Bottom-up approaches to strengthening child protection systems: Placing children, families, and communities at the center Academic research paper on "Law"

ELSEVIER

Research article

Bottom-up approaches to strengthening child protection systems: Placing children, families, and communities at the center*

Michael G. Wessells

Columbia University, USA

ARTICLE INFO ABSTRACT

Efforts to strengthen national child protection systems have frequently taken a top-down approach of imposing formal, government-managed services. Such expert-driven approaches are often characterized by low use of formal services and the misalignment of the nonformal and formal aspects of the child protection system. This article examines an alternative approach of community-driven, bottom-up work that enables nonformal-formal collaboration and alignment, greater use of formal services, internally driven social change, and high levels of community ownership. The dominant approach of reliance on expert-driven Child Welfare Committees produces low levels of community ownership. Using an approach developed and tested in rural Sierra Leone, community-driven action, including collaboration and linkages with the formal system, promoted the use of formal services and achieved increased ownership, effectiveness, and sustainability of the system. The field needs less reliance on expert-driven approaches and much wider use of slower, community-driven, bottom-up approaches to child protection.

© 2015 The Author. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction

Worldwide, the field of child protection in humanitarian settings is undergoing an historic shift toward strengthening child protection systems on a national scale (African Child Policy Forum et al., 2013; Davis, McCaffrey, & Conticini, 2012; UNICEF, UNHCR, & World Vision, 2013; Wulczyn et al., 2010). This approach aims to provide comprehensive child protection supports and promises to invigorate efforts to prevent problems of abuse, violence, exploitation, and neglect regarding children. This systemic approach is important and encouraging, but many challenges have arisen in implementing it. Many efforts at mapping and strengthening child protection systems have been top-down and failed to listen deeply to families and communities or to recognize adequately their contributions to children's protection and well-being.

A more comprehensive approach to child protection system strengthening is to intermix and balance top-down, bottom-up, and middle-out approaches. Top-down approaches help to ensure that governments have the laws, policies, and capacities that are essential in protecting vulnerable children. Bottom-up approaches work from grassroots level upward, feature community action, build on existing community strengths, and stimulate community-government collaboration.

* Author note: This article is based on a keynote address made at the International Society for the Prevention of Child Abuse and Neglect in Nagoya, Japan, September 2014. The research on which this article was based was supported by the ESRC/DfID Povery Alleviation Fund, the Oak Foundation, Plan International, Save the Children, UNICEF, USAID/DCOF, World Vision, and an anonymous donor.

Contents lists available at ScienceDirect

Child Abuse & Neglect

CrossMark

Article history: Received 9 April 2015 Accepted 9 April 2015

Keywords: Child protection Community-based systems Community ownership Children's rights

http://dx.doi.org/10.1016/j.chiabu.2015.04.006

0145-2134/© 2015 The Author. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).

Middle-out approaches, which emanate from actors such as city councils that are situated between the national and grassroots levels, embed the child protection agenda in regional centers of power. These three approaches are complementary.

Child Protection Systems

In humanitarian settings, children face myriad risks such as separation from caregivers, family violence, sexual abuse and exploitation, disabilities, violence in schools and communities, early marriage, justice-related issues, living and working on the streets, dangerous labor, trafficking, HIV and AIDS, and inability to meet basic needs. Before the advent of the systems strengthening era in 2008, most NGOs and practitioners developed projects aimed at protecting particular categories of vulnerable children. The resulting fragmented programming failed to account for the realities of the challenges children faced (e.g., a child who lived on the street one day might become a child soldier the next; Wessells, 2006; Zack-Williams, 2013). Also, the child protection sector had overemphasized the response aspects of child protection and devoted too little attention to prevention. The paradigm shift toward child protection systems serves partly as a corrective for these limitations and recognizes that systemic issues require an equally systemic approach.

What Are Child Protection Systems?

UNICEF (see UNICEF et al., 2013) has defined child protection systems as "certain formal and informal structures, functions and capacities that have been assembled to prevent and respond to violence, abuse, neglect, and exploitation of children" (p. 3). Primary components of child protection systems include laws and policies, human and financial resources, governance, means of data collection and system monitoring, child protection and response services, and nonformal supports of families and communities.

Diverse actors make up a child protection system. Government actors at different levels bear the primary responsibility for the protection of children within the state's territorial boundaries. Formal actors such as social welfare officials, police, government social workers, and magistrates lead the child protection system at national and sub-national levels. At grassroots levels, the role of nonformal actors is highly visible. As embodied in social ecological frameworks of child development (Bronfenbrenner, 1979; Dawes & Donald, 2000), children are frequently protected by nonformal actors such as families, communities, and leaders such as elders, teachers, or religious leaders. At societal levels, the media, government leaders, and civil society organizations play an important role. Because problems such as child trafficking cross international boundaries, international actors may also contribute to or support national child protection systems.

Child protection systems are inherently multidisciplinary and intersectoral. Although child protection was traditionally seen as the province of social welfare ministries and actors such as police and magistrates, other sectors play a vital role in child protection (Child Protection Working Group, 2012). For example, participation in education frequently protects children from exposure to other harms such as sexual exploitation or drug abuse (Wessells, 2011; Wessells, Kostelny, & Ondoro, 2014). Similarly, health workers frequently treat abused children or children who are vulnerable because of health issues such as HIV and AIDS or disabilities. Civil registration is essential because children who lack formal, government registration are at risk of discrimination and inability to access basic health and other services. The economic sector, too, has powerful implications, since poverty frequently contributes to child protection risks and also erodes protective factors at multiple levels (Collier et al., 2003). For these reasons, child protection systems strengthening requires a comprehensive approach that resists the deep divisions between sectors that permeate the humanitarian architecture.

Nonformal actors - including children, families, and communities - are important parts of child protection systems, although too often they are portrayed simplistically as beneficiaries or as part of the problem. Children show remarkable resilience amid adversity (Boothby, Strang, & Wessells, 2006; Fernando & Ferrari, 2013; Panter-Brick & Leckman, 2013; Ungar, 2008), navigate complex environments, and engage in self-protection. Children are social and political actors who may help peers, families, and communities to protect children. Although parents may not label their activities as "child protection," they do much to protect children by ordinary activities such as shielding infants from harm, and teaching children good behavior and how to avoid hazards. Communities include valuable protection resources such as religious leaders, teachers, elders, nurses, and natural helpers who respond to and prevent harms to children. To be sure, children, families, and communities can also be perpetrators of violations against children. Families often harm children through family violence or sexual abuse, and communities use harmful practices such as female genital mutilation. These problems, however, are not immutable, and they should not blunt our sensitivities to the central role that children, families, and communities play in child protection systems.

