having no

water, having no handwashing agent, and of forgetting. Further, action planning had a very low

prevalence in the communities. This factor was measured very narrowly: respondents were asked if

26

they had a specific place and facility for handwashing. Only 8% of the respondents answered yes to

this question, while 92% had no particular handwashing facility and location.

In summary, the prevalence of the following factor blocks had progression potential: risk factors, norm

factors, and self-regulation factors.

3.2.4 Qualitative results: some observations

During data collection, some qualitative results were gathered in addition to the presented quantitative

results, primarily by researchers’ observations.

First, it was observed that respondents washed hands without utilizing any sort of handwashing

station, and did not have a designated place and facility for handwashing. Rather, the usual

handwashing technique in the communities was to pour water out of the 20 liter jerry can into a mug,

from which the water was then poured over the hands. This technique seems to be rather

inconvenient, especially when there is no one present to help in pouring the water; therefore, washing

both hands properly is rather challenging. It was accordingly expected that respondents would rate

handwashing as an effort. However, the opposite was true: handwashing was rated by the majority as

being no effort at all. There are three possible explanations for this: 1) respondents are actually

satisfied with their handwashing technique, and indeed do not perceive handwashing as an effort or as

inconvenient; 2) respondents are not actually satisfied with their handwashing technique, but are used

to it, and are not aware of more convenient alternatives, and thus do not perceive it to be an effort; 3)

effort was understood by the respondents in a physical sense (exertion of force), and does not

adequately capture the aspect of inconvenience; other words, such as “inconvenient” or

“cumbersome”, might have been more appropriate. While explanations 1 and 3 cannot be completely

ruled out, it seems most likely that the second explanation is correct: respondents are not aware of

how much a handwashing facility would ease their handwashing practice. In support of this

interpretation, there is an evidence base that shows that a designated place and facility for

handwashing fosters this behavior: “If a busy mother needs to look for soap before preparing a meal,

or a child does not have easy access to water and soap after using the latrine, the probability of

handwashing with soap actually taking place is lower (…) hand-washing with soap is positively

correlated with having a designated place for family members to wash their hands” (Devine, 2010).

Further studies showed that a handwashing device strengthens women’s intentions and plans to wash

hands (Devine & Koita, 2010), can act as a reminder (Nguyen, Paynter, & Thi Nguyen, 2011), and

facilitates behavior change (Devine & Peschiera, 2010).

Second, observational results showed that handwashing behavior was by far not executed at all the

times when it should have been, and when it was stated to be executed. While socially desirable

answering was certainly responsible for this gap between reported and observed handwashing

behavior, it is likely that forgetting also contributed. Although respondents claimed not to forget

handwashing at all, it is likely that they did not realize that they forgot to wash their hands. Regarding

other public health behaviors, such as water treatment, forgetting is traceable, and becomes evident

rather quickly and easily: if someone forgets to filter the water there is eventually an absence of

filtered water, so it becomes evident that the execution of the behavior was forgotten. However,

forgetting to wash hands at a certain moment may never become salient. Therefore, it might be the

case that forgetting is severely underestimated in the communities, and is indeed a major obstacle to

regular handwashing behavior.

27

3.3 Selection of handwashing interventions and their implementation process

Promotion activities to effectively change behavior must influence the critical behavioral factors and

optimize their prevalence in the beneficiaries. According to the above-reported quantitative and

qualitative results, several behavioral factors would have had the potential for targeting by promotion

activities (see Table 7 for a summary).

Therefore, a number of promotion activities would have been conceivable for implementation, or

forgoing promotion activities, implemented during the drought response, could have been optimized by

targeting exactly these behavioral factors. Annex 2 provides an overview of these potential

handwashing promotions, and gives some ideas as to how the forgoing promotion activities could

have been adapted. However, in order to test the promotion activities in as controlled a manner as

possible in an approximated experimental design, only a limited number of promotions could be

selected and implemented in the field; that is only a limited number of behavioral factors could be

decisively targeted.

Table 7

Overview on behavioral factors relevant for handwashing promotions

Factor blocks Behavioral factors Type of evidence

Risks Severity

Vulnerability

Health knowledge

Regression results: ≤ .05 and B 0.10

Prevalence in community is suboptimal

Prevalence in community is suboptimal

Attitudes

Nurture – teaching Disgust

Regression results: ≤ .05 and B 0.10

Regression results: ≤ .05 and B 0.10

Norms Descriptive and injunctive norms Regression results: ≤ .05 and B 0.10/prevalence in community

Personal obligation Compliance with HEW/CHV

Prevalence in community is suboptimal

Prevalence in community is suboptimal

Abilities Action knowledge Hindrance/Ease of performance

Prevalence in community is suboptimal

Regression results: ≤ .05 and B 0.10/qualitative results

Self-regulation Action planning Coping planning

Forgetting/Remembering

Commitment (feces-related handwashing)

Prevalence in community is suboptimal

Prevalence in community is suboptimal

Qualitative results

Regression results: ≤ .05 and B 0.10/qualitative results

Note. Behavioral factors in bold were decisively planned for targeting, that is, promotions were selected on the basis of these. Behavioral factors in italics were also expected to be affected by the selected promotion activities.

Combining the quantitative and qualitative evidence, two groups of behavioral factors were chosen to

be targeted: descriptive and injunctive norms, and hindrance and ease of performance. While several

promotion activities would have been conceivable for tackling these factors, the construction of

handwashing stations and a public commitment intervention were selected.

28

3.3.1 Public commitment intervention

A public commitment is an oral or written pledge or promise to change behavior, and includes the

announcement of the pledge to the community. It can be done by mass media (e.g., local newspaper),

via a poster that is put up, or in a community meeting, where the individual orally commits in front of

the community members.

For the present research project, we decided to organize community meetings in which the

participants would orally commit (see Supplemental material 2 for the written instruction). In return, the

participants would receive a commitment sign, a scarf, which they would be asked to wear to signify

their commitment (see Figure 7). In addition, commitment certificates would be handed out, which the

participants would be asked to pin up in their homes.

Figure 7. A primary caregiver wearing the public commitment scarf.

Public commitment was selected as an intervention, as it was assumed to tackle the descriptive and

injunctive norms (1) by beneficiaries witnessing community members publicly committing during the

community meeting and, in the longer term, (2) by beneficiaries seeing community members wearing

the scarf as a sign of their commitment, and by seeing their certificates (see Figure 8). While norms

could have been targeted by alternative interventions (see Annex 2), two positive side effects of the

public commitment were anticipated, which made this intervention the first choice: it was expected that

the intervention would foster (1) remembering by beneficiaries, on seeing their own and others’ scarfs

and certificates, and (2) commitment, by beneficiaries actually committing publicly.

To our knowledge, public commitment has never been used in handwashing or hygiene interventions

to date. However, it has been successfully applied to promote the use of solar disinfection (Huber,

2008; Kraemer & Mosler, 2012), HIV-prevention (Perlini & Ward, 2000), energy conservation (Richard

D. Katzev & Johnson, 1983; Pallak & Cummings, 1976; Shippee & Gregory, 1982), and recycling

(Richard D. Katzev & Pardini, 1987-1988).

29

Figure 8. Expected influence of public commitment intervention on behavioral factors and thus behavior.

3.3.2 Promotion of constructing tippy taps

As a second intervention, it was decided to motivate study households to construct a handwashing

station for their family, and to train and support them in the construction (see Supplemental material 3

for the written instruction). The type of handwashing station selected for promotion was the so-called

tippy tap. The tippy tap is constructed out of branches and a jerry can; an example is displayed in

Figure 9. It is operated by a foot lever, so that (1) both hands are free for handwashing and (2) the

chance for bacteria transmission is reduced, as the user touches only the soap (tippytap.org, n.d.).

The tippy tap uses only 40 milliliters of water for handwashing, versus 500 milliliters using a mug,

which is the traditional handwashing technique used in the Borena zone. In addition, the used “waste”

water can go to plants or back into the water table. The tippy tap is a low-cost technology and can be

made with local materials; therefore replacement parts can easily be organized.

Figure 9. A tippy tap constructed by one of the research households.

The acceptance and applicability of the tippy tap in the Borena zone was pre-tested in one of the study

Kebeles, Madacho, whereby GPDI constructed a tippy tap in the presence of the community, and its

members were invited to try it out. The community’s reaction was very positive: they liked the tippy

30

tap, thought that it was appropriate and locally applicable, and that it made handwashing easier (Gayo

Pastoral Development Initiative, 2013).

Households were motivated for construction during visits, in which they were invited, and convinced,

to construct a tippy tap. They were informed of the materials17 they would need to obtain for the

construction, and that GPDI would contribute a jerry can. If the households agreed to the construction,

they were supported in selecting the most convenient place for the handwashing station; that is, a

place that is near the usual area of activities and that is visible, so as to remind people of

handwashing. Constructions took place after a community meeting, during which the construction was

explained by the building of a tippy tap in the group, and where the jerry cans were distributed.

It was decided to motivate the households to construct the tippy tap by themselves to the end that the

intrinsic motivation and commitment to use and maintain it would be higher than if a handwashing

device had been provided and erected.

The construction of tippy taps was selected as an intervention, as it was assumed that it would

facilitate the behavior performance and with that to lower hindrance (see Figure 10). Further,

possession of a designated place and facility for handwashing fosters action planning. Seeing their

own and others’ tippy taps should help in remembering to handwash. Seeing community members

using their tippy tap was assumed to tackle social norms.

Figure 10. Expected influence of the tippy tap construction and maintenance-planning on behavioral factors, and thus behavior.

The selection is supported by evidence from previous studies showing that a designated place for

handwashing is correlated with higher levels of handwashing and lower disease occurrence (Biran,

Tabyshalieva, & Salmorbekova, 2005; Devine, 2010; Luby & Halder, 2008). Further research showed

that a handwashing device facilitates behavior performance by making water and soap easily

accessible, and that it helps in intention-formation and planning, as well as serves as a reminder

(Devine, 2010; Devine & Koita, 2010; Devine & Peschiera, 2010; Nguyen et al., 2011).

17 The materials were: four branches of wood, rope, a nail, digging tools, a knife, a candle, and gravel.

31

Maintenance-planning

A tippy tap is only of use when there is water and soap present, so an additional intervention was

chosen to target a continuous provision of water and soap at the tippy tap; the maintenance-planning

intervention. It was planned to visit households that had constructed a tippy tap, and motivate the

primary caregiver to establish a routine regarding when to fill water and when to check if soap has to

be replaced. The primary caregiver was invited to develop specific and realistic plans and fill out a

planning file, which should help them to remember their plans. The file is displayed in Annex 3.

Households were asked to pin the file up in their home. The maintenance plan was expected to tackle

action planning, more precisely, to ensure that soap and water was always available and accessible.

