2.1 Ethics

The study protocol was reviewed by the Institutional Review Board committee of Emory University, Atlanta, Georgia, USA, which determined that the research was exempt from further review and approval (IRB00005955). It was approved by the Comité de Ética del Centro Universitario de Occidente, de la Universidad de San Carlos de Guatemala in Guatemala (Acta 1.23 C.E. DICUNOC); Universidad Nacional Autónoma de Honduras (CEIFCS-2023-P18) in Honduras; St. Paul’s University - Institutional Scientific Ethics Committee (ERB No. 38), the National Commission for Science, Technology and Innovation (NACOSTI/P/23/27117) in Kenya; and the Medical Research Council of Zimbabwe (MRCZ/A/3054) in Zimbabwe. All participants provided informed consent. In Guatemala, Honduras, and Kenya, participants consented verbally, and enumerators signed the consent form to affirm. In Zimbabwe, participants provided signed consents.

2.2 Study design and motivation

This convergent parallel mixed methods study [49] used (a) go-along, in-depth interviews (IDIs), (b) semi-structured observation, (c) activity-tracking smart watches, and (d) scales to elucidate women’s experiences going to water sources (Fig 1). Results are reported according to the Standards for Reporting Qualitative Research [50] (SRQR; See S1 Checklist).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Mixed-methods data collection activities to understand women’s water collection experiences.

https://doi.org/10.1371/journal.pgph.0004355.g001

Research took place across four countries (Guatemala, Honduras, Kenya, and Zimbabwe) in communities where World Vision is delivering its Strong Women, Strong World: Beyond Access (SWSW) programming. The SWSW program seeks to increase women’s empowerment and improve women’s well-being through an integrated intervention that includes water improvements and economic empowerment training and support. The SWSW program theorizes that by providing better water services, women will have increased access to water, saving them time and energy and enabling them to use water, time, and energy for other purposes, including economic activities. World Vision hypothesized that women in the SWSW program areas expend substantial time and energy to retrieve limited quantities of water. To test this hypothesis, Emory University worked with World Vision USA and local World Vision offices in Guatemala, Honduras, Kenya, and Zimbabwe to design the present study; Emory and the World Vision partners collectively agreed that research would be useful in all four countries where the program takes place to discern if and how experiences differed across these settings and to best inform the ongoing programs. We expected experiences to vary across countries based on known differences in geography, season, and water access. To carry out this research, each World Vision country office identified a local learning partner to contribute research expertise and lead data collection activities: Centro Universitario de Occidente de la Universidad de San Carlos de Guatemala, the National Autonomous University of Honduras, St. Paul’s University in Kenya, and Datalyst Africa in Zimbabwe.

2.3 Community and participant sample size and eligibility, and setting specific details

We aimed to engage at least six communities in each country to capture diversity in geography and population. Communities in Guatemala, Honduras, Kenya, and Zimbabwe were purposively selected by World Vision country offices for their engagement in World Vision SWSW programming. Almost all communities selected had water interventions planned, but not yet completed. In Kenya, World Vision had just completed construction of a borehole in one community, which was used by participants in that community and from two neighboring communities. As such, participants from three of six engaged communities in Kenya already had access to the new source.

We aimed to engage at least three women per community for a total minimum sample of approximately 18 women per country. Samples of 9–17 participants have been deemed to be sufficient for qualitative research [51], and we used this as a guideline per country sample size. Where resources allowed, we sought to increase the number per community to increase variability of data captured with the smart watches. In all settings, we recruited and enrolled a convenience sample of individuals who were eligible to participate if they were: women, bore responsibility for going to water sources for household needs, lived in a community selected as eligible, and were age 18 or older. Specific details around participant recruitment varied slightly in each country, as described below.