Community-Based Child Protection Mechanisms

Community-based child protection mechanisms (CBCPMs) are local-level groups or processes that respond to violations against children and work to prevent risks to children. CBCPMs are key parts of child protection systems since they operate at grassroots levels such as village level in rural areas and neighborhood level in urban areas, which is where children and families live and where children may be exposed to significant risks on an ongoing basis. Also, they are rich in potential child protection resources such as parents, teachers, and religious leaders, among others.

International NGOs frequently help to establish Child Welfare Committees (CWCs) or Child Protection Committees that consist of 10-20 women and men and also children. Having been trained, these Committees monitor, respond to, mitigate, and prevent various forms of child abuse. In emergency settings where supports for children have been weakened or shattered, CWCs are one of the most frequently used child protection interventions. CWCs are also used frequently in transitional and long term development settings.

Although these externally initiated mechanisms are valuable, it is a mistake to think of them as the main CBCPMs. Communities frequently have endogenous mechanisms that act locally, without facilitation or guidance from NGOs or the government. They may perform the functions of child protection even though they are not named as such. For example, in Southern Africa, where large numbers of children had been orphaned by HIV and AIDS, faith-based groups organized supports for orphans (Donahue & Mwewa, 2006; Foster, 2004). Similarly, in Sierra Leone, traditional chiefs and elders frequently help to resolve inter-family conflicts over the responsibilities of a boy or man who has impregnated a girl (Wessells, 2011). Whether CBCPMs are externally facilitated or endogenous, they should not be romanticized. In fact, there are reasons for being critical of each, and a considerable amount remains to be learned about both.

Critical Questions

The complexities inherent in child protection work and the omnipresent risk of violating the Do No Harm imperative make it important to ask critical questions such as the following.

- Are government managed child protection systems colonial impositions?

- How well do formal aspects of the child protection system fit the local context? Do they build upon or marginalize existing mechanisms or processes?

- At grassroots level, do people actually use formal means when severe cases of child abuse occur, or do they rely more on family and community supports?

- How well aligned are the formal and nonformal aspects of child protection systems?

- How effective and sustainable are community-based child protection mechanisms?

- Can one strengthen efforts to protect children at community level through community driven action?

In developing contexts, national child protection systems often fit poorly with the local context since they have been modeled after those of countries in the global North (Krueger, Thompstone, & Crispin, 2013). If governments or international NGOs impose systems that reflect outsider values and do not build sufficiently upon existing processes, local people will likely use nonformal processes that in some respects conflict with the formal aspects of the system. The resulting nonformal-formal misalignment impedes the coordination and congruence that is required for the system to function effectively.

At community level, a gnawing question has been whether CWCs are effective and sustainable. Also at question is how willing are citizens to use the expert-driven CWCs that NGOs or the government had helped to establish. Such questions prompted diverse child protection actors to conduct in 2009 a global evaluation of CBCPMs, the findings of which are relevant today.

A Global Interagency Review of CBCPMs

The review of CBCPMs was guided by a Reference Group of NGO, UN, and donor agencies that worked on child protection. The Reference Group was coordinated by Save the Children and included NGOs that often helped to set up CWCs in humanitarian settings. The review examined 160 published and mostly grey literature reports that evaluated CBCPMs mostly in Africa and Asia (see Wessells, 2009a). The reports included both NGO facilitated CBCPMs and endogenous processes, and a small number of evaluation reviews from the health sector related to children affected by HIV and AIDS.

The review found a very weak evidence base regarding CBCPMs. Many evaluations lacked robust designs, were conducted post hoc without baseline measures, and did not include the comparison conditions that are necessary for making causal attributions regarding the effectiveness of CBCPMs. Most evaluations focused on process and output indicators rather than on outcomes for children and families. For these reasons, the review sounded an urgent call for more systematic evaluations and attention to actual outcomes for children.

The review identified seven factors that contributed to the effectiveness and sustainability of CBCPMs: community ownership and responsibility; incorporating and building on local resources; leaders' support; genuine child participation; ongoing management of issues of power, diversity, and inclusivity; external training and resources; and linkages (including nonformal-formal linkages) with other parts of the child protection system. Since community ownership was the most important determinant of whether a CWC was effective and sustainable, it warrants additional discussion.

Community Ownership

Community ownership refers to community members' sense that the CWC or other CBCPM is 'ours' and that they have the power and the responsibility to support vulnerable children. Community ownership is, then, the key component of community-driven action. Ownership is frequently indicated in the language that local people use to describe the CWC.

Table 1

Four different means of engaging with communities and the resulting level of community engagement for each.

Category Description Ownership level

1 Direct implementation by agency: The agency is a service provider; community members are beneficiaries. Low

2 Community involvement in agency initiative: The agency is a promoter of its own initiative, a planner and a trainer, and community members are volunteers and beneficiaries Low to moderate

3 Community owned and managed activities mobilized by external agency: The agency is a catalyst, capacity builder, a facilitator of linkages, and a funder after community ownership has developed. The community members are analysts, planners, implementers, assessors, and also beneficiaries. Moderate to high

4 Community owned and managed activities initiated from within the community: The agency is a capacity builder and funder, and community members are analysts, planners, implementers, assessors, and also beneficiaries. High

Statements that "this is a UNICEF or NGO project" indicated a low or moderate level of community ownership. Higher levels of ownership were indicated by people saying, "This work is our way of supporting our children." Psychologically, high levels of ownership for a CBCPM entail a collective sense of responsibility that motivates individuals and groups to insure that the CBCPM is effective and yields lasting benefits for children.

An important question was whether the NGO facilitated CWCs had the high levels of community ownership that contribute to maximum effectiveness and sustainability. Because the level of ownership relates to how agencies engage with communities, we analyzed the evaluations using a previously developed typology (Behnam, 2008) regarding the NGO mode of engagement with the community.

Table 1 shows four broad categories or modes of community engagement. First is a direct service approach that generates low levels of ownership. For example, in a refugee camp where children face imminent danger, an NGO might form a CWC using staff who receive reports of abuse or violence against children and who respond through the authorities. This approach treats community people as beneficiaries and stimulates little community ownership since the CWC formation and work did not include community members. Category 2 is an expert-driven, partnership approach in which the agency conceptualizes what needs to be done, presents the idea to the community, invites them to participate as volunteers and beneficiaries, and provides training and funding as needed. For example, an NGO might raise awareness in a rural community about children's protection issues, suggest the need to have a CWC, and ask the communities to partner in its formation, development and work. The communities would help to select the CWC members, who volunteered their time and effort. Community members would likely experience some ownership over the CWC since they had partnered on its development and work. Because the idea, technical expertise, funding, and vision came from the NGO, however, local people would likely see it as an NGO project and would not experience a high level of ownership for the CWC.