3.3.3 Intervention implementation plan and base intervention

The selected interventions were implemented in the four study Kebeles, as follows:

The tippy tap construction and the public commitment intervention were applied in a fully crossed

design (see Table 8). Therefore, in a first Kebele, Kancharo, neither of the two interventions was

implemented; in a second, Magole, the public commitment was implemented only; in a third, Gololcha,

the tippy tap construction was implemented only; and in the last, Madacho, both interventions were

applied.

Table 8

Overview of the implementation of the different interventions within the four Kebeles

Tippy tap construction

Not

implemented

Implemented

Maintenance-planning

Implemented Not implemented

Public

commitment

Not implemented Kancharo Gololcha, zone 2 & 3 Gololcha, zone 1

Implemented Magole Madacho, zone 1 & 2 Madacho, zone 3

For ethical reasons, it was not feasible to leave one Kebele without any intervention at all. Therefore, a

base intervention had to be selected for application across all four Kebeles. For that purpose, an

intervention that is rather regularly used by relief and development agencies, and which focuses

mainly on knowledge provision, was selected; the f-diagram, which depicts the diarrheal disease

transmission. This promotion activity had already been applied in the drought response in 2011, and

was found to have no association with handwashing behavior. However, in the drought response, the

f-diagram was used in a teacher-centered form by which the health workers explained the route of

contamination by means of the f-diagram. In contrast, it has been suggested that the f-diagram be

used in an interactive learning setting (Krukkert, da Silva Wells, Shrestha, & Duwal, 2010; WASH

Cluster, 2011). This form was also selected for the present interventions, that is, f-diagram group

exercises were conducted (see Supplemental material 4 for the written instruction). The f-diagram

exercise was implemented in the form of a health education session during a community meeting. At

the meeting, participants were divided into small groups. Each group received a set of picture cards in

32

a random order and was asked to put the pictures into an ordered sequences representing the route of

contamination for diarrheal disease (see Figure 11).

To our knowledge, the f-diagram exercise had never been formally tested with regard to its

effectiveness.

Figure 11. F-diagram cards in ordered sequence.

Figure 12. Expected influence of the f-diagram exercise on behavioral factors and thus behavior.

Application of the above-presented implementation design made it possible to test whether the public

commitment intervention and the tippy tap construction, respectively, are more effective in changing

handwashing behavior than the f-diagram exercise alone, whether the public commitment intervention

or the tippy tap construction is more effective, and whether the combination of the public commitment

intervention and the tippy tap construction is more effective than the two interventions alone.

Further, half of the households in which a tippy tap was constructed received a maintenance-planning

intervention, while the other half did not (see Table 8). Thus, it was possible to determine whether

tippy tap construction was more effective with or without maintenance-planning.

33

3.3.4 Implementation process, implementation personnel, and their training

The implementation process began with the signing of the project agreement by GPDI and the

government offices. A launching workshop was held by GPDI in Yabello, to which officials from the

zonal and district health offices and the Kebeles were invited and, furthermore, in which additional

staff from GPDI, OA, and Eawag participated. Its aim was to inform the key stakeholders about the

project in detail, to get their approval with regard to the selected interventions and the implementation

plan, and to gain their active support for implementation of the interventions. The latter included the

warranty that no concurrent promotions by the government or other NGOs would be conducted during

the project phase. In addition, in each Kebele, the leaders and elders were visited to ensure their

approval and support. Their support was especially crucial, due to their influence within the

communities and their being role models therein.

The promotion activities were implemented by 10 Behavior Change Agents (BCA, two or three per

Kebele) employed by GPDI. These were supported by HEWs (one or two per Kebele) and CHVs

(between four and six per Kebele). In both Woredas, a Districts Level Health Supervisor (DLHS,

employed by GPDI) supported and monitored the BCAs, HEWs, and CHVs. The overall responsibility

for the promotion implementation lay with GPDI’s WASH coordinator (WC). He was supported by the

FRC, a local Eawag employee, whose responsibility was the external monitoring of the implementation

process.

The BCAs were trained in a 2-day workshop outlined by Eawag, which also provided detailed written

instructions for each promotion activity (see Supplemental material 2-4), and detailed implementation

and monitoring plans. The workshop was held by the WC, two OA agents, and the FRC, and its goals

were to ensure that the BCA knew why and how the interventions were conducted, were aware of their

role and responsibility, and were aware of the implementation plan. Therefore, the baseline results

were first briefly presented and it was explained how the selected interventions were deduced from

these results. Second, the promotion activities were explained in detail. Third, the promotion activities

were practiced by means of role plays, including the construction of the tippy tap. Fourth, the BCA’

role and responsibility was explained, as well as the reporting chain and monitoring files. Finally, the

implementation plan per Kebele was discussed.

HEWs and CHVs were trained in additional workshops at Kebele level, which were held by the WC,

the FRC, and the responsible DLHS and BCA in each Kebele. In this training, the interventions were

explained in detail, and learning occurred by means of role plays. In the case of the tippy tap

intervention, these were actually constructed. After training, HEWs and CHVs were highly motivated,

and promised to contribute to the activities (Gayo Pastoral Development Initiative, 2013).

3.4 The implementation process: some qualitative results

3.4.1 Some incidents and inconsistencies in the implementation process

The implementation process was marked by some shortcomings. First, the planned monitoring

process was restricted, with the necessary transportation means not being appropriately provided.

This led to a lack of monitoring visits by the DLHSs during the entire implementation phase, and

especially by the WC and FRC. Consequently, the quality assurance of implementation was not

continuously guaranteed, and potentially necessary corrective actions could not be taken at all times.

Further, in one Kebele, Magole, the filled monitoring files were inconsistent with the information

provided by the respondents: while it was reported in the monitoring files that all primary caregivers

34

interviewed during the baseline data collection had been reached, 11% of the interviewed individuals

stated that they had not experienced the intervention. In contrast, in the other three Kebeles, where

the monitoring files indicated that a small amount of households could not be accessed, only 1-4% of

the respondents stated that they had not experienced the intervention.

Second, not all the promotion material was provided as instructed and agreed upon. Color prints of the

f-diagram cards could not be produced in due time. Therefore, the f-diagram-exercises had to be

conducted using qualitatively inferior monochrome prints. Further, the maintenance-planning file had

not been translated from English into Afan Oromo. As most of the beneficiaries were illiterate, the file’s

content was delivered by pictures and text in unison. Therefore, it can be assumed that beneficiaries

understood the file’s content, regardless. However, the same file in a local or a foreign language may

have a different impact, due to socio-cultural factors, such as familiarity, acceptance, or trust.

Moreover, the required number of promotion materials was not always available: several study

households stated that they did not receive the promised materials, such as the public commitment

scarf or certificate, or the jerry can to construct the tippy tap. In contrast, in other households that were

hindered in participating in the promotion activity, the promotion material was sent through a neighbor

or a relative. In doing this, the purpose of the promotion activity and of the sent material was not

adequately delivered: the concerned respondents were not aware of the content and sense of the

public commitment and its sign.

3.4.2 A specific shortcoming in the implementation process: the public commitment sign

Part of the public commitment intervention was the presentation of a public commitment sign to the

committing beneficiaries. Several deficiencies emerged with respect to this sign.

The first of these concerns the selection of the sign, a process that, although extended over several

months, resulted in a suboptimal choice. The following criteria guided the selection: being unique and

prominent, but nevertheless appropriate for the local cultural context, belonging and relating to the

primary caregiver instead of the entire household, and with a cost maximum of US$ 2. From the long

selection process, a scarf to be tied around the head appeared to be the most appropriate

commitment sign. However, after implementation it appeared that this scarf was inadequate as a

public commitment sign: only five respondents were actually wearing it when being invited for

observation and interview. This might be explained by the fact that, in the Borena zone, tying a scarf

around the head is rather common for men, but not particularly so for women. As women were not

used to wearing scarfs, they would have had to change their dressing behavior to do so. Therefore, a

behavior change would have been necessary merely to bring the public commitment sign into force

and with that, to elicit the intended behavior change, namely handwashing. Thus, a public commitment

sign should be something that it is either already customary to wear (so that no behavior change is

necessary), is permanently wearable (such as some jewelry), or is stationary instead of worn (such as

a door curtain). Although the selection took a long time and involved many discussions, the selected

sign should have been pre-tested carefully in the field in terms of its appropriateness.

There are indications that the idea behind the public commitment sign went astray during the BCA

training. The researchers’ original idea behind the sign was that people expressed their commitment to

the community by wearing the scarf; that is, that the commitment process proceeds to continuously

trigger social norms and commitment. However, what was taught during training, and consequently

conveyed to the communities, was that the scarf serves as a reminder to always wash hands with

35

soap at key times.18 This is reflected in the result that 86% of the respondents stated that they wear

the scarf as a reminder of handwashing, hygiene, or cleanliness, whereas only 2.5% stated that they

wear the scarf as a sign of commitment. Although provision of a reminder is a valid alternative

promotion strategy, the original strategy got lost.

In one Kebele, Magole, there is evidence to suggest that the basic idea of wearing the scarf to foster

handwashing behavior was completely overridden in some cases: respondents said that they were

told to wear the scarf to be given a lift, or simply to always wear the scarf when people from outside

visited the village, without any other explanation for wearing the scarf being given. For both of these

incidents of distortion, the link between scarf and handwashing was removed, rendering the scarf

impotent for eliciting handwashing behavior.

3.4.3 Interference in the data collection

During data collection, the enumerators were repeatedly informed by interviewees and the community

that promotion implementers had come to their villages and had informed residents of the forthcoming

data collection. Targeted individuals were asked to act in a promotion-favoring manner during data

collection. Therefore, they were asked to wear the public commitment scarf, to use the tippy tap

frequently, and to give favorable answers. How far data were affected by these interferences cannot

be assessed, but it could have happened in two ways: (1) respondents followed the request and

answered more favorably; or (2) respondents were indignant at the influencing attempts and reacted

with reactance by answering more unfavorably.

The research team tried to mediate the effect of these attempts by (1) asking the data collectors to be

attentive to such statements, especially as they were often given outside the immediate interview

situation, by (2) trying to obtain additional information beyond the scope of the questionnaire whenever

indicated, and by (3) adapting some of the data collection methods, such as recording whether the

target individual was wearing the public commitment scarf, not only during the interview, but also

during the interview invitation when the visit was unexpected.

Further, contrary to agreement, GPDI organized some bush-clearing and pond-rehabilitation activities

during data collection. Consequently, some target individuals were absent during planned data

collection and had to be visited at another time, leading to additional transportation and loss of time.

3.5 Reach of the promotion activities

Table 9 displays the reach of the promotion activities. Around a fifth of the sample experienced the f-

diagram-only intervention and the public commitment-only intervention. Slightly less (17%)

experienced the tippy tap intervention in combination with the public commitment. A total of 30%

experienced the tippy tap-only intervention. Although the implementation design was intended to reach

the complete study sample, 10% stated that they had not experienced any of the promotion activities.