Research in Guatemala took place in seven rural communities in the departments of Quetzaltenango and Totonicapán, mountainous regions in the southern part of the country during the rainy season. In Guatemala, 95% of the population (91% rural; 98% urban) has at least basic drinking water service, defined as an improved source requiring no more than 30 minutes roundtrip for collection [4]. Guatemala is a multicultural, multiethnic, and multilingual country; the majority of the population (56%) is of mixed indigenous-Spanish descent and 42% are Mayan. The research team deliberately recruited members of the Mayan population; women from two different Mayan linguistic groups (K’iche’ and Mam) participated. Spanish is the official language of Guatemala and is spoken by 70% of the population; 8% of the Guatemalan population speak K’iche’ and 4% speak Mam [52]. In Guatemala, individuals were invited to voluntarily participate through local community authorities, with whom the research team discussed the research. Some women were also recruited by the recommendation of other participating women. World Vision staff did not participate in recruitment.

Research in Honduras took place in six communities across two areas—one rural and one peri-urban—in the department of El Paraíso in eastern Honduras during the rainy season. The urban area is hilly and forested while the rural area is mountainous. In Honduras, 96% of the population (91% rural; > 99% urban) has at least basic drinking water service [4]. Most (90%) of the population of Honduras is of mixed indigenous and European descent; 7% is indigenous. Spanish is the official language [53]. In Honduras, the local World Vision team introduced local community leaders (members of boards of trustees, water boards, churches) to the local research team members who explained the research project. The local leaders brought the research team to neighborhoods within the community, in which the research team went door to door to explain the research and recruit eligible and interested participants. Neither World Vision staff nor the local leaders participated in recruitment.

Research in Kenya took place in six rural communities across Isiolo and Samburu Counties during the dry season. The areas are arid and semi-arid and characterized by drought and water scarcity. In Kenya, 63% of the population (53% rural; 86% urban) has at least a basic drinking water service [4]. Kenya is also a multicultural, multiethnic, and multilingual country. English and Swahili are both official languages and there are numerous indigenous languages [54]. In Kenya, the local research team approached village leaders in the study area to explain the study and seek their help identifying potential participants. The village leaders approached potential participants to gauge their interest in participating and relayed this information back to the research team. The team then visited interested individuals in person to confirm their eligibility, provide a detailed explanation of the study, and initiate the consent process. World Vision staff did not participate in recruitment.

Research in Zimbabwe took place in six rural communities in Gwanda and Nyanga districts. The districts are drought-prone, and data were collected during the dry season. Gwanda has flat, agricultural lands whereas Nyanga is forested, mountainous and has flat agricultural lands in the low-lying areas. In Zimbabwe, 62% of the population (48% rural; 93% urban) has at least basic drinking water service [4]. Shona (80.9%), Ndebele (11.5%), and English (0.3%) are the official languages [55]. Members of the Gwanda and Nyanga World Vision offices introduced the local research team to local village leaders, who provided lists of households from which women were selected, ensuring that they lived far from each other to increase variability. The research team visited women to explain the research and invite participation. Neither World Vision staff nor village leaders participated in recruitment.

Additional information regarding the ethical, cultural, and scientific considerations specific to inclusivity in global research is included in the S2 Checklist. Data will be made publicly available upon publication at FigShare.[56]

2.4 Data collection tools and procedures

Data collection took place from June-September 2023 (Guatemala: July 11-August 2; Honduras: July 5-August 5; Kenya: June 22-July 24; Zimbabwe: July 15-July 22 and September 22-September 30.

All participants provided answers to a short survey on demographic characteristics and water access (e.g., primary water source). Women were also asked two questions informed by the Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS) to get their estimates of water collection time: [1] “How long does it take for members of your household to go to the drinking water source, get water, and come back?” and [2] “How long does it take for members of your household to go to the water source for other uses, get water, and come back?”

To understand women’s own perspectives about water collection and water work at sources, we carried out go-along in-depth interviews (See ‘a’ in Fig 1). Go-along interviews enable the interviewer and the participant to inhabit the space about which they are talking [57] and give participants active roles as guides through the space [58]. Because the environment can prompt discussion about what is observed and discussed, go-along interviews can produce richer insights than sedentary interviews [59]. This method has been used to examine women’s water fetching experiences and environmental risks in an informal settlement in urban Malawi [15] and to explore experiences of water use, access, and insecurity in Western Kenya [60].