Categories 3 and 4 feature greater community power and decision making and engender higher levels of community ownership. In category 3, the agency initiates discussions about CBCPMs but acts less as an expert and more as a facilitator of community planning and action. For example, an NGO might facilitate collective dialogue about how to support vulnerable children. The community itself might identify the key issues and decide to address them using the already existing Village Development Committee, working in tandem with women's groups, religious groups, and youth groups. Because the community members make the decisions and take primary responsibility for the work, this approach generates relatively high levels of community ownership. In category 4, communities initiate and manage activities from within, with little or no support from an external agency. For example, church groups in Kenya might act on their own initiative to support orphans and other vulnerable children. Because such action comes from the community itself, it generates high levels of felt responsibility and collective ownership.

Of the 160 papers reviewed, 112 had sufficient details about the mode of community engagement to enable classification by category. As Fig. 1 shows, the dominant or most frequently used mode of community engagement was a category 2 or partnership approach. This approach has discernible strengths such as such as rapid start up and time urgent response to harms to children. But it clearly lacks the high levels of community ownership needed for achieving maximum effectiveness and sustainability. This finding should serve as a wakeup call regarding the need to build higher levels of community ownership in child protection work.

The review also identified numerous factors that promoted community ownership. Primary among these was a sense of collective responsibility by community members. This was grounded in commitment to support vulnerable children, collective empowerment, and having the actual power to make key decisions and guide the work. The level of community ownership was higher when people identified the work as "ours" and took responsibility for its effectiveness. The level of community ownership was reduced by external attributions (e.g., "This is a Save the Children project"), which emphasized agency responsibility for the work.

Community ownership was boosted as well by the internal mobilization of community resources, notably the volunteer spirit. As parents, teachers, elders, youth leaders, religious leaders, traditional healers, and others empathized with vulnerable children and engaged in narratives such as 'these vulnerable children are our responsibility,' they decided to volunteer their time and used their networks and expertise to help support vulnerable children. In some cases, they collected money, food or clothing for vulnerable children, made land available for gardens or dwellings, or supported ceremonies that aided

Category 1 Category 2 Category 3 Category 4

Fig. 1. The percentage of reviewed papers according to their approach to engaging with communities.

vulnerable children. A circular relationship existed between resource mobilization and community ownership. The voluntary participation of diverse people in the work elevated the sense of community ownership, which in turn heightened their desire to devote their time, energy, and resources to supporting vulnerable children.

To promote community ownership, external agencies kept a steady emphasis on communities' decision making and action on behalf of children. Rather than acting as experts, their staff served as facilitators who listened, learned about local power dynamics, identified natural helpers, enabled child-focused dialogues, and patiently cultivated community awareness of and action to support vulnerable children. This method of patient cultivation was intentionally slow and reflected the community's own pace and readiness to include different people in dialogues, discuss issues iteratively, and take collective decisions. The advanced skills of facilitation needed for this process may not be present in many child protection agencies, because the humanitarian system prizes technical expertise and rapid, scalable action.

Factors that limited community ownership included the early introduction of large sums of money before the community had gained deep awareness of children's vulnerabilities or developed a sense of collective responsibility. The early introduction of money led local people to get involved in the work as a means of earning money rather than fulfilling the collective responsibility to protect children. Other limiting factors were agency-oriented modes of engaging with the community, the failure to build on local ideas and resources, and the use of didactic, top-down approaches. Little ownership developed if the external agency controlled the power, took the decisions, and "planted its flag" by, for example, posting its own signage and name. Ownership was also blunted when agencies acted as "experts' who used didactic methods to educate local people about child protection. Didactic approaches frequently led community people to see the work and the ideas as agency lead rather than their own. In some cases, there was backlash against outsider-imposed ideas such as "child rights" that did not fit the local culture.

A Critical Perspective

Although community ownership and people's power is important, it is also essential to ask who is the community and who holds the decision making power. It is not uncommon for NGOs to work with community leaders to convene several open discussions at which decisions are made in regard to forming CWCs. This approach is problematic because quite often there are marginalized people, including children and the poorest of the poor, who either do not attend such gatherings or remain voiceless when they do attend. Because the decision making process is not inclusive, it is ill advised to speak of community ownership. Further, the people who dominate such gatherings may be the chief and his relatives or other members of the local elite. Indeed, the review reported that unintended harm was sometimes caused by the reinforcement of local power structures that excluded or marginalized various people or sub-groups. These and other ethical issues may be managed by maintaining a critical stance and working with a social justice lens that makes inclusiveness a high priority.

Ethical issues arise frequently in work on child protection (Graham, Powell, Taylor, Anderson, & Fitzgerald, 2013; Wessells, 2009b) and surfaced in the review. In some cases, governments or child protection agencies imposed CWCs on communities, with little input or support from local people. This impositional approach evoked frustration and resentment, and it also undermined willingness to use the CWCs. CWCs were sometimes created in the manner of parallel systems that had few connections with existing community structures and resources. The unfortunate result was the weakening or marginalization of the existing supports, which had been most likely to be sustainable.

Community-Driven Action: An Exemplar from Sierra Leone

Following the global review, approximately 15 NGOs, UNICEF, and donors formed the Interagency Learning Initiative on Community-Based Child Protection Mechanisms and Child Protection Systems. Wanting to develop and test a different, community-driven approach to child protection, the Initiative chose to investigate whether CBCPMs can be made more effective through community driven action that strengthens linkages with selected formal aspects of the child protection system. We envisioned a bottom-up approach to system strengthening since the community drives the selection and nourishes the collaboration with formal actors. To build high levels of community ownership, we used Participatory Action Research (PAR) focused on supporting vulnerable children. PAR is a family of methods that are grounded in the work of Paulo Freire (1968) and others that feature local empowerment and social justice and that resonate with methods such and participatory rural appraisal (Chambers, 1994). In PAR methods, communities hold the power since they define the problem and then take self-designed steps to address it through community driven action. I will describe our work and interim findings in Sierra Leone, focusing on work done between January 2011 and June 2014, before the Ebola crisis began.

The Sierra Leone Context

Sierra Leone has a predominantly rural population of approximately six million people. It ranks near the bottom (177th) on the Human Development Index (UNDP, 2013) and has an average life expectancy of 46-47 years. In many rural areas, people engage in traditional beliefs and practices such as burial practices and conflict resolution by chiefs and elders. The country is divided into 144 chiefdoms, each led by a Paramount Chief.

From 1991 to 2002, Sierra Leone suffered a brutal war in which children were key actors (Denov, 2010; McKay & Mazurana, 2004; Wessells, 2006). To address the child protection issues during and after the war, diverse NGOs formed and trained CWCs nationwide. Wanting to support children's rights and well-being, the government enacted the Child Rights Act of 2007, which mandated the establishment of a 13-member CWC in each village. Their role was to monitor, respond to, and prevent violations against children and refer criminal violations to authorities such as the police. Subsequently, the government decided for practical reasons to form CWCs at Chiefdom rather the village level. NGOs aided the implementation by training CWC members and providing child rights education at community level. The child rights education was conducted primarily in a didactic, top-down manner in which NGO experts taught local people the concepts and language of international child protection and the harm done by customary practices such as corporal punishment.

