Care for left-behind children in rural China: A realist evaluation of a community-based intervention

In many countries, large numbers of left-behind children (LBC) grow up experiencing prolonged separation from their migrant worker parents. These children are known to be vulnerable to psychological and developmental problems. Drawing on qualitative interview data as well as ethnographic observations, a realist approach was applied in this study to evaluate the feasibility, acceptance, preliminary outcomes and potential sustainability of a community-based intervention to provide care and support to LBC, in a migrant-sending area of rural China. The intervention program comprised “ Children's Clubs ” in local communities, in which LBC participate in play and educational activities under the care and supervision of local volunteers. Twelve Clubs in 12 di ﬀ erent villages were evaluated for the study. In each village, semi-structured interviews were conducted with three community stakeholders, and children and primary caregivers from two families, to examine their perceptions and experiences with regard to the intervention. Our ﬁ ndings indicated that most Clubs adapted the initial program theory and implementation plan to speci ﬁ c community contexts, particularly the socio-economic si- tuation and the support from village leadership and other community members. Program implementation mechanisms consisted of integrating available resources, engaging local volunteers, and delivering various Club activities. Preliminary outcomes indicated the success in establishing a community care platform to bene ﬁ t the emotional and behavioral wellbeing of LBC, and to enhance the community support networks. To ensure pro- gram sustainability, the Children's Clubs should explore new funding schemes, expand the pool of quali ﬁ ed volunteers, and improve the curricular and activities at the Clubs.


Introduction
Over the past few decades, unprecedented migration flows in many parts of the world have had profound impacts on nuclear family structures and the home communities of migrants (Lam, Ee, Anh, & Yeoh, 2013). People migrate across countries, or within a country from rural to urban areas, mostly driven by the potential for improved economic circumstances and better opportunities for human development (de Haas, 2010). Such population movements often lead to separation between family members, due to stringent entry policies, financial constraints, and limited access to public goods in the destination (Valtolina & Colombo, 2012). This separation has raised concerns about the psychosocial wellbeing of children when migrant parents leave their children behind in the care of others, usually grandparents or other relatives (Asis, 2006). Physical inaccessibility and lack of communication tend to disrupt parent-child attachment, leading to negative emotional impact on children (Cassidy, 2008). The co-resident grandparents may also fail to fulfill childcare responsibilities, due to their old age, poor health, or lack of support (Ye & Pan, 2011).
The number of these "left-behind children" (LBC) is high in low-and middle-income countries. A review by Bryant (2005) estimated that there were about three to six million children in the Philippines left behind by parents working overseas, approximately one million such children in Indonesia, and half a million in Thailand. Nobles (2013) reported that more than one fifth of all children in Mexico experience a father's migration by the age of 15. In China, massive rural-urban migration has driven the number of children left behind in rural areas up to 61 million, accounting for 38% of children in rural China and 22% of all Chinese children (All-China Women's Federation, 2013). An estimated 29 million LBC live with neither parent, and over 2 million live alone (All-China Women's Federation, 2013).
While policy makers and academics have become increasingly aware of the economic role of migration, less attention has been paid to the psychosocial ramifications for left behind family members, particularly children (de la Garza, 2010). As children's wellbeing comprises many elements, the effects of parental absence are complex. Evidence has indicated that parental migration affects child wellbeing through a trade-off between increased family income and disrupted parental care. Empirical studies in many countries have shown that parental migration was associated with a range of psychosocial and developmental risks, including loneliness (Asis, 2006;Jia & Tian, 2010;Liu, Sun, Zhang, Wang, & Guo, 2010;Smeekens, Stroebe, & Abakoumkin, 2012), depression and anxiety symptoms (Graham & Jordan, 2011;He et al., 2012;Zhao et al., 2014), lower self-esteem (Luo, Wang, & Gao, 2012;Su, Li, Lin, Xu, & Zhu, 2012;Wen & Lin, 2012), risk behaviors (Gao et al., 2010;Hu, Lu, & Huang, 2014), as well as injury (Shen et al., 2009). Although remittances may reduce economic barriers to school enrolment in some settings, LBC were still faced with considerable challenges in academic performance due to lack of supervision (McKenzie & Rapoport, 2011) and increased domestic chores (de la Garza, 2010).
The quality of substitute care arrangements is an important factor determining the degree to which parental absence impacts children (Parreñas, 2008). When caregivers spare little time and attention on their safety, socio-emotional development, academic achievement, and daily living, LBC are at a significant risk of neglect and maltreatment (Liao, Lee, Roberts-Lewis, Hong, & Jiao, 2011). A recent study in China found that lack of family and social support exacerbated LBC's psychosocial difficulties (Zhao, Wang, Li, Zhou, & Hesketh, 2017). Although there is little research on the vulnerability of LBC to abuse, a number of extreme events such as rape, fatal injuries, murder, and suicide of unsupervised rural Chinese children have been reported by the media, with LBC also reported as perpetrators of crime (The Economist, 2015). These events together with increasing evidence about the social and developmental costs of migration, led the Chinese government to call on local authorities to expand the child protection services to cover more LBC (Man, Barth, Li, & Wang, 2017), and prioritize improvements in the education and care of LBC (Government of China, 2013).
To provide essential support to the families living apart and improve the wellbeing of their children, innovative programs targeting LBC should be developed and assessed in various local settings. Further, the sustainability of these programs is the key to address the childcare challenges across migrant-sending communities in China and globally. This study aimed to examine the feasibility, acceptance and preliminary outcomes of one such intervention program providing care and support to LBC, namely "Children's Clubs", in a county in Zhejiang Province, eastern China. This evaluation research particularly focused on perceptions and experiences of the stakeholders of the intervention. This study would also explore the sustainability of these Clubs, and thus contribute to refining the program design and improving its implementation, and potentially adapt the program in other migrantsending areas.