In our study, go-along interviews started in participant’s homes, continued while going to and from the water source (during pauses in walking or at the source), and concluded back at participant’s homes. Participants were asked to go to the water source they would normally be going to on that day and to carry out the activities they would typically carry out. The interview started while with participants were at their homes to build rapport before the journey and to not slow the participant or contribute to their exhaustion while walking. Example topics included those that pertained to participants’ typical water-collection related routines, like where and when participants typically collect water, if they go in groups or alone, what activities they carry out either at the source or en route, what challenges they have come across, and what they do with children while collecting. During the journey, the data collectors were guided to make specific observations and, as needed, ask clarifying questions based on observations, including if what they were seeing was typical. For example, data collectors were asked to observe and describe water containers used during the journey, the terrain traversed, the process of water collection, activities carried out in addition to water collection (e.g., breaks, feeding children), encounters with animals or other people, and if the participant experienced any injuries.

To capture additional information about experiences not gained through other methods, semi-structured observations were carried out during the journeys (See ‘b’ in Fig 1). Observation guides prompted team members to record information about, for example, the terrain, if women were accompanied, how water was extracted and transported home, and if women engaged in other activities on their journeys.

To determine total time, distance, elevation ascent, and caloric expenditure, participants wore Garmin Forerunner 255 activity-tracking watches (See ‘c’ in Fig 1). Watches were calibrated prior to the start of water collection using each participant’s weight, height, and age to estimate caloric expenditure; caloric expenditure is reported in kcal. A data collection team member also wore a watch as a back-up, which was calibrated using the participant’s weight, height, and age. Watches started recording when the participant left their home and finished upon return.

To determine the weight of items carried, Seca 876 scales were used (See ‘d’ in Fig 1). A data collection team member determined the weight of any items, such as a water container or a child, that each woman carried when she left home. The research team carried the scales to weigh anything acquired along the way. Items were re-weighed when women left the water source before beginning the return journey, such as water containers with water or wet laundry. If anything was difficult to place on the scale alone (e.g., clothing), team members weighed the woman while holding the item(s) and later subtracted her weight.

The protocol for go-along interviews, semi-structured observations, and research equipment and the tools for collecting participant demographic information and guiding go-along interviews and semi-structured observations are included as supplement (See S1 Text and S2 Text).

2.5 Data collection training

Emory team members trained the local research teams to carry out and supervise data collection. The five-day trainings provided background information about the SWSW program and guidance on ethical research conduct, principles of qualitative and quantitative research, data management, and how to use data collection tools and instruments (watches, scales, recorders). The local research teams also reviewed all tool translations to ensure consistent meaning and understanding across languages. In the last days of training, tools and procedures were piloted prior to data collection to ensure clarity, contextual appropriateness, and their usability. In Guatemala, Honduras, Kenya, and Zimbabwe, initial piloting was conducted among the enumerator teams to identify potential challenges in administering the tools. In Kenya and Zimbabwe, additional pilot testing was carried out with community members from areas adjacent to the main study sites. This process helped the teams identify questions that needed re-wording to strengthen clarity, required additional probing, or were poorly translated. Piloting with the community members also helped the teams in Kenya and Zimbabwe to estimate the duration of each data collection activity. Based on the pilot learning, the teams revised the tools and adjusted the data collection schedule accordingly. Data generated during the pilot phase was excluded from the final analysis.

2.6 Data collection and management

Two trained local research team members collected data from each participant. Typically, one team member explained and provided the watch to the participant, collected demographic data, and led the go-along, in-depth interview while the other team member took notes, led the semi-structured observation, and recorded data on all weighed items. Working in teams of two lessened the work burden and provided team members with a greater sense of security, particularly when journeying far in new locales.