The Research Design

As Fig. 2 shows, the research featured ethnographic learning and then used a two-arm randomized cluster design to test the community-driven intervention (Stark et al., 2014). The idea was to learn intensively about community processes and resources and to build on these in the intervention phase. In each of two districts, there were two clusters of three communities. These clusters were early intervention and delayed intervention conditions, respectively. In a mixed methods approach, a contextualized survey was used to measure children's risk and well-being outcomes at baseline (T1) and at two subsequent points (T2 and T3). I will discuss the T2 data since the Ebola crisis has postponed the T3 data collection. Qualitative data came from monitors living in the communities, group reflections by community members, and interviews with children, adults, and health workers.

A Multi-Partner Approach

To make the research more useful, achieve greater buy-in, and avoid an extractive approach, we collaborated extensively with local actors in all phases of the research. With UNICEF's support, we worked closely with the Ministry of Social Welfare Gender and Children's Affairs and the National Child Protection Committee (CP Com), which included representatives from child protection actors such as Save the Children, World Vision, Plan International, and Goal.

The CP Com helped to adapt the research to the Sierra Leone context, advise on research ethics, and collaborate on the selection of sites. To enable deeper learning, we focused on two districts only. However, we wanted to include ethnic and linguistic differences and work in areas regarded as typical of their respective regions. Following advice from CP Com members, we decided to work in Moyamba District, a southern, Mende-speaking area, and Bombali District, a northern, predominantly Temne speaking area. Consultation with the government and the child protection NGOs within each of those districts was instrumental in selecting two clusters of three villages each. The clusters were approximately matched (hard data were scarce) in regard to child protection issues, supports and services available, and socioeconomic status.

This collaborative approach in which the CP Com advised and the Columbia Group provided the research technical expertise set the stage for having a policy impact subsequently. Because the CP Com had seen the relevance of the research, it supported and sought to learn from the research.

Identification of 2 districts, 4 chiefdoms, 12 villages

Ethnography on all 12 villages

■ Intervention Cluster □ Comparison Cluster

Fig. 2. The action research design.

Ethnographic Phase

The purpose of the ethnographic phase was to build trust and learn deeply about children's issues, which could be fed back to communities in a way that sparked collective reflection and community-driven action. Trained Sierra Leonean researchers lived and worked in the villages for several weeks and were overseen by experienced mentors. The researchers avoided international terms such as child protection or child rights and attempted to capture local people's own categories and narratives. Positioning themselves as students, the researchers asked communities to teach them about their views of children and to answer questions such as: Who is a child? What are the main harms to children? What typically happens when a particular harm arises? What do people usually do if a criminal violation such as the rape of a child has occurred?

The researchers used methods such as participant observation, in depth interviews conducted one on one, and group discussions. The interviews and group discussions, which were conducted in local languages, asked open-ended questions and were highly flexible as they followed the participants' line of thought. Recordings of the interviews and discussions were translated into English, with the field mentor having checked their accuracy. Interviews were also conducted in English with child protection workers such as police, government social workers, and NGO workers.

To identify whether local views varied according to gender and age, the group discussions involved groups of ten people who were either all women, men, teenage girls, or teenage boys. In group discussions, participants first identified the main harms to children and ranked the top three harms that were most serious or concerning to people. For each of the top harms identified, the researchers probed and mapped the typical pathways of response and identified the actors, the decisions and action taken, the outcomes, and how different people viewed the outcomes.

The results showed that local people had very different views of children and harms to them than those enshrined in international standards (Wessells, 2011; Wessells et al., 2012). A child was defined not by age but as someone who was dependent on parents or not sexually active. The top harms to children were out-of-school children, teen pregnancy out of wedlock, heavy work, and maltreatment of children who did not live with their biological parents. Frequently mentioned

harms also included: child beating, cruelty, sexual abuse (including rape and incest), neglect and bad parenting, witchcraft, abduction and ritual murder, and child rights. Ordinary citizens did not mention problems such as female genital mutilation, which is widespread in Sierra Leone (UNICEF, 2013).

That child rights was frequently mentioned as a harm to children surprised us since child rights are intended to support and protect children (See also Behnam, 2011). Frustrated parents said that child rights had undermined their parenting since they could no longer use corporal punishment to teach children good values and behavior. Also, child rights workers were criticized for placing too little emphasis on children's responsibilities. Cases emerged in which people's reactions to child rights had caused harm. One upset mother told how she could no longer beat her willful girl because she feared being reported to human rights workers, so she now disciplined the child by denying her food.

The pathways of response to harms to children indicated a profound disconnect between the nonformal and formal aspects of the child protection system. Although child protection workers saw CWCs as being moderately functional, most ordinary citizens did not mention or report through the CWCs. For over 90% of the cases of harms to children, ordinary people preferred to use traditional processes through the chief or the extended family. For example, if a girl had become pregnant, the girl's parents identified the perpetrator, went to his family, and negotiated a compromise wherein the perpetrator's family supported the girl during her pregnancy, paid for her lost time in education, and obliged their son to marry the girl. Girls were not always happy with this outcome, though families saw it as protecting her and also the family's honor. Even in the case of rape of a child, ordinary citizens said that most people were unwilling to report the crime to police due to distance from the authorities, time away from farming, or concerns about inaction. For cultural reasons, people regarded their village as family and said they did not discuss family matters with outsiders.

These findings and others (Thompstone, 2010) indicated that the CRA was not working and that people overwhelmingly preferred to use nonformal pathways and supports rather than formal channels. They caution against the top-down imposition of CWCs that have low levels of community ownership and are poorly connected with the existing community strengths. They also point to the need for more respectful approaches for introducing child rights. A high priority then was to enable community driven links and collaboration with formal aspects of the child protection system.

The Planning Process

The mentors fed the ethnographic findings back to the communities at a large gathering and asked whether the findings were accurate. People confirmed the findings and thanked the team for sharing them. Having a holistic picture of the harms to children, people reflected on what they themselves should do about them. This was an important moment in terms of collective empowerment and local ownership.

The communities' planning process and the facilitation involved in it were a sharp departure from quick, expert driven approaches and merit description because they are not part of most agencies' child protection repertoire. To create an inclusive planning process that enabled meaningful participation by girls and boys, the team hired and trained two Sierra Leonean facilitators - one for Moyamba and one for Bombali - who spoke the local languages and understood the local context and social norms. Considerable training was provided on how to ask open ended questions that stimulated group awareness of an issue, probe for or even quietly challenge hidden assumptions, and enable problem-solving discussion on how to address the issue. Working subsequently with the senior mentors and also international researchers, the facilitators lived several days at a time in each village, rotating to other villages subsequently.