Since self-selection processes might be responsible for the nonparticipation, these respondents were

excluded from further analyses.

18 That the scarf could serve as a reminder was assumed to be a positive side effect. However, the main idea of the scarf was to serve as a sign of commitment.

36

Table 9

Reach of the promotion activities

Intervention groups

NI FDO PCO TTO total PC+TT total

N 58 114 110 166 95

Percentage 10.7 21.0 20.3 30.6 17.5

TTO TTMPO PC+TT PC +TTMP

N 74 92 52 43

Percentage 13.6 16.9 9.6 7.9

Note. NI = no intervention; FDO = f-diagram-only; PCO = public commitment-only; TTO = tippy tap-only; TTMPO = tippy tap with maintenance-planning -only; PC+TT = public commitment and tippy tap; PC+TTMP = public commitment and tippy tap with maintenance-planning; TTO total = TTO and TTMPO; PC+TT total = PC+TT and PC+TTMP.

3.6 Influence of the handwashing interventions on selected outcome variables

3.6.1 Influence on observed behavior

In terms of observed behavior, the longitudinal sample was too small for the purpose of drawing

statistically rigorous conclusions.

As a consequence, only tendencies in behavior change can be reported by looking at the rate in

observed handwashing behavior before (T1) and after (T2; see Figure 13 and 14 and Annex 4) the

handwashing promotions. In Figures 13 and 14, the rate of feces- and food-related handwashing,

respectively, are displayed. That is, the percentage of observed situations in which study participants

actually washed their hands out of all the observed situations in which they were supposed to wash

their hands is displayed.

Observed handwashing after contact with feces was substantially higher at T2, compared to T1, for

the groups who experienced the tippy tap-only intervention or the combined tippy tap/public

commitment intervention (see Figure 13). In contrast, for the group who experienced only the public

commitment intervention, the handwashing rate was lower at T2, compared to T1. For the group who

experienced the f-diagram-only, the handwashing rate was higher at T2, compared to T1 (where

handwashing was non-existent). However, their handwashing rate at T2 was quite low, compared to

the rates of the groups having experienced the tippy tap-only intervention or the combined

intervention.

Figure 14 shows that observed food-related handwashing was rather low for all intervention groups at

T2. In the groups experiencing the tippy tap-only or the combined intervention, behavior did not

change from T1 to T2. In contrast, there was a substantial behavior decrease in the group who

experienced only the f-diagram, and a considerable decline in the group experiencing only the public

commitment intervention (see chapter 4.2 for a discussion).

37

Figure 13. Mean rates of observed feces-related handwashing at T1 and T2. * Handwashing rate was zero at T1.

Figure 14. Mean rates of observed food-related handwashing at T1 and T2.

In summary, handwashing rates at T2 compared to T1 tended to be most satisfactory in the groups

who experienced the tippy tap-only intervention or the combined PC+TT intervention, while they were

least satisfactory in the f-diagram-only group, followed by the public commitment-only group.

38

3.6.2 Influence on self-reported behavior

3.6.2.1. Self-reported handwashing behavior

To examine changes in self-reported handwashing behavior, self-reported handwashing at T1 was

subtracted from self-reported behavior at T2. This measure of change was examined using a

MANCOVA, while controlling for the age19 of the respondents and self-reported behavior at T120 (see

Annex 5 for more detailed statistical parameters).

In terms of feces-related handwashing, groups who experienced the combined intervention with

maintenance-planning changed most positively from T1 to T2, followed by those experiencing the

tippy tap-only intervention with maintenance-planning (see Figure 15). In contrast, handwashing

behavior decreased in those groups experiencing the f-diagram-only or the public commitment-only.

Change was marginal in the group experiencing the combined intervention without maintenance-

planning and the tippy tap-only intervention. Planned contrasts revealed that means differed

significantly between the f-diagram-only group and the two groups experiencing significant positive

change, and between the public commitment-only intervention and the group experiencing the

combined intervention with maintenance-planning. The effect size of the mean difference (MD) in

change between the most and least successful intervention (PubComm+TippyTap+MP versus

FDiagramOnly) was high: MD = 27%, d = .80.

Figure 15. Mean changes in self-reported feces-related handwashing from T1 to T2. PubComm = public commitment; TippyTap = tippy tap promotion; MP = maintenance-planning; FDiagram = f-diagram.

Figure 16 shows that, with regard to food-related handwashing, the group experiencing the combined

intervention, with or without maintenance-planning, changed most positively, while behavior

decreased again in the groups experiencing only the f-diagram or only the public commitment.

Marginal changes were found in the groups experiencing the tippy tap-only intervention, with and

without maintenance-planning. According to planned contrasts, means differed significantly between

the f-diagram-only and both tippy tap-only groups and both combined groups. The public commitment-

19 Age was the only socio-demographic variable that emerged as explaining handwashing behavior, insofar as older respondents showed greater behavioral change. Thus, age was the only socio-demographic variable included as a covariate in the MANCOVA. 20 At T1, the four Kebeles differed in terms of their mean self-reported handwashing behavior, with lower levels in Madacho and Gololcha than in Magole and Kancharo. Consequently, behavior at T1 was included as a covariate in the MANCOVA.

39

only group differed significantly from the combined group with maintenance-planning. The effect size

of the mean difference (MD) in change between the most and least successful intervention

(PubComm+TippyTap+MP versus FDiagramOnly) was also high: MD = 35%, d = 1.03.

Figure 16. Mean changes in self-reported food-related handwashing from T1 to T2. PubComm = public commitment; TippyTap = tippy tap promotion; MP = maintenance-planning; FDiagram = f-diagram.

To summarize, the results for change in self-reported handwashing supported the tendency found in

the observed data: in groups experiencing the f-diagram-only intervention or the public commitment-

only intervention, there was a behavior decrease, while change in groups experiencing the tippy tap-

only intervention or the combined intervention was positive, or only marginal.

3.6.2.2. Imaginative-covered recall

Handwashing was measured by means of the imaginative-covered recall at T2 only. Thus,

comparisons were only possible between intervention groups at this time. These were realized by

means of a Kruskal-Wallis test (see Annex 6 for more detailed statistical parameters). Figures 17 and

18 show the median rate of recalled feces- and food-related handwashing; the percentage of

descriptions of daily routines, in which study participants stated that they washed their hands with

soap, out of all the sequences of daily routines that were presented to the participants, is shown.

Feces-related handwashing was highest for groups experiencing the tippy tap-only intervention, or the

combined intervention, and lowest for the group experiencing the f-diagram-only (see Figure 17). A

significant trend was revealed, in as much as rates significantly increased from the f-diagram-only to

the combined intervention.

An almost identical picture was found for food-related handwashing (see Figure 18). Median

handwashing was lowest for the f-diagram-only group and the public commitment-only group, and

highest for the tippy tap-only and the combined intervention groups. This linear trend was again

significant.

40

Figure 17. Median rates of imaginative-covered recall of feces-related handwashing at T2.

Figure 18. Median rates of imaginative-covered recall of food-related handwashing at T2.

In summary, and similar to the results presented above in terms of the self-reported data, according to

the imaginative-covered recall, handwashing rates were lower for the groups having experienced the f-

diagram-only or the public commitment-only than for the groups having experienced the tippy tap-only

intervention or the combined intervention.

3.6.3 Proxy measures: tippy taps in function

A crucial facilitator of handwashing is a designated place and facility for handwashing, and the tippy

tap intervention directly tackled this. However, merely providing a tippy tap to a household is

insufficient to enable handwashing behavior. First, household members must recognize and accept a

handwashing station as their designated place to wash their hands. This is better achieved when

households are motivated to construct their handwashing station by themselves, than when a

handwashing station is simply provided. Therefore, in the present interventions, households were

motivated and supported in constructing their own tippy tap. Second, they have to maintain the

handwashing station and make sure that soap and water is always present at the tippy tap, which was

directly addressed through the maintenance-planning intervention.

In the two Kebeles where the tippy tap intervention was applied, 94% (Madacho) and 99% (Gololcha)

of the respondents actually constructed a tippy tap. That is, almost 100% heeded the call and invested

material and time in the construction.

While only a small minority of households stated that they had a designated place and facility for

handwashing at T1, at T2, in those groups having constructed a tippy tap (with or without combination

with public commitment), almost 100% stated that they had a designated place and facility for

41

handwashing (see Figure 19). Therefore, nearly 100% not only constructed a tippy tap, but were still

using it, and actually recognized the tippy tap as their designated place for handwashing. This is quite

substantial, especially when considering that the interventions had ended 2-4 months prior to data

collection.

Figure 19. Percentages of households having a designated place for handwashing at T1 and T2. FDiagram = f-diagram; PubComm = public commitment; TippyTap = tippy tap promotion; MP = maintenance-planning.

At T2, those respondents who stated that they had a designated place and facility for handwashing

were asked to show the place (in most cases the constructed tippy tap), and those stating that they did

not have a designated place and facility for handwashing, were asked to show the place where family

members most often washed their hands (in most cases the place where the jerry can with water was

stored). In both cases, the data collectors checked whether there was water and soap available at the

place they were shown.

In those groups having constructed only a tippy tap, water was present in more than 90% of cases

(see Figure 20). In the group having experienced the combined intervention with maintenance-

planning, water was present in almost 80% of cases. For the groups having experienced the

combination without maintenance-planning, or only the public commitment, or only the f-diagram,

water was present in fewer than 70% of cases.

Figure 20. Percentages of households having water at their usual place for handwashing at T1 and T2. FDiagram = F-diagram; PubComm = public commitment; TippyTap = tippy tap promotion; MP = maintenance-planning.

42

With regard to soap, again, the groups that had constructed a tippy tap were best positioned, with

soap being present in around 80% of cases (see Figure 21). In groups having experienced the

combination, soap was present in slightly more than half of cases. In less than half of cases, soap was

present in groups having experienced the public commitment-only, or the f-diagram-only.

Figure 21. Percentages of households having soap at their usual place for handwashing at T1 and T2. FDiagram = f-diagram; PubComm = public commitment; TippyTap = tippy tap promotion; MP = maintenance-planning.

All in all, groups that were invited to construct a tippy tap not only accepted the invitation in high

numbers, but they also recognized the tippy tap as their designated place for handwashing. What is

more, in 50-80% of the households, the tippy tap was functioning, with water and soap being present.

In contrast, in groups having experienced the f-diagram-only or the public commitment-only, water and

soap was present at the place where they most often washed their hands in only 40-60% of the

households.