Study settings and intervention
Our study sites are in Kaihua County (population 350,000), in western Zhejiang Province, China. It is one of the poorest counties in the province, and a typical migrant-sending rural area. According to a survey by Kaihua Women 's Federation (unpublished), about 55% of school-age children in the county were left behind in 2012. Two-thirds of these were left-behind by both parents and typically live with grandparents who are illiterate. The other one third live with one parent, usually the mother, while the father migrates to work. Most young parents have one or two children, as a second child was allowed if the first was a girl, though this policy was lifted recently and now all couples are allowed two children.
For the intervention program we worked closely with local Women's Federation (WF), a Chinese government agency that assumes some responsibility for child welfare, with representation at each administrative level. In Kaihua, the WF runs at county, township, and village levels. Each village has a WF representative; this individual is usually a member of the Village Committee, a community mass organization of self-government (Shi, 1999).
In 2006 a six-year old girl left behind by both parents was drowned in her village in Kaihua during the summer holidays. This raised awareness in the local community of the vulnerabilities of LBC. In response, the Township WF established a "Left-behind Children's Club". This was set up in a village of Kaihua to provide care and supervision of LBC, and was the first of its kind in Zhejiang, and probably across China. Subsequently, the county government provided substantial policy and advocacy support to the initiative and many such Clubs were established across rural Kaihua. However, the initial enthusiasm waned and by 2012 most of them ceased to be operational. By early 2013 only three were left.
In early 2013 we started working with Kaihua WF on recommencing this Club model, as numbers of LBC in the county were increasing, along with concerns for their health and wellbeing. In April 2013 our intervention was launched in order to restart, reinforce, and expand the Children's Clubs program. To select participating communities, mutual agreements on program schemes were reached between the research team, county WF officials, and local township or village leaders. As a result, a total of 21 Clubs were set up in migrant-sending communities across the county. The aims of the Club program were to improve the psychosocial wellbeing of school-age children, who may be adversely affected by parental absence.
The enhanced Children's Club program comprised: (1) provision of a physical space and other necessary resources for out-of-school activities, (2) selection of Club volunteers to manage club activities, (3) workshop training of these volunteers to recognize and support the needs of children, (4) setting-up of the club where children participate in a range of age-appropriate activities: play, reading, and other learning and entertainment activities under the guidance or supervision of volunteers, and (5) development of a local community support network that provides the Clubs with sustainable funding and other resources. While the focus was on LBC, all village children were welcomed, and the Clubs was thus designated as a community asset for all children, their parents and caregivers. Hence the Clubs came to be called "Children's Clubs" or "Children's Centers". The target age range was age 5 to 15 years, though children outside this range were also welcomed. Villages were encouraged to adapt the model to local circumstances, needs and preferences.