The first task was to help communities choose which harm to children they should address. In this and subsequent work, the process was as important as the decisions taken and the results accomplished. To avoid the chief or other power broker controlling the planning, the mentors first talked with chiefs and explained the importance of real community ownership and participatory planning. The Paramount Chiefs agreed not to guide the process but had their authority respected by receiving regular updates and reports. Next, the facilitators asked questions designed to stimulate thinking about how to include all community members in the planning. The dialogue followed the general schema below:

Facilitator: "How could the whole community choose which harm to children to address?"

Participant: "We should all have a meeting in the Chief's baray."

Facilitator: "Does everyone participate in the Chief's baray?"

Participant: "Yes - everyone comes."

Facilitator: "I met a blind girl here. Does she come to the baray?"

Participant: "No. She cannot come to the baray."

Facilitator: "Are some people in this village much poorer than others?"

Participant: "Yes. Some are so poor they cannot feed their families."

Facilitator: "Do the poorest of the poor come to the baray meetings?"

Participant: "No. They have to work so long that they have no time."

Facilitator: "Do girls and boys come to the baray meetings?"

Participant: "Yes, and they are really happy to be there."

Facilitator: "Would an average girl, say a 10-year-old, speak at the baray?"

Participant: "Probably no. It's a matter of respect for her parents and the elders."

Dialogues with many people in groups and one-on-one led the communities to develop a three-pronged planning approach. First, a planning group of adults and children facilitated collective planning and made home visits to exchange ideas with people who were unable to participate in baray discussions. Second, sub-group discussions occurred among teenage girls, teenage boys, women, men, and elders, respectively, thereby enabling participants to discuss sensitive issues more openly. Third, regular baray discussions continued but with feedback from the small group discussions. This process, which gave voice to many people, became sufficiently popular that communities subsequently used it in making any community decision.

To facilitate planning within each cluster of three villages, communities formed an Inter-Village Task Force. Each village elected five members, one from each of the sub-groups identified above, to serve on this Task Force, making a total of fifteen members. To support the planning process, the Paramount Chief designated the Chiefdom Speaker, the operational head of the Chiefdom, to listen in on and encourage the Task Force discussions and report back. Having received suggestions from each community, the Task Force developed suggestions, which were then discussed by communities using the tripartite approach described above.

The planning discussions simultaneously considered intervention priorities, the likely nature of the intervention, and issues of feasibility and government collaboration. Local people did not want to select an issue without having an idea of the intervention and its feasibility. Wary of false promises, people wanted to select an issue on which the government could actually deliver promised support.

After extensive discussions, the communities in both districts selected teenage pregnancy as the priority issue to be addressed. This selection likely owed to the profoundly negative impact of teenage pregnancy (Coinco, 2010; Wessells, 2011), which led children to drop out of school and forced young mothers to engage in transactional sex in order to survive. Also, traditional approaches such as passing community by-laws had failed to stop teenage pregnancy. Numerous community members had heard of Marie Stopes' work and wanted their community to benefit from elements such as family planning, sexual and reproductive health education, and life skills. Discussions between the mentors and the district medical officers indicated that the government was willing to provide at no cost various forms of contraceptives such as condoms and implants, and to train the health post nurses how to do the implants safely. The selection of teenage pregnancy may have owed also to the sharp upsurge in national attention to the problem of teenage pregnancy. So widespread and significant was this problem that in 2012, the president of Sierra Leone declared a state of emergency and called on all citizens to help prevent teenage pregnancy.

For the detailed intervention planning, communities decided to add youth leaders to the previous Task Force members. This addition gave teenagers an even louder voice and more influential role and recognized that if teenage pregnancy were to be reduced, teenage boys and girls would need to play a central role in changing local social norms.

Following discussion with Marie Stopes and Restless Development (a UK-based NGO), the Implementation Planning Task Force identified the activities that were needed under each of the three elements of the intervention. Concerning family planning, they planned community education and dialogues about puberty, the causes of pregnancy, the value of family planning, and decisions of young people to use contraceptives and delay sexual activity. Concerning sexual and reproductive health, they planned for community members to learn about contraceptives and how to prevent sexually transmitted infections, address misconceptions such as the view that the use of contraceptives would impair one's future ability to have children, and manage healthy pregnancies when they occurred. Concerning life skills, they planned for teenagers and young people to become role models for social change. In particular, they wanted teenage girls to be able to say 'No' to unwanted sex and the advances of men in powerful positions, talk about and negotiate sexual behavior with peers, and avoid the customary practice of girls and boys drinking alcohol and having impromptu, unprotected sex. Ordinary teenage boys and girls and also adults were responsible for implementing activities, helping to change social norms, and obtaining positive results.

Each community selected five teenage girls and boys to be peer educators who worked with the Task Force members and diverse sub-groups within the communities to implement the intervention. The NGOs provided week long participatory workshops for the peer educators that taught basic concepts and stimulated thinking about key messages to be disseminated. The initial training took place in June-July 2013, with follow up training provided in March 2014. Following the initial training of the peer educators and the Government training of the health post staff, the full intervention began. However, there was a deliberate blurring of the planning and implementation processes, as the planning discussions contributed to the wider awareness raising and mobilization that supported the intervention.

The Community-Driven Intervention

The intervention process was documented by having the Sierra Leonean facilitators also work as monitors who used participant observation methodology to capture various activities. After the intervention had begun in June 2013, they focused 75% of their time on monitoring.

Table 2 shows the 10 key elements of the intervention process that emerged over time. At the heart of the process were collective dialogues and decision making. Throughout the intervention, task force members or village authorities continued to organize dialogues in the barray to discuss issues such as puberty, family planning and contraception, sexually transmitted infections, the importance of delaying sex or engaging in safe sex. The peer educators and task force members made home visits in order to learn the views of marginalized children and families. Many dialogues followed teenagers' performances of

Table 2

Ten key elements of the community-driven intervention.

Element

Description

Collective dialogue, awareness raising, and negotiation

Collective decision-making, empowerment, and responsibility

Linkage of communities with health services

Peer education

Use of culturally relevant media

Child leadership and

messaging Inclusion and outreach

Parent-child discussions Role modeling

Legitimation by authority

Ongoing collective reflection and dialogue occurred in the barray and in sub-group discussions among teenage girls, teenage boys, adult women, adult men, and elders about issues such as which issue should be prioritized, how to address the issue, and diverse aspects of teenage pregnancy. These dialogues raised collective awareness, negotiated points of disagreement, and created readiness to receive various messages associated with teenage pregnancy. The communities themselves decided which issue to address, how to address it, who should represent them on the inter-village Task Force, and other key issues. Seeing the decisions and intervention process as 'theirs,' they took responsibility for insuring its success. They empowered each other by encouraging participation, mobilizing different sub-groups, and creating public activities that engaged increasing numbers of people.