3.6.4 Influence of maintenance-planning on self-reported behavior and presence of soap and water

The above results revealed that in terms of self-reported handwashing and the proxy measures,

groups with maintenance-planning were best-positioned. Consequently, we tested whether the tippy

tap intervention with maintenance-planning was significantly more effective than without maintenance-

planning. However, no significant differences were found between the tippy tap-only intervention with

and without maintenance-planning, and between the combined intervention with and without

maintenance-planning, in terms of self-reported handwashing behavior or the proxy measures of

having water and soap present (see Annex 7 for detailed statistical parameters).

43

3.7 Mechanism of change: Psychological variables explaining the influence of the handwashing interventions on change in self-reported handwashing

With regard to the mechanism of change, it was hypothesized that the public commitment intervention

would change handwashing behavior by influencing norms, commitment, and remembering. In terms

of the tippy tap intervention, it was assumed that the intervention would affect hindrance, action

planning, norms, and remembering, and thereby behavior. Mediation analyses were conducted to test

these hypotheses; that is, whether the public commitment and tippy tap interventions and their

combination, compared to the f-diagram-only group, explain a change in these behavioral factors,

which then explain a change in handwashing behavior. As mediators, the change in behavioral factors

from T1 to T2 was included. In addition to the hypothesized factors, (additional) behavioral factors,

which emerged from regression analyses to significantly explain change in handwashing, were also

included in the mediation analysis to test whether additional factors can explain the intervention

effects.21 These were vulnerability and severity for both outcomes, nurture to keep children healthy

with regard to feces-related handwashing, and nurture as setting a good example, with regard to food-

related handwashing. The associated regression analyses are presented in Annex 8.

In Figures 22 and 23, mediation results for change in feces- and food-related handwashing,

respectively, are presented.22 On examination of the influence of the interventions, it appears that

public commitment-only, compared to the control group, had no significant influence on any of the

behavioral factors in both mediation models. In contrast, tippy tap-only and the combined intervention

(public commitment and tippy tap) led to a positive change in norms and action planning, and to a

reduction in hindrance and forgetting. In the feces-related handwashing model, the latter intervention

additionally influenced change in nurture to keep children healthy. While tippy tap-only and the

combined intervention explained food-related commitment, this was not the case for feces-related

commitment, the latter result contradicting the hypotheses. Regarding the associations between

hypothesized mediators and handwashing, only two behavioral factors significantly explained feces-

and food-related behavior change; norms and forgetting. In parallel with results from the baseline

study, commitment itself explained only feces-, but not food-related handwashing. Hindrance and

action planning, did not at all explain change in handwashing. When looking at mediation effects (blue

arrows), it appears that norms and forgetting significantly mediated the effect of tippy tap-only, and the

combined intervention. In the feces-related model, nurture was also found to mediate the influence of

the combined intervention.

To summarize, compared to the control group, tippy tap-only and the combined intervention had the

hypothesized influences on norms, hindrance, action planning and remembering, and thus behavior.

Commitment was tackled only in terms of food-related handwashing. However, public commitment-

only had no influence on any of the hypothesized behavioral factors, and thus, neither on behavior.

21 Factors that do not explain change in handwashing behavior cannot mediate an intervention effect. 22 Mediation tables are presented in Annex 9.

44

Figure 22. Mediation results comparing intervention groups to the control group, f-diagram-only: Effects of interventions on change in selected behavioral factors and of change in selected behavioral factors on change in feces-related handwashing. Numbers next to the arrows indicate unstandardized regression coefficients. Higher numbers signify stronger associations. Only significant coefficients with p < .05 are displayed. Behavioral factors that were hypothesized to mediate the interventions’ effect are highlighted in lilac. Significant indirect paths are marked in blue.

Figure 23. Mediation results comparing intervention groups to the control group, f-diagram-only: Effects of interventions on change in selected behavioral factors and of change in selected behavioral factors on change in food-related handwashing. Numbers next to the arrows indicate unstandardized regression coefficients. Higher numbers signify stronger associations. Only significant coefficients with p < .05 are displayed. Behavioral factors that were hypothesized to mediate the interventions’ effect are highlighted in lilac. Significant indirect paths are marked in blue.

45

4 Discussion

4.1 Baseline results: behavioral factors to be addressed and derived promotion activities

With regard to research question one (Which behavioral factors best explain handwashing with soap,

and thus should be targeted by handwashing promotions?), regression analyses revealed that the

following behavioral factors best explained handwashing behavior in our study, and should thus be

considered for future promotion activities:

severity

nurture to teach children

disgust

descriptive and injunctive norms

hindrance

commitment.

As a second selection criterion, mean levels of the behavioral factors were examined. Its goal was to

specify those behavioral factors that significantly explained handwashing behavior, while

simultaneously having only a small or mediocre mean level in the communities. This combined

examination suggested that descriptive and injunctive norms have the highest intervention potential

(high regression coefficient and only mediocre mean level), and should thus be targeted by

handwashing promotions.

Further, qualitative results suggested that a rather cumbersome handwashing technique, washing

hands with a mug, was prevalent in the community, and this seemed to inhibit an easy behavior

performance. In addition, forgetting might have been a major obstacle to regular handwashing

behavior.

Research question two read: By means of which promotional activities is it expected that these

behavioral factors will be changed? Several interventions were considered to target norms (see Annex

2). However, the first choice was a public commitment intervention, as it was anticipated that it would

not only change norms, but also, as a positive side effect, help in remembering. Easy behavior

performance was expected to be best met by households using a handwashing facility. Therefore, the

promotion of a tippy tap construction was selected. Again, a positive side effect of remembering was

anticipated.

4.2 Longitudinal study: summary of the interventions’ effect

When looking at the combined evidence drawn from observed and self-reported data and from proxy

measures, it was revealed that the tippy tap-only intervention and its combination with public

commitment have promising intervention potential with regard to handwashing: observed

handwashing behavior tended to be higher at T2 than T1 for these groups (feces-related), or did not

decrease (food-related), changes in self-reported handwashing behavior from T1 to T2 were positive

for these groups, and the median level in the imaginative-covered recall at T2 was highest for these

groups. Further, nearly 100% of the households that were asked to construct a tippy tap actually did

so, and recognized it as their designated place for handwashing. In addition, the vast majority of these

households had a functioning tippy tap, with water and soap being present, at T2; 2-4 months after

implementation completion. In contrast, for the group that experienced only the f-diagram, the

behavior decreased in frequency or was at a low level at T2. Further, these households rarely had a

46

designated place for handwashing, and soap and water were not easily accessible for handwashing in

more than half of the households. Although less pronounced, the same was true of the public

commitment-only intervention.

While the f-diagram exercise was selected as a base intervention in part because no substantial effect

on handwashing behavior was expected, it was not anticipated that frequency of behavior might

decline. Moreover, public commitment was selected specifically to boost social norms and, with that,

behavior. However, the opposite was true, with the public commitment-only intervention having a

marginal, or even detrimental, effect on handwashing behavior. There are several potential

explanations for these results, as follows:

1. The f-diagram exercise and the public commitment intervention had no impact and were not

capable of slowing the natural decrease in behavior. The baseline data collection took place

immediately after the drought response had finished. It is credible that handwashing was at a

rather high level, due to the handwashing promotions that had been conducted as part of the

drought response. However, there are studies that show that sustainability of behavior change

is not always a given (Arnold, Arana, Mausezahl, Hubbard, & Colford, 2009; Luby et al.,

2009). Thus, behavior might have decreased from T1 to T2 because the forgoing behavior

change was unsustainable. Having no impact, the interventions were not capable of slowing

this decline.

2. The f-diagram exercise had a genuinely negative impact. The f-diagram exercise assessed

the route of contamination as a whole. That is, it depicted not only how handwashing can

prevent diarrhea, but also how, amongst other things, sanitation and safe water are crucial.

These are not comprehensively guaranteed in the area. It is credible that participants were

over challenged by the sheer number of changes necessary, so they gave up instead of

focusing on the feasible changes. Whether this was the case could be further investigated in

focus groups.

3. The public commitment-only intervention was unsatisfactorily implemented, and was

accordingly unable to fulfill its potential and to slow a natural behavior decline (which would be

explained analogous to point 1). There are two reasons for concern with regard to the correct

implementation of the public commitment-only intervention. First, if it were the public

commitment intervention in itself that did not work, the combined intervention should have

performed the same, or poorer, than the tippy tap-only intervention. However, the opposite

was true. Second, it was in the Kebele with public commitment-only, Magole, where the

majority of concerns around implementation quality arose.

With regard to the maintenance-planning, groups experiencing this intervention had a tendency toward

more positive outcomes than those not experiencing it. However, these differences were not

significant. For maintenance-planning to be effective, it is crucial that the stated plans are specific and

realistic. It might be that some of the plans did not meet this criterion, which would lower the

effectiveness of this intervention. Further, as stated, the maintenance-planning file was, contrary to

instructions, not translated from English into Afan Oromo. It is possible that beneficiaries would have

more widely accepted a file written in their own language and thus, would have felt more committed to

following their plans. Moreover, to fulfill its intervention potential, the maintenance-planning file should

serve as a reminder to actually implementing the plans. The file, being a monochrome print, might

have been unable to awaken attention and thus, serve as a prompt. Future studies should address

these issues by investigating how specific and realistic are the established plans, by comparing the

impact of files in a foreign and local language, and by using eye-catching maintenance-planning files.

47

In terms of the mechanism of behavior change, it had been hypothesized that public commitment

would change handwashing behavior by influencing norms, commitment, and remembering. In terms

of the tippy tap intervention, it was assumed that it would affect hindrance, action planning, norms, and

remembering, and thereby behavior. Mediation analyses revealed that public commitment-only, when

being compared to f-diagram-only, had no associations with any of the hypothesized behavioral

factors, and thus, neither with behavior. Tippy tap-only and the combined intervention appeared to

have the hypothesized associations with norms, hindrance, action planning and remembering. This is

in line with previous research showing that a handwashing station (1) facilitates behavior performance,

by making water and soap easily accessible, (2) helps in planning, and (3) serves as a reminder

(Devine, 2010; Devine & Koita, 2010; Devine & Peschiera, 2010; Nguyen et al., 2011). Further, food-

related commitment was also associated with the combined intervention and the tippy tap-only,

although the latter association had not been hypothesized. Since hindrance, action planning, and

commitment with regard to food-related handwashing did not explain behavior, it emerged that only

norms and forgetting significantly mediated the association of tippy tap-only and the combined

intervention with behavior. Apart from the hypothesized mediating behavioral factors, only nurture to

keep children healthy was associated with an intervention, that is, the combined intervention, and

mediated the intervention’s association with behavior.

In conclusion, and in response to research question three (What is the intervention potential of the

selected handwashing interventions on handwashing behavior?) it can be stated that:

The tippy tap has a promising intervention potential: the tippy tap-only intervention and the

combined intervention had positive outcomes and explained the hypothesized behavioral

factors. This is in accordance with previous research showing that a designated place for

handwashing is correlated with higher levels of this behavior (Biran et al., 2005; Devine,

2010).