Study design
Evaluation of the Children's Clubs intervention drew on the principles of realist evaluation, an approach for complex social and healthcare programs. It is concerned with the identification of underlying causal mechanisms (M) and how they work under what contexts (C) toward the outcomes (O), described as the Context-Mechanism-Outcome (CMO) configuration (Pawson & Tilley, 1997). Underlying mechanisms in the pre-existing contexts tend to reproduce the current social structures; hence, the realist methodology explores what program inputs may be introduced to reconfigure or activate the mechanisms differently, to generate change or transformation (de Souza, 2013). In this study, besides the preliminary examination of intervention outcomes, we focused on how the implementation of the Children's Club program was shaped, enabled, and constrained by the interaction between the context and mechanisms. As the results from the refinement of CMO configuration may contribute to further cycles of inquiry and operate in "feedback loops" (Byng, Norman, & Redfern, 2005; Pawson & Tilley, 1997), we examined the sustainability of the Club intervention and potential ways to strengthen its impact.

Data collection
Ethical approval was received from both the Ethics Review Committee of University College London and Zhejiang University prior to the field study. For evaluation we purposively sampled 12 Clubs in 12 different townships, with the aim of including villages across the socio-economic continuum. Anonymous socio-demographic information about LBC who regularly attend the selected Clubs was provided by village leaders and WF staff members. Club participants whose one or both parents had been away for over six months were included in this study. Children of deceased parent(s) were excluded. Then a selective sampling approach was used, to recruit a diversity of LBC and their families in terms of socio-economic status, family structure, caring arrangements, and child age and sex.
Interviews were conducted during October 2013 to February 2014. Informed consent from each participant, including the primary caregiver of each interviewed child, was obtained before the interview. Home visit appointments were made one day in advance. The selected families were visited to recruit children, and primary caregivers who were present. The first author also visited each of the Clubs included in the study, and participated or observed Club activities except the ones only open in the summer. Children who were not at home were interviewed at the local Children's Club sessions. Key stakeholders, including community leaders, WF representative and volunteers, were also approached in each village, and interviewed in a private room. All participants were assured of the anonymity and confidentiality of the interview.
Twenty-four LBC (11 boys and 13 girls), 12 parents (nine mothers and three fathers) and 14 grandparents, from two families in each village, were interviewed. Children ranged in age from 8 to 14 (mean = 11.1, SD = 1.6). Half of these children were single child. Five children were left behind before turning one year old, and 13 children were left behind by both parents. Three children or their caregivers refused to participate. Semi-structured interviews were conducted to explore children's attitudes to the Club, experiences of Club activities, and interactions with volunteers and other children. Similar questions were asked to parents/caregivers to elicit their perceptions of the Club program and children's participation.
Thirty-two key stakeholders (two or three for each Club) in the program implementation were interviewed, including 20 village committee members or WF representatives, nine township officials, and three country WF leaders. Five declined or were unavailable at the time. Their interviews explored the relevant community contexts, resources and personnel for the Club, enabling and constraining factors of program implementation, specific program activities, and individual and community impact of the Club. All interviews were audio-taped and transcribed verbatim with specific consent from the participants.

Data analysis
Interview data were analyzed using the thematic framework approach, in which data were classified and organized according to key themes, concepts, and emergent patterns (Ritchie, Spencer, & O'Connor, 2003). In the first stage of data analysis, for each Club, qualitative data was organized based on different components of the intervention, and by contexts, mechanisms, and outcomes. An index of themes was developed and triangulated between different types of stakeholders within each Club. The second stage focused on the comparison and synthesis of themes across the Clubs, and identification of key variables in the CMO pattern. Main themes were further examined to elicit links and causalities between specific elements of the CMO pattern in different situations. Then an overall CMO configuration was proposed to illustrate a refined program theory, to be further tested and modified.

