Community-Driven Development in Conflict-Affected Contexts : Revisiting Concepts , Functions and Fundamentals

Community-Driven Development (CDD) is a popular aid delivery strategy in conflict-affected contexts. While the strategy remains appealing, the growing body of evidence suggests that CDD does not systematically deliver on all the desired outcomes. This may potentially be explained by the lack of clarity around the objectives and theoretical underpinnings of CDD interventions. This paper proposes ways to clarify the objectives, outcomes, theories of change and core processes of the CDD strategy in an effort to improve the design and evaluation of CDD interventions. We suggest schemas for prioritizing the function and outcomes of a given intervention, provide examples of reduced form theories of change and identify a set of ‘core processes.’ We hope these suggestions will assist practitioners in making the theoretical motivations, assumptions and trade-offs of their design choices that much more explicit and in so doing, improve our ability to deliver better interventions to conflict-affected populations. This paper forms part of a wider conceptual project supported by UK Department for International Development (DFID)’s Research and Evidence Division.


Introduction
Community-Driven Development (CDD) is a widely used aid delivery strategy that emphasizes community control over planning decisions and investment resources.It has seen increasing deployment for stabilization and development purposes in fragile and conflict-affected settings such as Afghanistan, Democratic Republic of Congo, Myanmar and Somalia.Recent rigorous impact evaluations of CDD programming across multiple locations have yielded limited or null findings.This has led to calls for a programmatic overhaul of CDD approaches, including more realistic goal-setting, greater specificity over intended outcomes and the explicit articulation of more robust change pathways linking CDD activities to desired outcomes in theoretically informed and credible ways (King 2013).This paper is part of a wider conceptual project 1 that seeks to respond to these, and other critical reflections, by interrogating core assumptions implicit in how CDD is generally practiced and explained, with a view to synthesising and advancing the intellectual labour necessary for the emergence of more transparent and clearly articulated CDD programming approaches.
Community-Driven Development (CDD) is a popular aid delivery strategy in conflictaffected contexts.While the strategy remains appealing, the growing body of evidence suggests that CDD does not systematically deliver on all the desired outcomes.This may potentially be explained by the lack of clarity around the objectives and theoretical underpinnings of CDD interventions.This paper proposes ways to clarify the objectives, outcomes, theories of change and core processes of the CDD strategy in an effort to improve the design and evaluation of CDD interventions.We suggest schemas for prioritizing the function and outcomes of a given intervention, provide examples of reduced form theories of change and identify a set of 'core processes.'We hope these suggestions will assist practitioners in making the theoretical motivations, assumptions and trade-offs of their design choices that much more explicit and in so doing, improve our ability to deliver better interventions to conflict-affected populations.This paper forms part of a wider conceptual project supported by UK Department for International Development (DFID)'s Research and Evidence Division.
This paper attempts to provide a framework for clarifying and proposing alternative ways of conceptualising CDD.In order to arrive at revised approaches to CDD, it is necessary to articulate the theoretical bases of these revisions and how they may lead to the desired changes.This in turn requires that the objectives of the intervention and the changes that are desired are conceptually clear.Taken together, these specifications along with information on context inform the range of theoretically motivated, contextually adapted design options available to practitioners and policymakers.This framework emerged primarily from reviews of theoretical, policy and evaluation literature published through peer-reviewed journals, working paper repositories and organisational websites.Individual and small group exchanges with a non-representative sample of practitioner and donor personnel in Myanmar and Somalia and feedback from a select group of academics in Economics and Political Science through reviews, conference presentations and discussions also contributed to the development of this framework.
The paper is organized as follows: an initial section introduces the CDD approach and the evidence associated with its effectiveness in conflict-affected contexts.This is followed by a focus on CDD objectives, with a clarifying schema proposed.The paper then proposes two preliminary, illustrative and reduced form theories of change around specific governance and social cohesion outcomes.The subsequent section highlights the core processes that constitute a CDD approach.The paper concludes with a discussion on the implications of the core processes, theory and context for the practical design and implementation of CDD interventions.