The potential of the public commitment intervention is uncertain: the public commitment-only

had negative outcomes, while the combined intervention had positive outcomes and did

explain, as hypothesized, food-related commitment. While the present study is, to our

knowledge, the first to apply a public commitment to promote handwashing or hygiene, our

inconsistent results are in contrast with research in other fields, where public commitment was

successfully applied (e.g. Huber, 2008; Kraemer & Mosler, 2012; Perlini & Ward, 2000).

However, one very recent study also found detrimental effects of a public commitment

intervention on well-switching in Bangladesh (Inauen & Mosler, 2013). Further research is

required to test the potential of a public commitment intervention in hygiene, which is

implemented in accordance with instructions.

The intervention potential of the f-diagram exercise is questioned: the f-diagram-only had

negative outcomes. To our knowledge, the f-diagram exercise has never been formally tested.

However, it was applied in a teacher-centered form during the drought response without

having any association with handwashing behavior (see chapter 1.3). An in-depth analysis of

the f-diagram exercise, including qualitative and quantitative approaches, would be valuable.

The intervention potential of maintenance-planning is uncertain. Although maintenance-

planning had positive outcomes, they were not statistically significant. Further research is

required to address this.

48

Of course, the question regarding sustainability of change is related to the intervention potential.

Several studies have addressed this question in the field of hygiene behavior change, revealing rather

inconsistent results. In some cases, behavior change persisted, sometimes even after time-spans of

several years (Cairncross & Shordt, 2004; Cairncross, Shordt, Zacharia, & Govindan, 2005; Wilson &

Chandler, 1993), while in others, behavior declined dramatically within 1 or 2 years (Arnold et al.,

2009; Luby et al., 2009). It is promising that the present research project revealed behavior change up

to 4 months after intervention termination. To assess the sustainability of change, a follow-up, 1 or

more years after intervention termination, would be necessary. However, due to the high government

and NGO activity in the Borena zone, it is questionable, whether differences between Kebeles could

be traced back to the interventions. This was possible for the present research, due to the zonal and

districts health offices’ warranty that no concurrent promotions by the government or other NGOs

would be conducted during the project phase.

Of course, for a lasting behavior change, external facilitators or barriers are also crucial. One of these

is adequate availability of water, which is not always guaranteed in the Borena zone. It is to be feared

that handwashing rates will drop in the case of another drought situation. However, the tippy tap might

have a leveling effect, by using only 40 milliliters of water for handwashing versus 500 milliliters if a

mug is used (tippytap.org, n.d.).

4.3 Limitations

The findings in this report are subject to the following limitations.

In terms of observed behavior, the longitudinal sample was rather small, which limited the options of

conducting analyses with these data. As already stated in chapter 2.3.1, the following points led to the

small sample size:

1. Due to budget and time constraints, it was decided to observe not all study households, but

only 200 out of 600.

2. At T1, only 151 observations could be conducted (a) due to difficulties in obtaining the

required number of households in each Kebele, because of low population density and high

absence of households during data collection days, and (b) due to the earlier termination of

the data collection in one Kebele, Kancharo.

3. During the observations, critical events happened with low frequency. That is, at T1 and T2

feces-related events were observed in only around 50% of the households; feces-related

events were observed in only 38 households at both T1 and at T2. Furthermore, the average

event frequency was below one for feces-related events and between three and four for food-

related events, meaning that the observed data drew on a limited number of observed events.

As a consequence, the main analyses had to draw on self-reported data. Several scholars have

previously emphasized the problem of inflated self-reports in terms of socially desirable behavior,

including handwashing behavior (Biran et al., 2008; Curtis et al., 1993; Halder et al., 2010; Manun'Ebo

et al., 1997). Nevertheless, it is worthwhile examining self-reported handwashing behavior. While an

overestimation in self-reports is beyond dispute, the four applied measures (i.e., observations, self-

reports, imaginative-covered self-reports, and proxy measures) all pointed in the same direction. Thus,

it is unlikely that self-reports and observations are completely disconnected. However, a validation of

the primary results by means of observational data would be preferable. Further, an in-depth analysis

49

of the interrelationship between the different applied measures would be valuable, and, in doing so,

factors potentially explaining over-reporting should be tested. While social desirability is certainly one

component, it might be the case that other factors are also involved, such as recall errors or inconstant

or interrupted domestic work.

Each of the interventions or intervention combinations, respectively, was applied in one Kebele. With

that, differences in behavior or behavior change that are genuine to a Kebele might blur the

interventions’ effect. To rule out such a cluster effect, a cluster-randomized controlled trial would have

been ideal. In such a design, each intervention is randomly assigned to several clusters. Such an

approach did not seem feasible in the present project, due to the following:

1. Because of time constraints at T1, and since it was only possible to work in Kebeles where

GPDI had conducted drought response23, it was not feasible to collect data in more than four

Kebeles.

2. Within Kebeles, the Ollas, although being rather scattered, are in constant and close contact.

Therefore, it did not seem practical to apply different interventions in different Ollas within the

same Kebele (that is, to work with the Ollas as clusters), as this might have raised social

tensions.

To level this shortcoming slightly, the analyses looked at the change in behavior, while controlling for

handwashing rates at T1.

Although we attempted to apply the handwashing interventions in a standardized and controlled

manner, it became apparent that monitoring was insufficient, and that the implementation quality was

sometimes questionable. As a consequence, it is not clear whether the public commitment intervention

had a genuinely small impact, or if it was prevented from fulfilling its potential, due to implementation

problems.

Although the public commitment sign was selected by means of a long consulting process, there are

indications that scarfs are suboptimal in the Borena zone as public commitment signs, especially as

the wearing of them would itself have required a behavior change. In future interventions, a public

commitment sign that is fixed or installed somewhere, while being publicly visible, or which is

permanently worn, should be used.

The targets of the study were primary caregivers. As a consequence, the sample contained only

women. Therefore, the research findings primarily apply to women. It might be that different

conclusions would be drawn for male beneficiaries. This could be tested in subsequent analyses.

23 That means it was not possible to select, for example, the 10 most easily accessible Kebeles to save time.

50

5 Conclusions and recommendations

The present research project tested the effectiveness of two handwashing promotions, which had

been selected on the basis of evidence and theory, a public commitment intervention, and the

promotion of tippy taps, in the Borena zone.

The implementation of the public commitment intervention-only had no influence on handwashing

behavior. However, since there is evidence to suggest that this specific intervention was not

implemented strictly according to instructions, it is doubtful that public commitment should be

abandoned as a handwashing intervention. Further research is necessary to better assess the

intervention potential of public commitment to promote handwashing behavior. If applied, it is

recommended to (1) carefully select the public commitment sign by thorough pre-testing and (2) to

ensure careful intervention implementation, according to instructions and close monitoring. The latter,

of course, is true of every intervention.

More importantly, the research project revealed the promising potential of a tippy tap promotion:

almost all study households were successfully motivated to construct a tippy tap, nearly all households

that constructed a tippy tap recognized it as their designated place for handwashing, and in a

substantial number of households, after 2-4 months of intervention termination, the tippy tap was

functioning, with water and soap being present. Maintenance-planning seemed to aid the positive

outcome. Therefore, further application of the tippy tap intervention can be highly recommended to

promote handwashing; if possible, in combination with maintenance-planning. The use of tippy taps is

advisable in a water-scarce region, such as the Borena zone, in particular; it uses less than one-tenth

of the water used by the traditional handwashing method; using a mug. When applying the promotion,

it should be ensured (1) that the beneficiaries contribute part of the materials and construct the tippy

tap themselves, so as to strengthen their commitment, (2) that the maintenance plans are specific and

realistic, so as to be applicable, and (3) that the maintenance-planning file is in the local language and

is eye-catching to serve as a reminder.

To conclude, the tippy tap intervention successfully induced a crucial prerequisite for handwashing: a

designated place and facility for handwashing, where soap and water is present.

51

References

Aboud, Frances E., & Singla, Daisy R. (2012). Challenges to changing health behaviours in developing countries: A critical overview. Social Science & Medicine, 75(4), 589-594. doi: 10.1016/j.socscimed.2012.04.009

Ajzen, Icek. (1991). The theory of planned behavior. Organizational behavior and human decision processes, 50(2), 179. doi: 10.1016/0749-5978(91)90020-T

Arnold, Benjamin, Arana, Byron, Mausezahl, Daniel, Hubbard, Alan, & Colford, John M., Jr. (2009). Evaluation of a pre-existing, 3-year household water treatment and handwashing intervention in rural Guatemala. International Journal of Epidemiology, 38(6), 1651-1661. doi: 10.1093/ije/dyp241

Aschalew, Belay (2013). [Handwashing promotions in the Borena zone].

Aunger, Robert V., Schmidt, Wolf-Peter , Ranpura, Ashish, Coombes, Yolande, Mukiri Maina, Peninnah, Nkatha Matiko, Carol, & Curtis, Valerie. (2010). Three kinds of psychological determinants for hand-washing behaviour in Kenya. Social Science & Medicine, 70(3), 383–391. doi: 10.1016/j.socscimed.2009.10.038

Bartram, Jamie, & Cairncross, Sandy. (2010). Hygiene, sanitation, and water: forgotten foundations of health. PLoS Medicine, 7(11), e1000367. doi: 10.1371/journal.pmed.1000367

Biran, Adam, Rabie, T., Schmidt, W., Juvekar, S., Hirve, S., & Curtis, V. (2008). Comparing the performance of indicators of hand-washing practices in rural Indian households. Tropical Medicine & International Health, 13(2), 278-285. doi: 10.1111/j.1365-3156.2007.02001.x

Biran, Adam, Tabyshalieva, Anara, & Salmorbekova, Zumrat. (2005). Formative research for hygiene promotion in Kyrgyzstan. Health Policy and Planning, 20(4), 213-221. doi: 10.1093/heapol/czi024

Cairncross, Sandy, & Shordt, Kathleen. (2004). It does last! Some findings from a multi-country study of hygiene sustainability. Waterlines, 22(3), 4-7. doi: 10.3362/0262-8104.2004.003

Cairncross, Sandy, Shordt, Kathleen, Zacharia, Suma, & Govindan, Beena Kumari. (2005). What causes sustainable changes in hygiene behaviour? A cross-sectional study from Kerala, India. Social Science & Medicine, 61(10), 2212-2220.