Results
The elements across the three domains of the Children's Club intervention program, namely contexts, mechanisms and outcomes, are illustrated in Fig. 1, based on our qualitative findings. Processes and potential causal links are also indicated within and across the program elements, from community to individual levels.
An estimated 800 children attended the Children's Clubs in Kaihua in 2013, according to the records kept by the WF. Most of the children were of primary school age. On average 40 children were registered at each Club (about 20 in smaller Clubs, and up to 70 in larger ones). The overall attendance rate per Club session was around 75%. A general overview of the effects of different variables in program contexts and mechanisms is presented in Table 1. Leadership strength, community socioeconomic status, and volunteer contribution were evaluated and rated by the investigators, based on qualitative findings from the interviews as well as ethnographic observations. For these three variables, "strong" indicates an enabling effect for the program in the respective village, "weak" indicates a constraining effect, and "medium" suggests there was no obvious effect. For example, for Club number 8, community leadership did not play significant roles; strong socioeconomic development helped with program operations, but weak volunteer engagement was a barrier. The Club was based at a former schoolhouse and open on weekends and summer holidays. No full-time teachers were involved in running it. In addition, Clubs 1, 4, and 10 have been operational for longer periods but did not perform significantly better than the other ones, due to the change of village leadership and other evolving contextual factors.

Contexts
The leadership of the village or township often played a key role, in the establishment, operation, and resource mobilization of the Children's Clubs. While the program was promoted vertically through the WF staff network, the WF did not have enough influence in the government system in rural China, partly due to its limited budget. In the township government and village committee, some grassroots WF leaders may have more power since they share responsibilities in family planning promotion, a traditionally high priority issue in rural China. Yet the support from village or township leaders was still critical. As a WF village representative (at Club 1) noted: As a Village Committee Member I'm doing my best to keep the Club running. But if the key village leaders, the Village Party Chief and Village Director, weren't supportive, there would be no way to roll this out in the first place. Their initiative is really important, otherwise it wouldn't work even if we had tons of money.
Since local village committee was the largest donor for most Clubs, the economic status of the community was still crucial to the context. Besides the local infrastructure, financial and human resources, betterdeveloped villages are often geographically close to the county or township center, with easier access to funding, policy pilots, and other benefits from the higher administrative level. A WF official in a poorer township (Club 4) described the situation: Clubs in the two major townships are different from uswe probably can't do better than them. They've got more money and other resources, favorable policies from the county, and more convenient transportation as well. It's also easier for them to recruit volunteers, or get help from the larger primary schools nearby.
Additionally, Clubs were often set up in more densely populated communities, partly due to the transportation barriers to the Club and safety concerns for the children as they walked to the Clubs, according to several interviewees. These challenges were especially problematic in Kaihua's poorer areas, where many families lived in a remote spot outside the village.

Resource mobilization
For each Club, a physical space was provided free of charge by the village committee. This mostly comprised underutilized public properties, such as former school/kindergarten, village temple, village committee office space, or community center. Compared to other necessary resources, providing such a space was considered less of a challenge, even for Children's Clubs in poorer villages. Then books, play equipment and materials for sports and learning activities were purchased for the Clubs. In deprived settings, resource integration was emphasized as a solution. As a village leader reported (Club 6): These books and other materials came from donations, the village committee or the elderly association (at the village). What we've achieved so far mostly relies on people pulling together all the existing resources.
The amount of funds offered by the village committee or township government to most Clubs was usually no more than 3000 Chinese Yuan (USD 490) annually, much less than what the Club volunteers would request. A few volunteers actually reported paying for the Club expenses themselves. Only in one village (Club 7), a Club fee of 20 yuan (USD 3) per day was collected. This Club was run by a college student and his mother, providing daycare and academic tutoring, only during summer holidays.
Views about payments for the clubs were mixed. The interviewed parents and caregivers were generally not against Club payment, although they emphasized it would be the educational benefits that they were willing to pay for. However, the Kaihua WF staff were opposed to payments in principle and practice. As a county WF leader said: We don't want to discriminate against children whose parents are not able or willing to pay, and a payment system would introduce the need for formal accounting systems, which would be time-consuming, leaving less time for the more important child-oriented tasks of the club volunteers.