Go-along, in-depth interview, observation, and demographic survey tools were translated into Spanish (Honduras and Guatemala), Samburu (Kenya), and Ndebele and Shona (Zimbabwe). Data collection team members in Guatemala who were native K’iche’ or Mam speakers verbally translated tools from Spanish into K’iche’ and Mam, as these languages are not typically written.

Digital voice recorders were used to record go-along interviews. Audio recordings in K’iche’, Mam, and Spanish were first transcribed into Spanish by trained members of the local research teams and then translated into English by a hired translator. Audio recordings in Samburu, Ndebele, and Shona were simultaneously transcribed and translated into English by trained team members.

Demographic and observation data were recorded on paper forms. Trained team members in Kenya and Zimbabwe typed up observation notes in English. In Guatemala and Honduras, observation notes were typed up in Spanish and then translated into English by a hired translator. Demographic and select numeric or short answer data from observation sheets, which included weight data, were double entered into a database (by different members of the Emory team) and then compared for accuracy. Discordant data were checked against scanned files and corrected.

Data from the watches were synced to the ‘Garmin Connect’ web platform. No names were linked to the watches; each watch had an ID that the team used for tracking which watches were used for each journey (‘observation’). On the online platform, each observation was renamed with a unique participant ID; this ID did not contain any identifiable information. While some participant information (weight, height, age) was used to calibrate the watch in advance of the observed journey, this information was updated for each new journey observed and did not remain on the watch. Data were then downloaded and exported from the Garmin web platform as CSV and TCX files and uploaded onto a secure OneDrive folder. Data were deleted from the watches after it was uploaded.

All data, including recordings and scanned documents (e.g., surveys, notes), were de-identified, uploaded to secure OneDrive folders, and stored using unique identifiers that linked all data for each participant.

2.7 Data analysis

To learn about women’s practices, experiences, and perceptions of water collection, and the environments and factors that influence them, (aim 1; activities ‘a’ and ‘b’ in Fig 1), we conducted a rapid thematic analysis of all transcripts from go-along interviews and observation notes. The themes were chosen a priori based on the prompts in the data collection tools. Themes selected included: timing of trips to water sources, number of trips to water sources, methods used for water carriage, terrain encountered, perceived risks and experienced harms due to water collection/visiting water sources, the accompaniment of children when going to water sources and why (e.g., need to care for children, children assisted), additional work carried out at or on the way to or from water sources, perceived quality of water and any steps taken to treat water. Team member and co-author MP used Excel to organize data by creating four country-specific matrices that present themes by community. MP read each transcript and all observation notes, and any information relevant to a theme was summarized in the appropriate theme cell for community.

After data summarization by community for each country was complete, the matrices were carefully reviewed. MP and BC discussed the matrices and made iterative updates in three cases to combine themes that were closely related for clarity. Specifically, the themes “Number of trips” and “timing of trips” were combined because the timing of trips was dependent in part on the number of trips. “Terrain” and “risk” were combined because the risks identified by participants were closely related to the terrain and read as redundant when separated. “Perceived water quality” and “water treatment” were combined because participants typically spoke of the two themes together and because we had limited data on water treatment. After each country matrix was adjusted, MP created summaries for each theme by country. These summaries were then used to create one table with a country profile for each theme.

To determine actual water collection time, distance, caloric expenditure, total elevation ascent, and water volume and weight carried for the entire water journey (both to and from water sources) (aim 2; activities ‘c’ and ‘d’ in Fig 1), we calculated descriptive statistics of watch and weight data. The volume of water collected was derived from the weight data; the weight of containers was subtracted from the total weight of the water-filled containers to determine weight of the water alone. The weight of water was then converted to volume (1 kg = 1L). We generated graphs to visualize trends related to participant journeys, with attention to time, distance, caloric expenditure, water volume, and the source visited (improved or unimproved).