Contzen, Nadja, & Mosler, Hans-Joachim. (2013). Impact of different promotional channels on handwashing behaviour in an emergency context: Haiti post-earthquake public health promotions and cholera response. Journal of Public Health, 21(6), 559-573. doi: 10.1007/s10389-013-0577-4

Curtis, Valerie A., Cairncross, Sandy, & Yonli, Raymond. (2000). Review: Domestic hygiene and diarrhoea – pinpointing the problem. Tropical Medicine & International Health, 5(1), 22-32. doi: 10.1046/j.1365-3156.2000.00512.x

Curtis, Valerie A., Cousens, S, Mertens, T, Traore, E, Kanki, B, & Diallo, I. (1993). Structured observations of hygiene behaviours in Burkina Faso: Validity, variability, and utility. Bulletin of the World Health Organization, 71(1), 23-32.

Curtis, Valerie A., Danquah, Lisa O., & Aunger, Robert V. (2009). Planned, motivated and habitual hygiene behaviour: An eleven country review. Health Education Research, 24(4), 655-673. doi: 10.1093/her/cyp002

Curtis, Valerie A., Schmidt, Wolf, Luby, Stephen, Florez, Rocio, Touré, Ousmane, & Biran, Adam. (2011). Hygiene: New hopes, new horizons. The Lancet Infectious Diseases, 11(4), 312-321. doi: 10.1016/s1473-3099(10)70224-3

Debsu, Dejene Negassa. (2013). Local institutions, external interventions, and adaptations to climate variability: The case of the Borana pastoralists in southern Ethiopia. In K. Pfeifer (Ed.), Oxfam America Research Backgrounder series.

Devine, Jacqueline. (2010). Insights from designing a handwashing station for rural Vietnamese households Water and Sanitation Program: Learning Note: Water and Sanitation Program.

52

Devine, Jacqueline, & Koita, Seydou Nourou. (2010). Senegal: A hand-washing behavior change journey Water and Sanitation Program: Learning Note: Water and Sanitation Program.

Devine, Jacqueline, & Peschiera, Rocío Flórez. (2010). Peru: A handwashing behavior change journey Water and Sanitation Program: Learning Note: Water and Sanitation Program.

Federal Ministry of Health Ethiopia. (2011a). Ethiopia national hand washing communication campaign strategy. Addis Ababa: Federal Democratic Republic of Ethiopia.

Federal Ministry of Health Ethiopia. (2011b). National hygiene & sanitation strategic action plan for rural, peri-urban & informal settlements in Ethiopia. Addis Ababa: Federal Democratic Republic of Ethiopia.

Floyd, Donna L., Prentice-Dunn, Steven, & Rogers, Ronald W. (2000). A meta-analysis of research on protection motivation theory. Journal of Applied Social Psychology, 30(2), 407-429. doi: 10.1111/j.1559-1816.2000.tb02323.x

Gayo Pastoral Development Initiative. (2013). Hygiene Behavioral Change Promotion and Drought Early Warning Surveillance (DEWS): Oxfam.

Halder, Amal K , Tronchet, Carole , Akhter, Shamima , Bhuiya, Abbas , Johnston, Richard B, & Luby, Stephen P. (2010). Observed hand cleanliness and other measures of handwashing behavior in rural Bangladesh. BMC Public Health, 10(545). doi: 10.1186/1471-2458-10-545

Huber, Alexandra. (2008). The modes of operation of prompts and public commitment: A field study in Bolivia. (Master of Science), University of Zurich, Zurich.

Humphrey, Jean H. (2009). Child undernutrition, tropical enteropathy, toilets, and handwashing. The Lancet, 374(9694), 1032-1035. doi: 10.1016/S0140-6736(09)60950-8

Inauen, Jennifer, & Mosler, Hans-Joachim. (2013). Developing and testing theory-based and evidence-based interventions to promote switching to arsenic-safe wells in Bangladesh. Journal of Health Psychology. doi: 10.1177/1359105313493811

ITAB CONSULT PLC. (2011). WASH & EFSL Baseline survey in five districts (Miyo, Dire, Teltele, Dillo and Moyale) of Borena zone. Final report.

Iyer, Parameswaran, Sara, Jennifer, Curtis, Valerie, Scott, Beth, & Cardosi, Jason. (2005). The handwashing handbook. A guide for developing a hygiene promotion program to increase handwashing with soap. Washington, DC: Water and Sanitation Program.

Katzev, Richard D. , & Pardini, A. U. . (1987-1988). The comparative effectiveness of reward and commitment approaches in motivating community recycling. Journal of Environmental Systems, 17, 93-113.

Katzev, Richard D., & Johnson, Theodore R. (1983). A social-psychological analysis of residential electricity consumption: The impact of minimal justification techniques. Journal of Economic Psychology, 3, 267-284.

Kraemer, Silvie M., & Mosler, Hans-Joachim. (2012). Effectiveness and Effects of Promotion Strategies for Behaviour Change: Solar Water Disinfection in Zimbabwe. Applied Psychology, 61(3), 392-414. doi: 10.1111/j.1464-0597.2011.00475.x

Krukkert, Ingeborg, da Silva Wells, Carmen, Shrestha, Yubraj, & Duwal, Mangal Dash. (2010). Hygiene Promotion for Men - Challenges and Experiences from Nepal. Paper presented at the Asia Hygiene Practitioners' Workshop, Dhaka, Bangladesh.

Luby, Stephen P., Agboatwalla, Mubina, Bowen, Anna, Kenah, Eben, Sharker, Yushuf, & Hoekstra, Robert M. (2009). Difficulties in maintaining improved handwashing behavior, Karachi, Pakistan. American Journal of Tropical Medicine and Hygiene, 81(1), 140-145.

Luby, Stephen P., Agboatwalla, Mubina, Feikin, Daniel R., Painter, John, Billhimer, Ward, Altaf, Arshad, & Hoekstra, Robert M. (2005). Effect of handwashing on child health: a randomised controlled trial. The Lancet, 366(9481), 225-233. doi: http://dx.doi.org/10.1016/S0140-6736(05)66912-7

Luby, Stephen P., & Halder, Amal K. (2008). Associations among handwashing indicators, wealth, and symptoms of childhood respiratory illness in urban Bangladesh. Tropical Medicine & International Health, 13(6), 835-844. doi: 10.1111/j.1365-3156.2008.02074.x

53

Manun'Ebo, M., Cousens, S., Haggerty, P., Kalengaie, M., Ashworth, A., & Kirkwood, B. (1997). Measuring hygiene practices: A comparison of questionnaires with direct observations in rural Zaïre. Tropical Medicine & International Health, 2(11), 1015-1021. doi: 10.1046/j.1365-3156.1997.d01-180.x

Mosler, Hans-Joachim. (2012). A systematic approach to behavior change interventions for the water and sanitation sector in developing countries: A conceptual model, a review, and a guideline. International Journal of Environmental Health Research, 1-19. doi: 10.1080/09603123.2011.650156

Nguyen, Nga , Paynter, Nat , & Thi Nguyen, Minh Hien (2011). Vietnam: A handwashing behavior change journey for the caretakers’ program Water and Sanitation Program: Learning Note: Water and Sanitation Program.

Pallak, M. S. , & Cummings, N. (1976). Commitment and voluntary energy conservation. Personality and Social Psychology Bulletin, 2(1), 27–31.

Perlini, A. H., & Ward, Christine. (2000). HIV prevention interventions: The effects of role-play and behavioural commitment on knowledge and attitudes. Canadian Journal of Behavioural Science, 32(3), 133.

Prüss-Üstün, Annette, Bos, Robert, Gore, Fiona, & Bartram, Jamie. (2008). Safer water, better health: Costs, benefits and sustainability of interventions to protect and promote health. Geneva: World Health Organization,.

Rosenstock, Irwin M. (1974). Historical origins of the health belief model. Health Education Monographs, 15, 175-183. doi: 10.1177/109019817400200403

Schmidt, Wolf-Peter, Aunger, Robert, Coombes, Yolande, Mukiri Maina, Peninnah, Nkatha Matiko, Carol, Biran, Adam, & Curtis, Valerie A. (2009). Determinants of handwashing practices in Kenya: The role of media exposure, poverty and infrastructure. Tropical Medicine & International Health, 14(12), 1534-1541. doi: DOI: 10.1111/j.1365-3156.2009.02404.x

Scott, Beth, Curtis, Valerie, Rabie, Tamer, & Garbrah-Aidoo, Nana. (2007). Health in our hands, but not in our heads: Understanding hygiene motivation in Ghana. Health Policy and Planning, 22(4), 225-233. doi: 10.1093/heapol/czm016

Shippee, Glenn, & Gregory, W. Larry. (1982). Public commitment and energy conservation. American journal of community psychology, 10(1), 81-93. doi: 10.1007/bf00903306

tippytap.org. (n.d.). Retrieved May 14 2012 http://www.tippytap.org/the-tippy-tap

UNICEF. (2007). Supplementary materials for the hygiene promotion orientation package. Retrieved from http://www.humanitarianreform.org/Default.aspx?tabid=343

UNICEF, & WHO. (2009). Diarrhoea: Why children are still dying and what can be done. New York: Unicef.

WASH Cluster. (2011). WASH visual aids library. Retrieved 20. July, 2012, from http://ceecis.org/washtraining/index.html

WHO. (n.d.). Ethiopia situation reports. Humanitarian Health Action Retrieved June 26 2013, from http://www.who.int/hac/crises/eth/sitreps/en/

WHO regional Office for Africa. (2010a). Ethiopia: Factsheets of Health Statistics. Retrieved June 26 2013 http://www.afro.who.int/en/ethiopia/country-health-profile.html

WHO Regional Office for Africa. (2010b). Ethiopia: Health statistics profile 2010. Retrieved June 26 2013 http://www.afro.who.int/en/ethiopia/country-health-profile.html

WHO Regional Office for Africa. (n.d.). Ethiopia: Integrated management of childhood illnesses. Retrieved June 26 2013 http://www.afro.who.int/en/ethiopia/country-programmes/integrated-management-of-childhood-illnesses.html

Wilson, Jane M., & Chandler, Glenys N. (1993). Sustained improvements in hygiene behaviour amongst village women in Lombok, Indonesia. Transaction of the Royal Society of Tropical Medicine and Hygiene, 87(6), 615-616. doi: doi:10.1016/0035-9203(93)90260-W

54

Annex

Annex 1

Overview of the behavioral factors: questions, sample size, means, standard deviations and Cronbach’s alphas for scales

Behavioral factor group Behavioral factor Questions N M SD Range

Risk factors Perceived vulnerability (2 question) How high do you feel is the risk that the child under the age of 5 in your household gets diarrhea?

463 1.31 1.33 .90 04

Conditional vulnerability: not washing hands (1 question)

How likely is it that you get diarrhea if you DO NOT wash your hands with soap and water after contact with feces and before handling food?

463 2.80 1.09 04

Conditional vulnerability: washing hands (1 question)

How likely is it that you get diarrhea if you DO wash your hands with soap and water after contact with feces and before handling food?

463 1.56 1.50 04

Perceived severity (4 questions) Imagine that you contracted diarrhea, how severe would be the impact on your life in general?