Community-Driven Development: Concept and Evidence
Community-Driven Development is a popular aid delivery strategy that emphasizes community control over planning decisions and investment resources.CDD operates on the principles of community and local government empowerment, decentralisation, accountability (downward and horizontal), transparency, learning by doing or enhanced local capacity (Davis 2004).CDD interventions ' emphasize giving communities and locally elected bodies the power, information and skills to determine the best use of development resources' (Wong and Guggenheim 2005: 254).
The underlying premise is that local communities are best placed to identify their development needs and the corresponding solutions to them (Cliffe, Guggenheim and Kostner 2003).Also, by giving ' control of decisions and resources to community groups' (Dongier et al. 2003: 3), CDD projects will better meet communities' needs and enhance ownership.It is also hypothesised that engaging in CDD's participatory processes will empower communities, increase capacity for local development and governance and improve social cohesion (Chase and Woolcock 2005;OED 2005).When CDD is implemented in conflict-affected contexts (often referred to as CDR -communitydriven reconstruction), there is a recognition that ' countries face an even stronger imperative to rebuild social capital, empower and provide voice to communities and generally rebuild the social fabric torn apart by violent conflict' (Cliffe, Guggenheim and Kostner 2003: iv).
Despite the democratic underpinnings and the captivating promise of improved governance, welfare and social cohesion or capital that characterize CDD or CDR interventions, evidence of their effectiveness is mixed at best.In a review of seventeen of its social fund and CDD programmes for which rigorous evaluations were conducted, the World Bank reported generally positive effects of CDD programming on poverty targeting and reduction and access to services but little, if any, evidence of impact on governance, social capital or conflict (Wong 2012).Results from rigorous impact evaluations vary across outcomes and contexts.
Positive short-term welfare outcomes were observed in Aceh (Barron, Humphreys, Paler and Weinstein 2009) and Sierra Leone (Casey, Glennerster and Miguel 2011); positive social cohesion outcomes in Liberia (Fearon, Humphreys andWeinstein 2008, 2009), positive impacts on some welfare outcomes and mixed impacts on governance outcomes in Afghanistan (Beath, Christia and Enikolopov 2013) and null results (i.e.no measured effect) across all outcomes in the Democratic Republic of the Congo (Humphreys, Sanchez de la Sierra and van der Windt 2012).An extensive critical review of the family of participatory strategies to which CDD belongs reveals more of the same mixed results with the degree of effectiveness dependent on a number of factors including support and responsiveness of state government and national and local historical, political, geographic and social contexts (Mansuri and Rao 2013).The evidence base also includes a number of studies on the impact of design components of the CDD strategy: monitoring and audit strategies (Olken 2007), election and project selection strategies (Beath, Christia and Enikolopov 2009) and women's participation (Beath, Christia and Enikolopov 2012).The general take-away seems to be that CDD sometimes yields some of its purported outcomes, under some circumstances.