Volunteer engagement
Besides the WF representative, other Village Committee members and community volunteers also managed Club activities; typically one to five people were involved regularly across different Clubs. The volunteers were mostly the elder, educated villagers, such as retired village leaders and teachers.
In addition to the effects on material resources for running the Clubs, it was notable that contextual factors in community network and leadership also affected human resources in Club operations, particularly in engaging volunteers. According to a Village Committee member (Club 2): We have a strong, cohesive community with close connections between our people, so everyone would be there to help [with the Club] if needed. It's also mainly because of our Village Chief; people admire and listen to him, because he works really hard and sets an example for everybody.
In villages where the community leadership offered little support, some dedicated volunteers who were passionate about LBC's wellbeing were still able to run the Clubs rather independently, without demand for remuneration. One of them, a retired teacher, said to us (Club 11): Initially the village committee invited me to be a tutor for the Club, but later on I became the only volunteer in charge of the Club. In fact, I really enjoy doing this and I am able to contribute. I won't ask for payment and will be a volunteer forever as long as the Club exists.
Yet in some villages, the community volunteers received a small allowance, around 10-20 yuan (USD 1.6-3.3) per participation. Not all volunteers were satisfied with this: A middle-age villager implied that the allowance from the government or community should have been more generous (Club 9): We only get 20 yuan per Club open day from the township WFit's basically selfless dedication…We do have to be accountable here and make sure kids are safe and sound, and that requires careful attention and hard work.
This Club volunteer used to be a local kindergarten teacher. Some teachers who were working at local primary schools or kindergartens at the time were also engaged in organizing Club activities, especially in villages nearby. However, with a demanding full-time job, they were often unable to independently manage all Club activities. A township WF leader talked about the sustainability challenges: Everyone talks about "volunteer spirit", but there's a limit to it and you can't ask people to volunteer forever. The government has to step up its roles, because this is really a social problem.

Delivery of Club activities
The open days and hours of the Clubs were dependent on the availability and willingness of the volunteers, as well as the types of activities being organized. Most Clubs were open at weekends and during the summer holidays, but some were only able to open in the summer. In just one village the volunteer reported that the Club opened every day after school.
The content of Club activities varied depending on the capabilities of volunteers, the available resources in the Club, and the age range and distribution of the children. Volunteers with certain skills were able to organize simple art workshops. A kindergarten teacher at Club 12 would instruct painting and paper crafting, with free watercolor pens and colored papers offered by the Club. At Club 9, a volunteer would play songs on an electronic keyboard, and lead the children to sing along. Children enjoyed these extracurricular classes that were not available or affordable to them elsewhere. Additionally, some bettereducated volunteers helped children with their school homework, and even gave lessons or individual tutoring at the Club.
Yet due to limited resources, simple group games focusing on physical exercise or interactive play, rather than learning, were more common among the Clubs. These included various chasing games and running around games. While some volunteers were only facilitating, others joined the children in the games. Sports that do not require much space, such as half-court basketball and badminton, were also organized and played at some Clubs.
Children also spent time on their own activities, such as reading and playing board games with each other, when the volunteers were there just to keep order and ensure safety. The older children would often help or guide the younger ones in Club activities as well. A volunteer described a typical day in her Club (Club 10): Children do their homework and read the books herethey do enjoy the books a lot. When they're a bit tired, they play chess or board games with each other. Sometimes I tell stories to them, or take them outside and do some sports.
The long hot summer holiday period was the time that children most needed activities and supervision. Student groups from universities in Zhejiang Province provided significant support during the summer time. They would live in the village, sometimes in local residents' houses or village committee properties (such as temporary lodges turned from office spaces), and take charge of running the Club activities for one to two weeks. Several universities established partnership with a township or village in Kaihua, for their students to carry out "social practices", a type of extracurricular fieldwork program that university students are required or highly encouraged to participate in. As a WF representative mentioned (Club 4): The children were really happy when the college students were herethey teach them all kinds of stuff that our kids don't know about, including some life skills. They offered great help with homework as well.