463 3.63 0.48 .76 04

Health knowledge (3 questions) Can you tell me what causes diarrhea? 463 0.71 0.33 .68 04

Attitude factors Instrumental – Effort (1 question)a Do you think that washing hands with soap and water takes a lot of effort?

463 0.01 0.19 04

Instrumental – Expensiveness (1 question) a

Do you think that washing hands with soap and water is expensive?

463 0.31 0.76 04

Instrumental – Distance handwashing facility (1 question) a

Do you think that the hand washing facility is far away from your usual area of activity?

462 0.05 0.33 04

Instrumental – Time consume (1 question) a

Do you think that washing hands with soap and water is time-consuming?

463 0.06 0.35 04

Instrumental – Attractiveness (1 question)

Do you feel more attractive when you wash your hands with soap and water?

463 3.44 0.65 04

Instrumental – Nurture: Health children (1 question)

Do you think that you risk the health of your children if you don’t wash your hands with soap and water?

462 3.05 0.92 04

(continued)

55

Annex 1 (continued)

Behavioral factor group Behavioral factor Questions N M SD Range

Instrumental – Nurture: Teaching (1 question)

Do you wash your hands with soap and water because you wish to set a good example to the children?

463 3.26 0.67 04

Instrumental – Return (1 question) Considering all the benefits and efforts related to washing hands with soap and water, how much do you think is it worthwhile for you to wash hands with soap?

463 3.40 0.78 -44

Affective – Liking and pleasantness (3 questions)

How much do you like or dislike washing hands with and water?

463 3.65 0.59 .76 -44

Affective – Perfume spoiling food (1 question)

Do you think that if you wash your hands with soap before eating the perfume of the soap spoils the taste of the food?

463 0.42 1.02 04

Affective – Cleanliness (1 question) Do you think that soap is only needed when you have stubborn dirt on your hands that can’t be re-moved only with water?

463 0.68 1.19 04

Affective – Disgust (1 question) – feces

Do you feel dirty if you don’t wash your hands with soap and water after defecation?

463 3.28 0.90 04

Affective – Disgust (1 question) – food

Do you think it is disgusting not to wash hands with soap and water before handling food?

463 3.28 0.73 04

Norm factors Norms – descriptive and injunctive –feces (4 questions)

People who are important to you, do they rather approve or disapprove if you wash your hands with soap and water after contact with feces?

463 2.79 0.85 .72 -44

Norms – descriptive and injunctive – food (3 questions)

How many people of your community wash hands with water and soap before handling food?

463 2.77 0.84 .67 -44

Personal norm – obligation – feces (1 question)

Do you feel a personal obligation to wash hands with soap and water after contact with feces?

463 2.95 0.74 04

Personal norm – guilt – feces (1 question)

Would you feel guilty if you didn’t wash hands with soap and water after contact with feces?

463 3.11 1.03 04

(continued)

56

Annex 1 (continued)

Behavioral factor group Behavioral factor Questions N M SD Range

Personal norm – obligation – food (1 question)

Do you feel a personal obligation to wash hands with soap and water before handling food?

463 2.96 0.79 04

Personal norm – guilt – food (1 question)

Would you feel guilty if you didn’t wash hands with soap and water before handling food?

463 3.11 0.96 04

Compliance with HEW/CHV (1 question)

Do you wash your hands with soap and water because the Extension or Volunteers told you to do it?

463 2.25 1.22 04

Ability factors Perceived behavioral control –difficulties general handwashing (2 questions)

How difficult is it to get as much water as you need to always wash hands with soap and water?

463 0.78 1.14 .82 04

Response efficacy (1 question) How certain are you that washing hands with soap and water after contact with feces and before handling food prevents you and your family from getting diarrhea?

463 3.30 0.69 04

Self-efficacy – Maintenance and recovery – handwashing in general (4 questions)

How confident are you that you can wash hands with soap and water even if urgent tasks arise which interfere with handwashing?

463 2.28 1.18 .87 04

Self-efficacy – food (1 question) Do you think you are able to always wash hands with soap and water before handling food?

461 3.02 0.83 04

Self-efficacy – feces (1 question) Do you think you are able to always wash hands with soap and water after contact with feces?

462 3.01 0.85 04

Hindrance (3 questions) When you think about the last 24 hours: how often did it happen that you intended to wash hands with soap and water but were hindered in doing so?

463 0.57 0.71 .74 04

Self-regulation factors Action control (2 questions) How aware were you of your goal to wash hands with soap and water?

463 3.01 0.66 .52 04

Action planning – handwashing facility (1 question)

Do you have a specific place and facility for handwashing? 453 8%b 0 / 4

(continued)

57

Annex 1 (continued)

Behavioral factor group Behavioral factor Questions N M SD Range

Action knowledge – hand washing (1 question)

Can you show me how you usually wash your hands? 407 2.83 0.82 04

Coping planning – No water (1 question)

Do you have a plan what to do when there is no water for handwashing?

461 0.71 0.95 04

Coping planning – No hand washing agent (1 question)

Do you have a plan what to do when there is no soap for handwashing?

460 1.10 1.32 04

Coping planning – Forgetting (1 question)

Do you have a plan what to do to avoid forgetting to wash hands with soap and water?

452 0.20 0.69 04

Forgetting (1 question) a When you think about the last 24 hours: how often did it happen that you intended to wash hands with soap and water and then forgot to do so?

463 0.22 0.43 04

Commitment– feces (1 question) Do you feel committed to wash hands with soap and water after contact with feces?

463 3.08 0.80 04

Commitment– food (1 question) Do you feel committed to wash hands with soap and water before handling food?

462 3.11 0.74 04

Note: a Variables excluded from further analysis due to low variance. b Instead of the mean the percentage of households having a specific place and facility for handwashing is presented. In bold are presented the means of promotion activities with low prevalence in the population which could be optimized by future promotion activities.

58

Annex 2

Overview on potential behavior change techniques and tools to target specific behavioral factors

Behavioral factor Behavior change techniques Tools

Information techniques

Vulnerabilitya

Give personal risk information.

Ask beneficiaries to appraise their own susceptibility discussion on their false beliefs about their invulnerability.

Group discussion moderated by HEW/CHV

Severityb

Give personal risk information: Illustrate personal, social as well as economic consequences of diarrhea.

Explicitly arouse fear by threat inducing arguments or graphic illustrations, e.g. of severely deceased children or even death.

Fear inducing pictures or movies

Health Knowledgea

Transfer factual knowledge by:

• F-diagram as a group sorting task • Three pile sorting as an interactive learning tool • Diarrhea leaflet: re-distribute and hang up as a

prompt

F-diagram

Three pile sorting

Diarrhea leaflet

Persuasive techniques

Nurture: teachingb

Strengthen the motive with persuasive arguments causal and functionality explanations:

Remind beneficiaries of their parenting responsibilities. Explain how children learn by observation (imitating): Parents don’t handwashing children don’t learn to handwashing by imitating diarrhea contraction is more likely.

Learn the causal chain interactively by means of a sorting task using picture cards displaying good and bad observational learning chains.

Sorting task to explain the causal chain

Disgustb

Affective beliefs: attach aversion to an unhealthy behavior, e.g. disgust:

Make visible the contamination of hands with fecal bacteria (“Glow Germ”) Explicitly point out, that not washing hands with soap means eating feces (“feces”-movie or “feces eating”-task)

“Glow Germ”-demonstration

“Feces”-movie

“Feces eating”-task

Normative techniques

Descriptive & injunctive

norma&b

Norms which already exist or which do not exist yet have to be highlighted:

Reference to important others, to other communities, other regions, or even other countries. Public commitment: first the Kebele leader and village elders and then the community ‘Model household’ approach (Ethiopian government).

Commitment sign:

Flag on housetop

Coloring a finger

Nail polishing

Finger stall/ring

Publicly visible handwashing station

Personal obligationa

Bring beneficiaries to think about their responsibilities towards their families.

Invite them to imagine the concerns and regret after not handwashing and thus not conforming to their personal norm.

Interactive discussion with HEW/CHV

Compliance HEW/CHVa

Compliance is higher towards important, credible, well-liked persons select accordingly or train the staff

Staff training

(continued)

59

Annex 2 (continued)

Behavioral factor Behavior change techniques Tools

Infrastructural, skill and ability techniques

Action knowledgea

Behavioral trials where the beneficiaries handwashing:

Not in form of a mass event (joint handwashing). But guided practice in an intimate surrounding: demonstrate the behavior and observe and assist in behavior performance individual feedback

Guided practice at home visits

Motivate to provide social help in form of passing on action knowledge.

Activate self-help

Hindranceb

Infrastructure as a precondition for behavior performance:

Construction of handwashing stations to simplify performance. Assess the need and preferences prior to construction: Handwashing station wished? What type of handwashing station Where to install the handwashing station?

To deal with hindrance, maintenance self-efficacy could be targeted also see coping planning

Material to construct a handwashing station

Instruction how to construct and maintain

Motivate to provide social help in form of material (e.g. soap or water) or verbal social support

Activate self-help

Planning and relapse prevention techniques

Action planninga

Help to plan handwashing execution in more detail: Where? With what means?

Store water and soap next to each other In a place which is easy accessible Next to the area where most of the critical behavior is performed.

Plan performance with HEW/CHV’s help

Material for handwashing station and instruction how to construct and maintain

Coping planninga

Motivate and support to identify barriers and to develop detailed but realistic coping plans against them.

Develop plans with HEW/CHV’s help

Remembering/Forgetting (qualitative result)

Target remembering by means of cues or reminders:

Put soap and water together in a visible place Install handwashing station prominently in the household or compound. Make hands more salient: coloring a finger, nail polishing, wearing a finger cap or ring etc. Let children observe and alert their parents each time when they should wash their hands but don’t do it.

Coloring a finger

Nail polishing

Finger stall/ring

Good visible handwashing station

Commitmentb

Make a contract with the beneficiaries where they oblige themselves to handwashing (self-commitment).

Contract

Note. a Behavioral factor suggested to be targeted based on prevalence in the communities. b Behavioral factor suggested to be targeted based on regression results.

60

Annex 3

Maintenance-planning file

61

Annex 4

Observed handwashing after feces contact – mean observed behavior at T1 and T2

Intervention groups

FDO PCO TTO PC+TT

Point in time N M SD N M SD N M SD N M SD

Before (T1) 2 0.00 0.00 22 0.34 0.44 26 0.08 0.27 21 0.16 0.28

After (T2) 57 0.15 0.35 54 0.16 0.35 59 0.34 0.44 36 0.33 0.48

Note. Observed handwashing ranging from 0 = 0% handwashing to 1 = 100% handwashing. FDO = f-diagram-only; PCO = public commitment-only; TTO = tippy tap-only; PC+TT = public commitment and tippy tap.