Time, distance, total elevation ascent, and caloric expenditure data were manipulated for select participants in each country based on identified data collection challenges, enabling us to retain and use data collected. Specifically, in Honduras, one water journey was only recorded one-way. For this observation, time and distance were doubled to estimate a round-trip. In Kenya, data for two water journeys are the mean of the participant and enumerator watches as these were not properly distinguished during data collection (individual observations were similar). Watch data from data collectors were used as proxies for two participants (Guatemala, Zimbabwe) because participant water journeys were accidentally not recorded.

To determine if women’s estimated and actual water collection times were aligned (aim 3), we calculated descriptive statistics (mean, range, SD) of the watch-measured times and women’s estimates of water collection time, specifically among the subset that collected water. We calculated the difference between estimated and actual times and identified the number and proportion of observations by country and type (drinking, other uses) that were aligned and that were discordant. We leveraged the go-along interview and observation data to determine and classify women’s intended use of the water they were collecting. Water was classified as for ‘drinking’ if a participant indicated that any of the water would be used for drinking, even if they also would use some of the water for other purposes. Water was classified as for ‘other uses’ (e.g., bathing, washing dishes) if women did not explicitly mention using any of it for drinking. (We acknowledge that the WHO/UNICEF Joint Monitoring Programme for Water Supply, Sanitation, and Hygiene (JMP) considers ‘drinking water’ to include ‘drinking, cooking, personal hygiene and other domestic uses [61]. However, we separated as described because we expect participants responded to what was asked of them, not to otherwise determined categories (See section 2.3).

Quantitative analyses were done in Stata (version 18.0), Microsoft Excel, and R (v4.2.1); qualitative analyses were facilitated using Microsoft Excel.

2.8 Risk mitigation

Like all research, we acknowledge that this research could pose some unintentional risks. Specifically, we anticipated that the wearing of smart watches could cause conflict within families or communities or impose a safety risk to participants. We took steps to mitigate these risks by making it clear within communities that devices were not a gift but were to be returned to the research team. We clearly explained risks during consents so that women could opt out if they perceived that temporarily possessing a watch would create risk. Additionally, research can pose risks to those who collect data. To ensure the safety of our data collection team members, we hired and trained individuals from the broader research areas. We never had only one research team member accompany women on water journeys; water journeys were completed in teams of two. Finally, the research team intentionally worked with World Vision teams to ensure that research would not take place in areas with high perceived risks of banditry, where participants and research team members may be targets.

2.9 Reflexivity statement

This manuscript includes authors from the United States, Guatemala, Honduras, Kenya, and Zimbabwe. The authors are both men and women and represent a range of experiences, including in program delivery and in research (both globally and specific to each country). Members of the Emory team were engaged by World Vision US to a learning partner for the Strong Women Strong World Program. In that capacity, members of the Emory team facilitated meetings with each World Vision country team to identify their research priorities; assessing water collection was a priority across country teams. Acknowledging the distance they should keep from the research given ongoing programs, World Vision country teams engaged local learning partners to lead data collection. Still World Vision country team members helped identify communities within which to collect data. Both members of the World Vision country teams and members of local learning partner teams leant their expertise about the country and communities by providing feedback on and strengthening the tools and research protocols that Emory drafted. Members of the Emory team led trainings in each country. Research participants only had direct contact with members of the local learning partner teams, who were all from the country they worked in and were fluent in the local languages. Analyses of data were led by the Emory team members who have extensive training in qualitative and quantitative analyses and the least assumptions about the data from each country. Once data analyses were completed, Emory led online workshops with members of the World Vision country and learning partner teams to discuss and raise questions about the findings (one in English with Kenya and Zimbabwe; one in Spanish with Guatemala and Honduras). In these meetings, Emory presented data and all present engaged in discussion about the data and its implications, which informed the current manuscript, particularly the discussion.