Child experiences
The children we interviewed unanimously expressed that they enjoy coming to the Children's Clubs. Children simply appreciated having a place to go and be with friends, rather than being "stuck at home", often with elderly grandparents. As a 12-year-old girl mentioned (Club 3): I feel lonely sometimes at home with my grandparents. With the Club, "left-behind children" like me can gather here, play games and have other activities. It's really fun, and cheers us up.
One 11-year-old girl specifically mentioned that attending Club activities made her happier, and described so she could cope better with being alone (Club 12): When I was home alone, I used to be afraid there might be things like rats somewhere around, so I would hold my cat and let it stay with me in my room. Now after spending so much time at the Club, I don't feel that afraid any more. I feel like there's something else I can rely on.
Children also tended to develop strong bonds with peers and volunteers in Club activities. A 10-year-old boy mentioned (Club 1), "We hang out together a lot here; we will definitely still be friends after we go to middle schools in the county town." When asked about the most important people in her life, an 11-year-old girl pointed to the volunteer, "She isshe is the same as my mom" (Club 1).
Additionally, children were particularly interested in Club events and activities hosted by the university students in the summer. The students engaged very well with the children organizing lively and entertaining activities. Within only a couple of weeks, some of the children became quite attached to the students and reacted emotionally when they left. Several interviewees reported that children would often cry when saying goodbye to these students. A girl, 12, said (Club 7): I had a lot of fun with the big brothers and sisters from the university. When they were here I would go to the Club whenever it's open. I miss them a lot. Really hope they'll come again next summer.
Many children would like the Club to open more frequently for longer hours, especially if it was only running during the summer. A girl at Club 8 said sometimes she went to the Club only to find it closed, as she was not receiving any notifications. Children generally spoke highly about the activities at the Clubs, although some of them wished the Club had offered games and classes of their specific interests.

Perceptions of family and community members
All parents and grandparents were supportive of the Clubs and very happy that their children had the opportunity to attend. Many caregivers were especially supportive of the Clubs' role as a safe, supervised environment for out-of-school activities. Parents also commented on the potential role of Children's Clubs in improving interpersonal skills and overall development. A mother of two children attending Club 10 said: It helps the child communicate with others. Nowadays families only have one or two children. Children who go the clubs play together, so they are less likely to be antisocial; the games and activities at the clubs can benefit their development.
Although no specific educational outcome was measured, most parents emphasized the value of the educational support provided by the Clubs. According to one mother (Club 8): We hope that the Club will help children get better scores at school. Especially since no one at home is able to help them with homework and things, we appreciate the help from the Clubteachers, or even other kids. Study is the first priority.
Further, a county WF leader noted that a good Club can help strengthen the community network and reinforce solidarity, "it [the Club] tackles many families' top concern [the care of their LBC] for everyone's benefit, and it actually makes the community more united as a whole". Among issues related to LBC, a specific community-wide concern was vandalism. Several caregivers or community leaders reported the benefits of Clubs in preventing such behavioral problems. A WF official said (Club 1): Before the Club was there, kids often went to the farmland and vandalized it…They would use a stick to poke holes in cabbages and ruin the whole farm lot. It was really annoying, and caused many conflicts between the families, which were difficult to deal with. Now during holidays kids are mainly in the Club, so they stay out of trouble.