Towards More Specific Objectives
One of the central criticisms of CDD interventions, as they are generally designed and implemented, is that there is a lack of clarity and coherence in the specification of programme objectives.Proponents of CDD have argued that the approach can have a positive impact on multiple aspects of social life, yielding improved welfare, governance and social cohesion outcomes, almost always in combination.This has led to the critical observation that CDD designs have 'been plagued by a panacea-type approach to goals' (King 2013: 3), wherein 'grand, interconnected' objectives are set (King 2013:31), often representing the ' confluence of reasons' that motivated the selection of CDD as an intervention type in the first place (King 2013: 48).This section of the paper examines these criticisms in more detail and proposes several ways of defining clearer, more precise and relevant objectives, as an essential step in designing more theoretically informed, robust CDD interventions that are more adaptable to context and more amenable to rigorous evaluation.
There is an apparent contradiction between a call for greater focus and clarity in defining the objectives of an approach and a desire for that same approach to be demand-(community-) driven.The decision to fund a CDD intervention to achieve specific objectives is not made by the intended participants.It is, therefore, a form of supply-driven demand-driven programming.These supplyside choices can constrain or promote differing degrees of demand-side choice making it all the more important for those choosing to fund and design CDD interventions to be clear about their motivation and objectives for doing so.
The lack of precision around CDD objectives gives rise to several concerns.Foremost among these is that in the absence of clearly formulated objectives, it is impossible to construct a plausible argument about how CDD interventions, and specific elements thereof, contribute to desired outcomes.Clarity of objective is a logical precursor to the development of a theory of change.Absent a theoretically grounded and contextually relevant set of hypothesised causal pathways, it is very difficult to make sense of results (positive, negative or null), to modify programme designs to potentially improve outcomes or to accurately measure the effect of any such changes.All of these problems trace back to the lack of clarity about CDD objectives.
There are other ways in which imprecise objectives can confuse or complicate an understanding of what is occurring within and as a result of a CDD intervention.By treating multiple outcomes as equivalent products of a CDD approach and weighting them equally within a logical framework, potentially complex causal relationships between the outcomes may be obscured.For example, it is possible that progress towards representation might unintentionally undermine social cohesion through stimulating debate, revealing and intensifying intergroup tensions, which were either latent or non-existent under previous decision-making arrangements.Another problem that arises from bundling the common CDD outcomes is that these outcomes might require different time frames for coming to fruition or might require significantly different numbers of repetitions to generate a perceptible effect.This in turn has implications for the timing and structure of measurement and evaluation of the intervention.Similarly, it is difficult to understand whether or not sequencing of certain activities is important to achieving objectives, without clarifying objectives and separating out different outcomes, their associated causal pathways and hypothesised timeframes.
CDD interventions have also been criticised for over-ambitious targets, both in relation to the context of intervention and to the scale of programming inputs, whether in invested resources or time devoted to specific stages of the CDD intervention.While these concerns do not fully derive from the imprecision of CDD objectives and are at least partially rooted in limited understanding of the context, a clarification and narrowing of the focus of the approach would encourage a closer analysis of the degree to which projected changes are realistic relative to context and the scale of investment that would be most appropriate.
There are three categorical schemas that the authors have found useful in thinking about CDD objectives and outcomes.The first is an attempt to disaggregate the standard outcomes of CDD -welfare, governance and social cohesion.The second introduces the distinction between 'product' and 'process' and focuses on what the CDD intervention primarily seeks to affect.The third identifies whether the intervention seeks to improve administrative efficiency, provide a temporary substitute for other processes (historical, anticipated or both) or to transform aspects of the context in which it occurs.

Welfare, governance or social cohesion
Distinguishing between welfare, governance and social cohesion outcomes provides a useful starting point for a discussion of how to introduce greater precision into CDD objective setting.As previously stated, it is important to disaggregate these outcomes to determine whether and how they might work with or against each other.It also provides an opportunity to specify a primary objective and to weight design choices accordingly.This does not preclude the possibility that CDD could have multiple effects across different outcomes, but it forces the prioritisation of outcomes in such a way as to encourage greater precision in the development of change hypotheses and corresponding designs.For example, there is a non-trivial difference between attempting to improve local accountability practices, by instrumentally using resource allocation to stimulate local decision-making processes on the one hand, and seeking to maximise welfare outcomes, while instrumentally using local consultation to more appropriately align investments with the revealed preferences of programme participants on the other.This difference could have implications for a range of design choices, such as the size of resource allocations to communities, the amount of time and resources spent on community 'sensitisation,' the types and extent of participant engagement required or the extent to which efficient use of funds is important to the intervention.
However, within these broad categories of 'welfare,' 'governance' and 'social cohesion' there is a similar need to further disaggregate outcomes in order to achieve greater precision of objective and associated theories of change.There is a difference between seeking to stimulate collective action and seeking to increase the legitimacy of local government institutions.Both might be termed governance outcomes, but the former could occur in isolation from existing local government structures, particularly if they are weak, non-functional or parasitic, and might entail engaging other forms of social organisation, such as traditional kinship structures or churches, seeking to build greater self-reliance among local communities.The latter firmly embeds the CDD intervention within or in close association with existing government structures.The former might emphasise the development of capacity to manage and implement programming by citizens, the latter would embed this function in the civil service.Despite the varying and sometimes opposing design implications, attempting to address collective action problems while claiming to build state legitimacy is not an atypical goal for CDD interventions.This illustration underscores the need to specify even more clearly what is meant by improving governance, welfare, or social cohesion as a precursor to developing and testing the causal hypotheses that link intended interventions with these more clearly defined outcomes.