3.1 Participant information

Ninety-four women from 25 communities in four countries participated in data collection (Table 1). For 93 women, a single journey to a single water source was recorded. One woman in Guatemala participated in two journeys to capture that participant’s different water sources.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Demographic and water characteristics of water journey participants, by country.

https://doi.org/10.1371/journal.pgph.0004355.t001

Across all settings, the mean participant age was 37.1 years (SD = 13.7). The largest proportions of participants in Guatemala (12; 55%) and Honduras (11; 65%) were unmarried with a partner or single; the majority in Kenya (17; 77%) and Zimbabwe (23; 70%) were married. The average household size was 5.8 members (SD = 2.9). The greatest proportion to never attend school was in Kenya (20; 91%) and the greatest proportion to engage in economic activities was in Guatemala (20; 91%).

3.2 Reported and observed water collection practices and experiences

Sources.

Based on the demographic survey, women reported a variety of sources from which they would primarily collect water for drinking and ‘other uses.’ Overall, most participants, including all from Kenya and Zimbabwe, reported that their primary drinking water source was located outside of their dwelling, yard, or plot (83; 89%) and almost half (42; 46%) indicated that they collected drinking water from a source that we classified as unimproved (as per JMP definition [61]). However, there was variation between countries. Over half of the participants in Guatemala (15; 71%), Honduras (11; 65%), and Zimbabwe (19; 58%) reported collecting drinking water from a source considered improved, compared to just 24% [5] in Kenya. Surface water was the most reported unimproved drinking water source in Guatemala (5; 24%), Honduras (4; 24%) and Zimbabwe (13; 39%). In Kenya, most participants reported that they typically collect drinking water from unprotected dug wells (15; 71%). On average, women reported collecting water almost daily (5.6 days/week); though the mean number of reported collection days was lower in Honduras compared to the other countries (Guatemala 6.4 days; Honduras 4.1 days; Kenya 5.7 days; Zimbabwe 5.9 days) (See Table 2b and section ‘Frequency of trips’ for further elaboration from participants about how frequently they collect water and why).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. a. Thematic descriptions of women’s water collection experiences and practices, by country. b. Thematic descriptions of women’s water collection experiences and practices, by country. c. Thematic descriptions of women’s water collection experiences and practices, by country. d. Thematic descriptions of women’s water collection experiences and practices, by country.

https://doi.org/10.1371/journal.pgph.0004355.t002

While all women in Kenya reported that they would use the same water sources for drinking and ‘other uses’, sizable proportions of women in Guatemala (91%), Honduras (41%), and Zimbabwe (46%) said they visited different sources for drinking and ‘other uses.’ The proportions of women using surface water for ‘other uses’ were greater than the proportions using it for drinking in both Guatemala (+21 percentage points) and Zimbabwe (+23 percentage points). No women reported using rainwater for drinking, but approximately one third reported using it for ‘other uses’ in both Guatemala (6; 33%) and Honduras (5; 29%).

On the days they were followed, women visited improved and unimproved sources. In each of the three communities in Kenya where participants had access to a recently installed borehole, water was not always available. In each of these communities, two participants went to the new borehole and two went to an existing unprotected source. Participants also described other sources they would frequent depending on need or season (discussion of season was spontaneous, not probed). For example, participants from all countries reported collecting rainwater during the rainy season. Women from some communities discussed how they sometimes got water from family or neighbors (Honduras, Guatemala) or by paying someone, typically men (Zimbabwe). Table 2 provides thematic summaries of the qualitative data by country; S1 Table provides country-specific summaries by community.

Activities at the source.

The primary reason for most (83; 87%) participants to visit water sources was to collect water. Among those, four women (2 Guatemala, 2 Kenya) also brought clothing to wash (Table 3). Ten women (9 Guatemala, 1 Kenya) did not collect water; they visited sources to wash clothes, which they brought with them.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Water journey activity, source, time, distance, caloric expenditure, and elevation ascent, by country.

https://doi.org/10.1371/journal.pgph.0004355.t003

The methods women used to extract water varied depending on the source. In Guatemala women used buckets or ‘tinajas’ (clay pots) to pull water from a well or filled containers by dipping them into rivers. Some women in both Guatemala and Honduras used hoses to fill containers. Other methods for extracting water were more arduous. Women in both Honduras and Zimbabwe used hand pumps, which women said required strength or as noted in Zimbabwe, assistance from other women also present at the sources. In Kenya and Zimbabwe, women dug for water at riverbeds. The process required them to dig deep enough to reach water, wait for water to pool, and then remove it with small vessels until it ran clear enough to be deemed suitable for collection. Women in Kenya said they sometimes needed to dig so deep during the dry season that they would be fully inside the holes.