The District Medical Office agreed to supply various contraceptives and train health post nurses to do procedures such as implants. At local levels, supportive partnership developed between local people and health post staff. People visited the health post for contraceptives and invited nurses to visit the villages and help to educate people about issues related to puberty, sex, reproductive health, and pregnancy.

Community selected Peer Educators educated peers and adults on issues of family planning, sexual and reproductive health, and life skills. Less formally, peer education occurred also through discussions of parents with each other, of women with each other or with men, etc.

Peer educators and others conducted culturally appropriate educational activities such as song and drama. Afterwards, teenagers and adults discussed the realistic nature of the song or drama, debated the implications, and affirmed the benefits associated with young people making good life decisions.

Teenage girls and boys were prominent actors in developing and implementing the intervention. They created and spread messages that were likely to influence other young people.

Representatives of diverse sub-groups (teenage girls, teenage boys, adult women, adult men, and elders) took part on the Task Force that facilitated much of the community driven work to prevent teenage pregnancy. Home visits brought forward the voices and perspective of marginalized people such as out of school children and children with disabilities.

Parents and children began to discuss puberty, sexual and reproductive health, sex, and teenage pregnancy prevention. In some cases, the children helped to correct parental misconceptions.

By taking part in activities such as dramas and singing, young people, including teenage boys, signaled that they wanted to prevent teenage pregnancy. Parents and elders, too, provided positive role models by showing their commitment to reducing teenage pregnancy.

The Paramount Chiefs publicly supported the work to prevent teenage pregnancy and encouraged people to get involved in the intervention. Other community leaders such as teachers and religious leaders also encouraged support for preventing teenage pregnancy.

dramas and songs, and spontaneous dialogues occurred over meals, in schools, and during daily activities. Both inside and outside the barray, dialogues occurred that contributed to social change. Initially, elder men resisted the use of contraceptives on the grounds that this would corrupt girls' morals. Over time, though, mothers persuaded the men that their daughters were getting pregnant and that contraceptives were needed.

At each point, the community drove the intervention and decided whether, when, where, and how to implement the intervention. The task force facilitated this process, yet there were many smaller decisions made as well by various community members. Although the implementation followed a community-designed plan, it deliberately did not follow a protocol but had an improvisational quality that drew on the agency and creativity of teenagers and other people. This flexible approach enabled the intervention process to adapt to new opportunities and challenges that arose in response to increased community awareness or, alternately, resistance. Following a social norms change approach, insiders drove the process and decided which activities were needed and how to implement them.

Throughout the intervention, awareness of and knowledge about various aspects of family planning, sexual and reproductive health, and life skills was promoted through youth driven messages and the use of culturally appropriate media. The peer educators, who included leaders of youth groups, conducted various activities such as songs and dramas (vignettes) which fit the local culture and stirred keen interest. In one activity, girls and boys called the community members to a two-part drama. In Part 1, a girl and boy went to the video hall, drank and smoked, and had unprotected sex. The girl became pregnant and both she and the boy dropped out of school, thereby "losing their futures." In Part 2, the same girl and boy were shown discussing their dreams of "getting an education" and what it would take to achieve that goal. Having discussed how an early pregnancy would destroy that dream, they agreed to use contraceptives. This drama evoked animated discussions with community members, who explored the problems of teenage pregnancy and the benefits of the second scenario for individuals, families, and communities.

Child leadership and messaging were central to the intervention since young people know better than adults how to communicate with and influence other young people. In Moyamba, teenagers constructed the message "5920," which meant that for 5 min of pleasure, one gets 9 months of pregnancy followed by 2 years out of school, and at the end, they have nothing because their educations have been lost. Teenagers' use of this message in small gatherings or in public meetings typically evoked laughter and excitement, yet its meaning was serious. Teenagers' delivery of such messages was supported by adults, who shared power with teenagers and increasingly saw them as having good values and making good decisions.

During the planning stages, elders commented that before the war, parents had talked with their children about puberty, sex, and pregnancy and that those discussions had benefited children and families. In order to rekindle this practice, public discussions in the barray and also the NGO trainings encouraged parent-child discussions that enabled mutual education

and support. In some cases, parents led the way by helping younger children to understand the changes in their bodies, explaining how girls become pregnant, and discussing the importance of preventing teenage pregnancy. Children also initiated discussions with parents and even became 'teachers' who helped to correct misconceptions such as the idea that male condoms should not be used because they come off and get lodged in the woman's vagina.

Role modeling served as a means of increasing collective motivation to get involved and of teaching new forms of social interaction that helped to prevent teenage pregnancy. Observational learning likely contributed to people becoming involved in the intervention. As children sang songs and performed dramas, they not only provided information but served as role models who encouraged other children to stay in school, resist the pressure from boys, and avoid early pregnancy. Adults also provided important role models by giving advice on how to avoid teenage pregnancy. The vocal support by Paramount Chiefs and the Chiefdom Speakers helped to legitimate the intervention and to motivate community members to participate in it. The support of Government workers such as health post staff likely contributed as well to people's ongoing concern and involvement.

Preliminary Results

The preliminary results included positive outcomes related to child protection, the community process, and system strengthening.

Community Ownership. High levels of community ownership were evident throughout the planning and implementation phases. Community members and chiefs regularly referred to the intervention as "ours" and stated that NGOs and the government support them but do not lead the intervention. People have repeatedly volunteered their time and energy, without any monetary or material compensation. Over time, many people beyond the task force members and peer educators have participated in or contributed to organizing the intervention activities. Thus the intervention was the collective work of the community.

Nonformal-Formal Linkage and Collaboration. The intervention process significantly improved communities' collaboration and linkage with the local health posts. Both the health workers and ordinary citizens reported that before the intervention, teenagers and other community people had rarely visited the health posts and that nurses had seldom been invited to the villages. The intervention, however, reversed this situation. Many teenagers and/or their parents visited the health posts regularly for contraceptives or advice. Also, villages frequently invited nurses and other health staff to visit in order to educate villagers about puberty, pregnancy, sexual and reproductive health, and how to prevent teenage pregnancy.

Contraception. The district medical officers fulfilled their promise to supply the contraceptives and train the health staff. The health post staff talked excitedly about how the intervention had increased demand for contraceptives, particularly implants for girls. Significant increases in demand for contraceptives occurred also in the comparison communities, possibly as a result of the president's initiative to end teenage pregnancy. Although large numbers of male condoms were distributed, the survey results showed only small increases in men's willingness to use condoms on a regular basis. Yet there were encouraging signs of change. Relative to the comparison condition, teenagers in the intervention communities reported increased intent to use condoms regularly and increased willingness to ask their partner to use a condom. These are precursors, we hope, of wider changes in behavior and social norms related to sex.