Product vs. process
A second categorical distinction can be made between CDD as (primarily) a means to deliver products to people and CDD as a means of changing processes.This is not to suggest that only the former involves the delivery of products and only the latter potentially influences social processes but rather that the purpose or objective in using a CDD approach is radically different in the two cases.Clearly all CDD interventions involve the delivery of resources to people.Similarly, it can be argued that the act of requiring that decision-making power be given to local communities may likely have some effect on various social processes in and across those communities.However, applying this distinction at the level of objectives yields a potentially different focus and associated definitions of success, with consequent implications for design and measurement.
Whether the product is social infrastructure, a means to generate income, vouchers for poorer people to access services or some other output, CDD intended primarily as a mechanism for the delivery of a product to people focuses on the relative efficacy of the delivery mechanism and those aspects of the CDD process that support that delivery.The participation of community members in various stages of the CDD intervention is important to the extent that this yields a more appropriate delivery of the most appropriate product(s) to the most appropriate people.This engagement would therefore be focused around the revelation and articulation of preferences, the resolution of different views on product choice and the smooth delivery of products.The decision to invest in a CDD 'product delivery' approach could hinge on comparison with other delivery channels (such as capital investment projects or cash transfers) and the potential added value of this type of participatory delivery mechanism.Similarly, measurement strategies to determine the effectiveness of such an intervention could make use of other aid delivery mechanisms as an appropriate counterfactual.
Conversely, CDD as a means to affect processes shifts attention to how the intervention influences attitudes, behaviours and norms.In this approach, the role of product delivery is primarily to provide a means for people to experience the preference revealing, decision-making and broader collective action opportunities that CDD offers.The objective is to influence a specific process or set of processes that can be simulated within the CDD intervention such that experience with CDD can be plausibly expected to have an effect on similar processes in the broader social context.An example would be the expectation that the encouragement and involvement of women in CDD-related decision-making processes through quotas on leadership committees or earmarked funding for women's-only projects would lead to a greater involvement of women in wider household or communal decision-making around resources.
If the objective of a given CDD intervention is defined in terms of affecting processes, those elements of the intervention that potentially influence behaviour, attitudes and norms become the crucial focus of design, implementation and measurement strategies.Sound contextual knowledge of existing socio-political and social psychological processes becomes significantly more important than it would be for a 'product delivery' approach.Likewise, the timeframe of a 'process change' intervention would be arguably longer.The relevant framework for CDD as a strategy to change processes could entail comparison with an entirely different set of alternative interventions, such as public information campaigns or a range of transparency and accountability initiatives.Similarly, such alternative approaches could yield more appropriate counterfactuals to determine the relative effectiveness and cost effectiveness of CDD as a means to affect processes.

Efficiency, substitution, transformation
A third way that one can attempt to bring greater precision to the objectives of a CDD intervention is to be as specific as possible about the extent to which the intervention seeks to improve efficiency, or provide a temporary substitute or transform norms and institutions.These three functions are often bundled together, with a transformative aspiration typically implied in the framing of the intervention (sometimes for reasons of political or other contextual sensitivity).Each provides a distinct flavour to a CDD approach, however, and separating them out provides another lens through which to clarify and prioritise objectives.
The efficiency function entails the deployment of a CDD approach to improve how an already existing process or delivery mechanism works.For example, one might consider CDD to be adept at revealing people's preferences regarding community investments and deploy this as a programming strategy to improve the targeting of public funds.Conversely, the substitutive function concerns the use of CDD to address system failures or the absence of functioning systems through temporary measures.This function is often observed in fragile or conflict-affected settings in which an absence of central government investment in infrastructure in certain areas of the country or the lack of de facto decentralisation can lead to the adoption of CDD as a means of programming social investments at scale.This approach has been used in Eastern Congo and Somalia and may be present in planning for future programming in South Sudan.A government may adopt, in collaboration with the international aid and development community, a CDD approach as a temporary stopgap given prior absence of investment in areas due to war, political neglect, economic policy failures or other reasons.
A transformative function is focused beyond the amelioration of existing systems or addressing their failures.It intentionally seeks to transform some aspect or aspects of social life.This function is particularly prevalent in combination with governance outcomes: the notion of improvements in governance practices generally means the promotion of specific democratic norms of governance.This is unsurprising given that the basic definition of CDD is to give people greater choice over how resources are allocated and the assumption that this degree of popular engagement with resource allocation decisions was either non-existent or atypical in many of the contexts in which CDD is implemented.Where CDD is implemented in a conflict-affected setting, there is often a transformative objective with regards to social cohesion, particularly in explicit 'peace-building' or ' conflict-mitigation' programming.Here the notion is often that relations between previously warring groups can be transformed by non-violent participatory decision-making on resource issues of common interest.
Being more explicit and honest about the transformative vs. substitutive vs. efficiency aspirations of a CDD intervention helps to determine the relative weighting of different components of programme design and the most appropriate measurement and evaluation strategies.There is also an important injection of realism encouraged by being explicit about transformative aspirations and tracing these back into a theory of change.As a result, this conceptual lens is particularly useful in driving CDD design discussions towards more humble and realistic goals.