In addition to water collection or washing clothing, many women carry out other activities at or on the way to and/or from water points. For example, at least one woman in each country was simultaneously caring for a child, carrying them on their back or front; in Guatemala and Honduras, at least one woman in each community reported that children regularly accompany them to water sources. Women noted older children occasionally accompany them and help carry water if strong enough. While at water points women also washed the clothing they were wearing (Guatemala, Kenya, Zimbabwe) and eating utensils (Guatemala); bathed themselves or children (Honduras, Kenya); and tended to and watered small vegetable gardens near the source (Zimbabwe). While journeying to or from water points, women discussed or were observed collecting firewood (Honduras, Zimbabwe, Kenya), going to the store (Honduras), harvesting mopane worms (Zimbabwe), and watering goats (Kenya), all of which can lengthen the journey time.

3.3 The burden of collection

The time and distance traveled, the elevation gained, the terrain and risks encountered, the energy expended, the weights carried, and the volumes of water collected varied within and across countries.

Distance.

The mean round-trip distance for all water journeys was 3.5km (SD = 3.9) (Table 3). Mean distances were shortest in Guatemala (1.1km) and Honduras (1.1km), greater in Zimbabwe (2.8km), and greatest in Kenya (8.5km). Within each country, the shortest journeys were all less than 1km (Guatemala: 0.2km; Honduras: 0.2km; Kenya: 0.7km; Zimbabwe: 0.2km) and the farthest were all over 2km, with the longest in Kenya (Guatemala: 3.7km; Honduras: 2.7km; Kenya: 15.8km; Zimbabwe: 8.7km). As Fig 2 shows, those walking the greatest distances and taking the greatest amount of time to collect water were largely doing so to get water from unimproved sources.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Time, distance, and (a) caloric expenditure, (b) volume of water collected, and (c) source type among participants followed in Guatemala, Honduras, Kenya, and Zimbabwe.

(Note: The graphs include data from a woman from Zimbabwe who spent more time, went farther, expended more calories, and collected more water than other women in Zimbabwe. She tended a garden by the source; her work hauling water to it resulted in these increased, outlier values.).

https://doi.org/10.1371/journal.pgph.0004355.g002

Weight of loads carried and volumes of water collected.

The weight and volumes of water women carried varied depending on why they were going to the source (e.g., water collection, laundry), if they were carrying children, and whether they were traveling to or returning from sources (Table 4).

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Water journey transport method and item weight, by country.

https://doi.org/10.1371/journal.pgph.0004355.t004

The mean weights of what was carried to water points were higher in Guatemala (6.2 kg) and Zimbabwe (3.4 kg) compared to Honduras (1.4 kg) and Kenya (2.0 kg). Empty water containers, which were carried by over half of women followed in Guatemala (12/22) and by most women followed in Honduras, Kenya, and Zimbabwe weighed less than 1 kg on average (range: 0.2 kg in Honduras to 3.3 kg in Kenya). Children were carried in all countries, contributing to the loads carried to and from the sources. More women were observed carrying children in Zimbabwe [7] and Guatemala [6] than in Honduras [1] and Kenya [1]. Among these women, the mean child weight was 9.3 kg (range: 5.6-12.5 kg, both in Guatemala). Dry laundry was carried to water sources by 13 women, all from Guatemala [11] and Kenya [2]. The two in Kenya carried less than 2kgs of dry laundry weight. In Guatemala, the mean dry laundry weight carried was 6.5kgs (range: 2.5-14.3 kg). Sixteen women (from Guatemala, Kenya, and Zimbabwe) carried other items [range: 0.2 kg for a rope (Kenya) or item to scoop water from dug wells (Zimbabwe) to 11 kg for a bucket of manure carried one-way for gardening (Zimbabwe)].