Life Skills. Teenage girls reported that because of the intervention, they said 'No' more frequently to unwanted sex. Both girls and boys said that they had learned how to discuss and negotiate with their partners in regard to sex, and also how to plan their sexual activities in light of wider, life goals. In addition, boys said openly that they had a responsibility to prevent teenage pregnancy. This responsibility taking contrasted sharply with the boys' previous behavior.

Teenage Pregnancy. Participant observations and interviews with health post staff, monitors, teenagers, and adults indicated a significant decrease in teenage pregnancies. In the intervention communities in both districts, participants reported that in an average pre-intervention school year (September-June), there were five or six teenage pregnancies. In contrast, in the 2013-2014 school year, fewer teenage pregnancies had occurred. During that period, half the communities reported no new teenage pregnancies, and the other half reported only one new teenage pregnancy. Grandmothers, who are respected community figures, assured that it is impossible to hide pregnancies in the villages. Both grandmothers and health post staff reported no increase in abortions during the intervention period. These encouraging results still need to be triangulated fully with data obtained from other methods.

Spinoffs. The community-driven intervention had numerous positive spinoff effects. Participants said that school dropouts had decreased, probably owing either to reductions in teenage pregnancy, the many conversations about "keeping one's dream" of getting an education, or a combination of the two. Also, some villages had spontaneously begun to discuss the problem of early marriage. This marked a change relative to the ethnographic phase, in which people had indicated that the problem was teenage pregnancy out of wedlock, as if teenage pregnancy in the context of marriage were unproblematic. Having learned more about the adverse effects of teenage pregnancy, villagers had begun to question the appropriateness

of any teenage pregnancy and also of early marriage. Subsequent evaluations will examine whether these changes are harbingers of wider shifts in views of girls and young women and in social norms that had prized being married and having children at an early age.

Limitations

Our ability to generalize these findings is limited since the research did not include a national, representative sample or a large number of clusters. The results presented above are best regarded as preliminary, because the endline data collection was postponed as a result of the Ebola crisis. Also, the intervention approach is not a "silver bullet," as the group discussions at its heart could, in zones of political violence, be seen as political organizing that could lead to violence. Further, humanitarian emergencies may demand immediate child protection responses rather than a slower, dialogue guided intervention. In urban environments with fluid populations, there may be low social cohesion and no discernible sense of community.

This intervention approach may also not be the method of choice for changing all harmful social norms and practices. In this research, people never identified female genital mutilation as a harm to children or a problem to be addressed. Fortunately, internally guided approaches to changing such harmful practices in countries such as Sudan and Ethiopia exist (Ahmed, Al Hebshi, & Nylund, 2009; Dagne, 2009). It is essential to complement the community-driven approach that I have presented with social transformational work on ending harmful practices such as FGM. How to mix these approaches effectively is an important question for future research.

Implications for Practice and Policy

This action research has valuable implications for practitioners. Its slow, patient facilitation stimulated community driven action on behalf of children and achieved much higher levels of community ownership than are usually achieved through expert-driven approaches. The preliminary findings suggest that the community-driven approach is quite effective in promoting positive outcomes for children. The research also confirms the potency and utility of bottom-up approaches to strengthen system strengthening. Through community-driven action, communities developed greater willingness to engage with and learn from formal health workers and to use and the formal services to address teenage pregnancy. Sustainability will likely be high since the intervention is founded on voluntary, community led action and people, children included, are highly motivated to address teenage pregnancy.

To some extent, this approach fits with and is a logical extension of current practice. After all, skilled child protection practitioners have long prioritized community mobilization and encouraged communities to actively support and engage in child protection. However, important differences also exist. First, this approach involved much deeper, nonjudgmental listening to different people than is usually done. Second, this community-driven approach requires that child protection actors shift their roles from that of experts to that of facilitators who are also co-learners. When external child protection workers enter the community as experts, community understandings and practices tend to be marginalized, and communities lapse into a familiar role as beneficiaries. The facilitative approach highlighted above viewed communities fundamentally as actors, albeit imperfect, who given the appropriate space, opportunities, and encouragement for reflection motivated each other to take collective action that reduces harms to children. Third, this approach entailed a slower pace that enabled high levels of participation in the decision making and intervention work, the inclusion of marginalized people, and a collective process that included different subgroups. NGO workers frequently ask how it is possible to take a slower approach when donors and their agency headquarters demand rapid results. Useful strategies in this regard include obtaining diverse funding streams, including multi-year ones, educating the donors using data such as those presented above, or negotiating the use of a small percentage of funds to pilot test a community-driven approach in selected areas. Also useful is taking a self-critical stance that heightens awareness of the harms caused by top-down and excessively rapid approaches.

The meaningful participation of children in community decision making and action warrants special attention. Too often, NGOs enable child participation only in children's domains, such as child clubs or youth groups. Although valuable, these efforts do not fully enable children's voice in community-decision making or fully engage the more vulnerable children who typically do not take part in such children's venues. With good facilitation, however, communities can reconfigure their planning processes to be more inclusive, reach the more vulnerable children, and make children's agency a strength that helps to guide informed community action on behalf of children.

With respect to policy, this research highlights the importance of supporting community owned and driven efforts as part of strengthening child protection systems. Fortunately, the government of Sierra Leone, with the aid of UNICEF and other CP Com members, is taking these findings to heart. The ethnographic findings, together with those of other research (e.g., Thompstone, 2010), have formed the basis for a new child and family welfare policy that limits the "add-a-structure" orientation to system strengthening and aims to support family and community action on behalf of children. Also, UNICEF and a technical working group of the agencies that have been most directly involved in this research is leading the plan to scale up the community driven approach described above. Initially, the work will be extended throughout Moyamba and Bombali districts, and, pending its success, it will be rolled out on other districts countrywide. The lateral spread of ideas across communities and chiefdoms will be important in this process, as will a spirit of ongoing learning together with local people.

Conclusion

A frequently expressed frustration by child protection practitioners is that despite their vigorous efforts to help establish CWCs, train CWC members, and teach local people how to protect children in accord with national laws and international standards, local people tend not to use what they have been taught and rely on nonformal practices that are poorly aligned with the formal system. Clearly, the dominant approach to strengthening CBCPMs is not working as intended.

What is needed is a fresh approach to strengthening CBCPMs and enabling their collaboration and alignment with formal aspects of the child protection system. The action research in Sierra Leone represents such a fresh approach of community-driven action that is animated by community ownership, builds on community assets and resources, and features the role of children and young people as change agents. When communities themselves drive the process of linking with formal stakeholders, local people develop a new sense of ownership for formal services and a strong sense of partnership with formal stakeholders. This bottom-up approach to system strengthening supports the alignment of nonformal and formal elements and enables the uptake and use of formal services. Such bottom-up approaches not only complement the more widely used top-down approaches but also help to unlock the prodigious creative and practical capacities of communities. Collectively, we will do a better job of protecting children if we step out of our expert role and facilitate the community-driven action and the related social transformation that supports vulnerable children.