Discussion
Our results suggested that the implementation of Children's Club program was feasible and highly acceptable in a poor rural area of China, and seemed to benefit the psychosocial wellbeing of LBC and their families and communities, through the Context-Mechanism-Outcome configuration that is shown in our conceptual framework. Based on the pre-existing community contexts, this intervention program re-defined the use of available resources, roles of active community members, and practices and processes in childcare. These interrelated mechanism components were modified in various ways across the Clubs to achieve to our program outcomes.
Despite the considerable challenges in mobilizing financial resources and seeking support from local authorities, in many villages, stakeholders of the Children's Club were able to adapt the general program theory into specific local contexts. Various strategies and tactics were developed as implementation mechanisms, by taking advantage of the enabling factors and counterbalancing the constraints. As a result, Children's Clubs created a platform of community care, providing psychosocial and education support through a range of activities. With effective and sustainable implementation, the program may considerably improve the caring environment of LBC, and help children and families cope with the distress due to parental absence.
First, under different contexts, local communities managed to develop operative implementation strategies, with currently available resources. In resource-poor villages, dedicated volunteers with exceptional passion and adequate skills were still able to run a successful Club, even with little support from the community. Nevertheless, endorsement of the local leadership and solidarity of the community were important contextual factors. These ensured active resource mobilization and engagement of various stakeholders, which led to strong efforts to establish the Club and improve its performance. Meanwhile, effective funding schemes significantly contributed to program success in some better-off villages. Connections with the local school also benefited the program implementation.
In the second domain of our proposed framework, innovative mechanisms to ensure financial sustainability of the Clubs should be explored, as insufficient funding was a common challenge in Kaihua's villages. For example, advocacy and fund-raising schemes should be developed to enhance resource mobilization. When the social benefits from the economic costs for the Club program become more evident and community awareness is increased, advocacy initiatives should be able to raise funds from both public and private sectors, across the county and the province. Better-off villages may explore community funding schemes with contributions from migrant families, with the caveat of ensuring universal access to the Club. Additional resources should particularly target vulnerable LBC such as those who live in extreme poverty or face transportation barriers to the Club. Also, with regard to human resources, more retired teachers and other educated elderly from the community should be engaged to take the lead in running the Clubs, because they generally have more spare time, and have pension income that reduces economic stress. Most WF representatives were dedicated and capable in organizing Club activities, yet lack of secondary education may limit their roles in offering academic guidance for children. Given the available funding, in addition to maintaining essential Club facilities and equipment, funding allocation should prioritize remuneration for capable and responsible volunteers who do not receive salaries from the government. Further involvement of teachers and college students may add to the educational and developmental value of the Club activities. Certain training by local volunteers may help college students who did not have much relevant experience to better interact with young rural children. Then with inputs from child development experts, a handbook of Club activities or curricula for Children's Clubs may be developed, to offer stronger evidence-based guidance to Club organizers and volunteers. Meanwhile the Clubs may still opt for what would be feasible and relevant in their own contexts. To better serve those in need, schoolteachers, parents and caregivers may help identify children with behavioral problems and encourage them to attend Club activities. The activities should also aim to address challenges in family care and relationships, through activities that promote communication with migrant parents, and improve interactions with the resident parent and/or grandparents. Nevertheless, it is important to inform the parents that the Club cannot fully replace parental care.
In the domain of outputs or outcomes, the program model seemed effective in delivering the intervention components, and in relieving LBC's distress due to parental absence and their emotional and behavioral problems. Preliminary qualitative findings suggest that the Clubs contributed to decreased loneliness, less antisocial behaviors, as well as improved school achievement. The results also suggest benefits for the community in terms of enhanced cohesion and the development of a support network for children's wellbeing. This may in turn positively affect the community contexts, in terms of the attitudes of Club stakeholders, and local support and resource mobilization for the Clubs, which are crucial for program sustainability.
To our knowledge there has been only one previous published paper on an intervention targeting LBC. This is run by a non-government organization in Mexico that builds up supportive community networks by engaging LBC, caretakers, and community promoters in life skills training programs based in community centers (Givaudan & Pick, 2013). Over 2000 children participated in the program during 2009 to 2012, in rural areas of two Mexican States. However, our study is the first to study the implementation and evaluation of a community support program specifically targeting LBC.
There are certain limitations in this study. First, the outcome measures relied only on qualitative data, without quantitative metrics of child wellbeing benefits. Second, although our results indicated the positive change due to the intervention, we were unable to specifically collect baseline data to be compared at follow-up. Additionally, findings regarding the sustainability of this ongoing program may be strengthened by future follow-up assessments of the psychosocial outcomes in the longer term.

Conclusion
Our initial findings from the Children's Clubs program provide insights to its further improvement, as this Club program will soon be expanded from the initial 21 villages to all villages and other counties. Our results from these Kaihua communities across a wide economic spectrum may also inform interventions in other low-and middle-income settings, especially in rural China where the sociocultural contexts are similar. Further, the realist approach to developing and assessing programs providing care and support for LBC may be applied across migrant-sending communities globally.
Challenges to Club sustainability need to be addressed across the elements of the program mechanisms, and further studies are necessary to formally measure child and family level outcome indicators, as well as specific community impacts of the program. Policy-makers should take into account the social impact of migration in national and regional development strategies, and take the initiative to safeguard the psychosocial wellbeing of left-behind children and families. Institutional systems and accountabilities of government agencies should be established, to ensure the sustainability of social programs for the benefits of LBC.

Funding
This work was supported by the UK Economic and Social Research Council (ES/L003619/1) and the UBS Optimus Foundation (6346-UBSOF).

Conflicts of interests
None.