The mean total weight of loads carried or transported from water sources was 19.3 kg. The greatest loads were observed in Guatemala (43.7 kg) and Zimbabwe (50.1 kg) compared to Honduras (31.7 kg) and Kenya (33.6 kg). The mean proportional increase in weight carried from the source compared to the journey to the source was 1512%. The increase in weights carried for the return journey is attributable to water, whether women brought back water for later use or wet laundry. The mean weight of containers filled with water during return trips was 17.0 kg. The smallest weight of a filled container was observed in Honduras (5.0 kg) and the greatest was observed in Zimbabwe (40.1 kg). Women in Guatemala who brought wet laundry back from the source carried a mean weight of 18.3 kg (range: 8.8-36.5 kg), representing a mean increase 11.8 kg over the dry laundry (range: 5.6-26.6 kg).

The overall mean volume of water collected was 16.1L. The smallest volume of water collected was observed in Kenya (3.7L) and the most was in Zimbabwe (38.2L). Across countries, the mean volumes of water collected and carried were greatest in Zimbabwe (19.6L) followed by Guatemala (15.3L), Honduras (14.8L) and Kenya (12.0L). One women from Zimbabwe reflected on the limited quantity of water she collects and how she manages: “[If] we had a lot of water, we would be using water for many different reasons; but due to unavailability of water we are minimizing the water so that it can accommodate all of the activities that we are undertaking. Therefore, we normally get water from the far sources; and yet we are able to carry a small jerrican.” (Zimbabwe)

On average, women expended 17.4 calories per liter collected; the mean number of calories expended per liter of water carried were comparable in Guatemala (7.8kcal/L), Honduras (11.1kcal/L), and Zimbabwe (11.8kcal/L), and were much greater in Kenya (39.3kcal/L). The least amount of calories exerted for one liter of water collected was in Zimbabwe (1.9kcal/L), and the greatest was in Kenya (79.5kcal/L).

The water weight that women carried or transported was equivalent to substantial proportions of their body weight. The mean proportion of collected water weight to participant body weight was 28%. Across countries, values were comparable (Guatemala: 29%; Honduras: 23%; Kenya: 25%; Zimbabwe: 32%). The smallest proportion was recorded in Kenya (9%) and the largest proportion in Guatemala (72%). (Note, the Guatemalan woman bearing this weight was carrying it unassisted (see Table 4)). As Fig 2 shows, there is a trend between the journey time, distance, and volume (or weight) carried. Specifically, women in Kenya, who were experiencing a drought, walked the greatest distances, took the greatest time fetching water and collected smaller volumes of water. Conversely, higher volumes of water were collected by those traveling shorter distances for less time.

3.4 Alignment of estimated and measured water collection time

On average, those who indicated they would use the water for drinking (n = 30) over-estimated by 39 minutes and those who collected for ‘other uses’ (n = 6) under-estimated by 11 minutes (Table 5). Average estimates from participants in Honduras, Zimbabwe, and Guatemala were not far from the average measured times, and times were greatly over-estimated by Kenyan participants. Specifically, participants underestimated across all water collection activities by 19 minutes in Guatemala and by 2 minutes in Honduras and overestimated by 11 minutes in Zimbabwe. In Kenya, participants overestimated by an average of 85 minutes. About half (47%) of participants estimated their water collection time to be within 15 minutes of the measure time (S2 Table). Among those whose estimates were greater than 15 minutes off, a greater proportion overestimated (42%) than underestimated (11%).

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Estimated and measured times to collect water overall (n = 36), stratified by if for drinking (n = 30) or other uses (n = 6).

https://doi.org/10.1371/journal.pgph.0004355.t005