Acknowledgements

The author wishes to thank the Sierra Leonean research team, especially David Lamin, Dora King, and Marie Manyeh; Sarah Lilley, who coordinated the global Reference Group; and Lindsay Stark, who led the work on outcomes measurement.

References

African Child Policy Forum, African Network for the Prevention and Protection against Child Abuse and Neglect, Environnement et Developpement du Tiers-monde, International Social Service, Mouvement Africain des Enfants et Jeunes Travailleurs, Plan International, Regional Inter-agency Task Team on Children and AIDS, Regional Psychosocial Support Initiative, Save the Children, SOS Children's Villages International, Terre des hommes, UNICEF, & World Vision International. (2013). Strengthening child protection systems in Sub-Saharan Africa: A call to action. Dakar, Sengal: Author. Ahmed, S., Al Hebshi, S., & Nylund, B. V. (2009). Sudan: An in-depth analysis of the social dynamics of abandonment of FGM/C (Innocenti Centre Working Paper 2009-08). Florence, Italy: UNICEF.

Behnam, N. (2008). Agencies, communities, and children: A report of the Interagency Learning Initiative: Engaging communities for children's well-being. Washington, DC: USAID Displaced Children and Orphans Fund. Available online at. http://www.usaid.gov/our_work/humanitarian_assistance/ the.funds/pubs/comaction.html

Behnam, N. (2011). Awkward engagement: Friction, translation, and human rights education in post-conflict Sierra Leone (Unpublished doctoral dissertation).

Philadelphia PA: University of Pennsylvania. Boothby, N., Strang, A., & Wessells, M. (Eds.). (2006). A world turned upside down: Social ecological approaches to children in war zones. Bloomfield, CT: Kumarian Press.

Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press. Chambers, R. (1994). The origins and practice of participatory rural appraisal. World Development, 22,953-969. Child Protection Working Group. (2012). Minimum standards for child protection in humanitarian action. Geneva, Switzerland: Author. Coinco, E. (2010). A glimpse into the world of teenage pregnancy in Sierra Leone. Freetown, Sierra Leone: UNICEF.

Collier, P., Elliott, V., Hegre, H., Hoeffler, A., Reynal-Querol, M., & Sambanis, N. (2003). Breaking the conflict trap. Washington, DC: World Bank.

Dagne, H. G. (2009). Ethiopia: A comparative analysis of the social dynamics of the abandonment of harmful practices in four locations (Innocenti Centre Working

Paper2009-07). Florence, Italy: UNICEF. Davis, R., McCaffery, J., & Conticini, A. (2012). Strengthening child protection systems in sub-Saharan Africa: A working paper. Dakar, Senegal: Inter-agency

Group on Child Protection Systems in sub-Saharan Africa. Dawes, A., & Donald, D. (2000). Improving children's chances. In D. Donald, A. Dawes, &J. Louw (Eds.), Addressing childhood adversity (pp. 1-25). Cape Town,

South Africa: David Philip. Denov, M. (2010). Sierra Leone's Revolutionary United Front. Cambridge, UK: Cambridge University Press.

Donahue, J., & Mwewa, L. (2006). Community action and the test of time: Learning from community experiences and perceptions. Case studies of mobilization

and capacity building to benefit vulnerable children in Malawi and Zambia. Washington, DC: USAID. Fernando, C., & Ferrari, M. (Eds.) (2013). Handbook of Resilience in children of war. New York: Springer.

Foster, G. (2004). Study of the response by faith-based organizations to orphans and vulnerable children. New York, NY: World Conference of Religions for Peace and UNICEF.

Freire, P. (1968). Pedagogy of the oppressed. New York, NY: Herder & Herder.

Graham, A., Powell, M., Taylor, N., Anderson, D., & Fitzgerald, R. (2013). Ethical research involving children. Florence Italy: UNICEFOffice of Research-Innocenti. Krueger, A., Thompstone, G., & Crispin, V. (2013). Learning from child protection systems mapping and analysis in West Africa: Research and policy

implications. Global Policy, 5, 47-55. McKay, S., & Mazurana, D. (2004). Where are the girls? Montreal, Canada: International Centre for Human Rights and Democratic Development. Panter-Brick, C., & Leckman, J. (2013). Editorial commentary: Resilience in child development - Interconnnected pathways to wellbeing. Journal of Child

Psychology and Psychiatry, 54, 333-336. Stark, L., Macfarlane, M., King, D., Lamin, D., Lilley, S., & Wessells, M. (2014). A community-driven approach to reducing teenage pregnancy in Sierra Leone:

Midline evaluation brief. London, England: Save the Children. Thompstone, G. (2010). Mapping and analysis of the child protection system in Sierra Leone. Hong Kong, China: Child Frontiers. UNDP. (2013). Poverty and development indicators, Sierra Leone. Retrieved from hdrstats.undp.org. Ungar, M. (2008). Resilience across cultures. British Journal ofSocial Work, 38, 218-235.

UNICEF. (2013). Female genital mutilation/cutting: A statistical overview and exploration of the dynamics of change. New York, NY: Author.

UNICEF, UNHCR, Save the Children, & World Vision. (2013). A better way to protect ALL children: The theory and practice of child protection systems, conference

report. New Delhi, India: Author. Wessells, M. G. (2006). Child soldiers: From violence to protection. Cambridge, MA: Harvard University Press.

Wessells, M. G. (2009a). What are we learning about protecting children in the community? An inter-agency review of the evidence on community-based child protection mechanisms in humanitarian and development settings. London, England: Save the Children.

Wessells, M. (2009b). Do No Harm: Toward contextually appropriate psychosocial support in international emergencies. AmericanPsychologist, 64,842-854.

Wessells, M. G. (2011). An ethnographic study of community-based child protection mechanisms and their linkage with the national child protection system of Sierra Leone. New York, NY: Columbia Group for Children in Adversity.

Wessells, M., Lamin, D., King, D., Kostelny, K., Stark, L., & Lilley, S. (2012). The disconnect between community-based child protection mechanisms and the formal child protection system in rural Sierra Leone: Challenges to building an effective national child protection system. Vulnerable Children and Youth Studies, 7(31), 211-227.

Wessells, M., Kostelny, K., & Ondoro, K. (2014). A grounded view of community-based child protection mechanisms and their linkage with the wider child protection system in three urban and rural areas of Kenya: Summary and integrated analysis. London, England: Save the Children.

Wulczyn, F., Daro, D., Fluke, J., Feldman, S., Glodek, C., & Lifanda, K. (2010). Adapting a systems approach to child protection: Key concepts and considerations. New York, NY: UNICEF.

Zack-Williams, T. (2013). When children become killers: Child soldiers in the civil war in Sierra Leone. In C. Fernando, & M. Ferrari (Eds.), Handbook of resilience in children of war (pp. 83-94). New York, NY: Springer.