Observed handwashing before handling food – mean observed behavior at T1 and T2

Intervention groups

FDO PCO TTO PC+TT

Point in time N M SD N M SD N M SD N M SD

Before (T1) 4 0.25 0.50 44 0.19 0.35 49 0.12 0.24 33 0.14 0.28

After (T2) 111 0.10 0.23 107 0.13 0.28 164 0.09 0.22 92 0.11 0.23

Note. Observed handwashing ranging from 0 = 0% handwashing to 1 = 100% handwashing. FDO = f-diagram-only; PCO = public commitment-only; TTO = tippy tap-only; PC+TT = public commitment and tippy tap.

62

Annex 5

MANCOVA results of promotion activities on self-reported handwashing behavior

Using Wilks’ statistic, there was a significant effect of the promotion activities on self-reported handwashing behavior while controlling for age and behavior at T1, = (.94), F(10, 758) = 2.55, p < .01, η2 = .03. Separate univariate ANOVAs on self-reported handwashing behavior revealed significant intervention effects on feces-, F(5, 380) = 2.88, p < .05, η2 = .04 and food-related behavior, F(5, 380) = 4.79, p < .001, η2 = .06. Age had not a significant effect on self-reported handwashing behavior, = (.99), F(2, 379) = 3.06, p > .05, η2 = .02. In contrast,

a significant effect was found for feces-related behavior at T1, = (.66), F(2, 379) = 96.37, p < .001, η2 = .34, and

food-related behavior at T1, = (.64), F(2, 379) = 105.42, p < .001, η2 = .35, insofar as high behavior at T1 was associated with lower behavior change and vise versa.

In terms of feces-related handwashing, planned contrasts comparing the f-diagram-only group with all other groups revealed that handwashing rates for this group were significantly lower than for the tippy tap-only with maintenance-planning and the combined intervention with maintenance-planning. Planned contrasts comparing the combined intervention with maintenance-planning to all other groups revealed that for this group handwashing rates were significantly higher than for the f-diagram-only and the public commitment-only groups.

Means and SD in self-reported feces-related handwashing and results from simple contrasts comparing each group with the f-diagram-only group and each group with the combined intervention with maintenance-planning

Intervention groups N M SD Contrasts FDO Contrasts with PC+TTMP

FDO 19 -.14 .36 – p < .01

PCO 110 -.12 .29 p > .05 p < .01

TTO 74 -.03 .30 p > .05 p > .05

TTMPO 91 .06 .35 p < .05 p > .05

PC+TT 52 .04 .36 p > .05 p > .05

PC+TTMP 43 .13 .30 p < .01 –

FDO = f-diagram-only; PCO = public commitment-only; TTO = tippy tap-only; TTMPO = tippy tap-only with maintenance- planning; PC+TT = public commitment and tippy tap; PC+TTMP = public commitment and tippy tap with maintenance-planning.

In terms of food-related handwashing, planned contrasts comparing the f-diagram-only group with all other groups revealed that handwashing rates for this group were significantly lower than for both tippy tap groups and both combined groups. Planned contrasts comparing the combined intervention with maintenance-planning to all other groups revealed that for this group handwashing rates were significantly higher than for the f-diagram-only and the public commitment-only groups.

Means and SD in self-reported food-related handwashing and results from simple contrasts comparing each group with the f-diagram-only group and each group with the combined intervention with maintenance-planning

Intervention groups N M SD Contrasts FDO Contrasts with PC+TTMP

FDO 19 -.23 .34 – p < .001

PCO 110 -.16 .27 p > .05 p < .01

TTO 74 -.04 .32 p < .01 p > .05

TTMPO 91 .01 .35 p < .01 p > .05

PC+TT 52 .08 .32 p < .01 p > .05

PC+TTMP 43 .12 .30 p < .001 –

FDO = f-diagram-only; PCO = public commitment-only; TTO = tippy tap-only; TTMPO = tippy tap-only with maintenance-planning; PC+TT = public commitment and tippy tap; PC+TTMP = public commitment and tippy tap with maintenance-planning.

63

Annex 6

Kruskal-Wallis- and Jonckheere-Terpstra-Test results of promotion activities on imaginative-covered recall of handwashing behavior

A Kruskal-Wallis test was conducted to evaluate differences among the intervention groups on median change in feces- and food-related handwashing. The test was significant for feces-related handwashing, H(3) = 25.35, p < .001, and food-related handwashing, H(3) = 18.74, p < .001.

Jonckheere-Terpstra-Test was applied to test for a linear trend in the data. This test revealed a significant trend in the data: when moving from f-diagram to the combined intervention, the median handwashing frequency increased. Feces-related handwashing: J = 1651, z = 4.30, p < .001, r = .20; and food-related handwashing, J = 1656, z = 4.03, p < .001, r = .18.

64

Annex 7

MANOVA results of maintenance-planning on self-reported handwashing behavior and on imaginative-covered recall of handwashing behavior: TTO compared to TTMPO and PC+TT compared to PC+TTMP

Using Wilks’ statistic, non-significant effects of maintenance-planning on self-reported handwashing behavior and on imaginative-covered recall of handwashing behavior were revealed for the TTO and TTMPO groups, = (.97),

F(4, 161) = 1.38, p > .05, and for the PC+TT and PC+TTMP groups, = (.97), F(4, 90) = 0.75, p > .05 . That is, in terms of self-reported handwashing and imaginative-covered recall of handwashing the group experiencing maintenance-planning in addition to the construction of a tippy tap (and a public promotion) did not differ significantly from the group not experiencing maintenance-planning.

2 test comparing availability of water between TTO and TTMPO

2 test revealed a non-significant effect of maintenance-planning on availability of water, 2 (1) = .15, p > .05. That is, availability of water did not differ significantly between the group experiencing maintenance-planning in addition to the construction of a tippy tap, and the group not experiencing maintenance planning.

2 test comparing availability of water between PC+TT and PC+TTMP

2 test revealed a non-significant effect of maintenance-planning on availability of water, 2 (1) = .67, p > .05. That is, availability of water did not differ significantly between the group experiencing maintenance-planning in addition to the construction of a tippy tap and the public commitment, and the group not experiencing maintenance-planning.

2 test comparing availability of soap between TTO and TTMPO

2 test revealed a non-significant effect of maintenance-planning on availability of soap, 2 (1) = .63, p > .05. That is, availability of soap did not differ significantly between the group experiencing maintenance-planning in addition to the construction of a tippy tap, and the group not experiencing maintenance-planning.

2 test comparing availability of soap between PC+TT and PC+TTMP

2 test revealed a non-significant effect of maintenance-planning on availability of soap, 2 (1) = .02, p > .05. That is, availability of soap did not differ significantly between the group experiencing maintenance-planning in addition to the construction of a tippy tap and the public commitment, and the group not experiencing maintenance-planning.

65

Annex 8

Linear regression analysis for change in behavioral factors explaining change in self-reported handwashing

Regression analyses were conducted as follows: only behavioral factors correlating significantly and with an r ≥ .20 were included into a first regression analyses; based on this analysis, subsequent analyses were run in which only those behavioral factors with a significant regression coefficient within the preceding analysis were included. Results for the last analysis for food- and feces-related handwashing are presented.

Handwashing type

Feces-related Food-related

Behavioral factor group Behavioral factor

Constant -0.36*** -0.07***

Risk factors Perceived vulnerability -0.56*** -0.10***

Perceived severity 1.60*** 0.23***

Attitude factors Instrumental – nurture: Health Children 0.76*** –

Instrumental – nurture: Teaching – 0.21***

Norm factors Norms – descriptive and injunctive 1.62*** 0.54***

Self-regulation factors Commitment 0.72*** –

Forgetting -0.09*** -0.03***

Adjusted R2 .47 .52

Note: *p ≤ .05, **p ≤.01, ***p ≤.001.

66

Annex 9

Feces-related mediation results comparing intervention groups to the control group: Effects of interventions on behavioral factors, and total indirect, direct and total effects of interventions on feces-related handwashing

Intervention group

PubCommOnly TippyTapOnly PubComm+TippyTap

Behavioral factors Effect on Effect via Effect on Effect via Effect on Effect via

vulnerability 0.10 -0.02 -0.03 0.00 -0.08 0.01

severity -0.01 -0.00 0.06 0.02 0.08 0.03

nurture – health 0.07 0.01 0.11 0.02 0.17 * 0.03

norms 0.10 0.04 0.22 *** 0.09 0.26 *** 0.10

hindrance 0.13 -0.01 -0.26 * 0.01 -0.26 * 0.01

action planning 0.00 0.00 0.96 *** -0.03 0.96 *** -0.03

commitment -0.04 -0.01 0.06 0.01 0.08 0.01

forgetting -0.04 0.01 -1.75 ** 0.03 -1.83 *** 0.03

Total indirect effect 0.03 0.15 0.20

Direct effect 0.01 0.03 0.04

Total effect 0.05 0.19* 0.25**

Note. N = 380. R2 = .46. Effect on = Effect from intervention on mediator. Effect via = Indirect effect from intervention on feces-related handwashing through mediator. Displayed are unstandardized coefficients. * p < .05. ** p < .01. *** p < .001. = p < .05, based on bootstrapped, bias corrected and accelerated 95% confidence interval with a bootstrap sample of 10’000. PubComm = Public Commitment; TippyTap = Tippy Tap Promotion.

Food-related mediation results comparing intervention groups to the control group: Effects of interventions on behavioral factors, and total indirect, direct and total effects of interventions on food-related handwashing.

Intervention group

PubCommOnly TippyTapOnly PubComm+TippyTap

Behavioral factors Effect on Effect via Effect on Effect via Effect on Effect via

vulnerability 0.10 -0.01 -0.03 0.00 -0.08 0.01

severity -0.01 -0.00 0.05 0.01 0.07 0.01

nurture – example -0.00 -0.00 0.04 0.01 0.08 0.02

norms 0.06 0.04 0.23 *** 0.13 0.24 *** 0.14

hindrance 0.13 -0.00 -0.25 * 0.00 -0.25 * 0.00

action planning 0.00 -0.00 0.96 *** -0.00 0.96 *** -0.00

commitment 0.03 -0.00 0.14 * -0.01 0.14 * -0.00

forgetting -0.23 0.01 -1.60 ** 0.05 -1.69 ** 0.05

Total indirect effect 0.03 0.20 0.22

Direct effect 0.04 0.02 0.10

Total effect 0.07 0.22** 0.33***

Note. N = 379. R2 = .53. Effect on = Effect from intervention on mediator. Effect via = Indirect effect from intervention on food-related handwashing through mediator. Displayed are unstandardized coefficients. * p < .05. ** p < .01. *** p < .001. = p < .05, based on bootstrapped, bias corrected and accelerated 95% confidence interval with a bootstrap sample of 10’000. PubComm = Public Commitment; TippyTap = Tippy Tap Promotion.