Using clarified objectives
These three schemas represent a small sample of possible ways to clarify the motivation and objective of CDD interventions -a necessary step in improving theory, design and measurement.The schemas are not intended to be mutually exclusive; neither do the authors argue for one schema being prioritised over another.In fact, it is possible to envision these schemas as a set of layered, if not sequential, decision-points for the donors, policymakers and implementers involved in the elaboration of a given CDD intervention.For instance, one may first choose the function of the CDD intervention -transformative -then choose the substance of the intervention -processthen choose the theme or outcome category to be prioritized -governance.The subsequent decision would be around the type of governance improvement or change that is sought and the corresponding outcomes to be measured e.g.equity or accountability.Increasingly, more specific knowledge of the context and theory become necessary as one progresses through this chain of decisionmaking points.It is at this point that practitioners can attempt to articulate theories of change, which would then influence choices over the types of activities, inputs, and strategies that could plausibly lead to the desired outcome or effect.Likewise, as a result of the narrowing and specification exercises described above more appropriate measurement and evaluation strategies can be identified and developed.

Towards More Explicit Theories of Change
Another criticism of the CDD approach is that it lacks explicitly articulated theories of change (King 2013).This section of the paper includes preliminary attempts to articulate potential theories of change associated with specific outcomes.A theory of change can be thought of as ' a comprehensive picture of the early and then intermediate changes that are needed to reach a long-term goal' (Anderson 2005:12).
Theories of change for the CDD approach have been previously articulated: a conceptual framework (World Bank 2003), a generalised theory of change (King 2013: 14) and a theory of change for CDD in conflictaffected and fragile contexts (Barron 2010).These theories of change have been helpful in understanding the general pathways through which the overall approach, with all of its embedded processes and tensions, may lead to the achievement of the main outcomes: improved welfare, governance and social cohesion.However, to the extent that the CDD approach is complex and requires deliberate clarification and prioritisation, so do the theories of change that must underpin it.The intention is to develop theories of change from which practical implications for design can be drawn.Theories of change at lower levels of abstraction with arguably more specificity would clarify hypothesised causal mechanisms and processes that could, in theory, render CDD interventions more effective and more amenable to rigorous evaluation.Theories of change affiliated with a single, more precisely defined outcome may further help policymakers, practitioners and evaluators to think more critically about intervention design choices and measurement strategies.
For the purposes of illustration and to ground the theories of change in the schema discussed above, this section of the paper is based on a hypothetical CDD intervention that assumes a transformative function, a focus on processes and with governance as the prioritised outcome in the first theory of change and social cohesion in the second.The decision to focus on governance and social cohesion outcomes is motivated by the limited evidence on CDD's impact on these outcomes.While there is evidence of relatively positive effects of CDD on poverty targeting and reduction and access to services, there is little and mixed evidence of impact on governance, social capital or conflict (Wong 2012).For governance, the theory of change focuses on attitude change.For social cohesion, the theory of change focuses on collective action.These reduced outcomes have also been selected because they seem to be embedded in the goals of many CDD interventions, particularly those implemented in fragile and conflict-affected contexts.In addition to the impetus to address the urgent welfare needs of affected populations, CDD interventions are also typically motivated by implicit assumptions of mistrust and strained relations among population subgroups and between the state and the population.CDD in conflict-affected areas often seek to not only improve intergroup relations and attitudes towards democratic forms of decision-making as ends in themselves, but also as a means to minimise the incidence of future conflict.
The transformative function of this hypothetical CDD intervention demands that the following assumption be made: the desired changes in governance and social cohesion outcomes are expected to be transportable and observable beyond the substantive, logistical and temporal confines of the intervention itself.An intervention that elicits only short-term shifts such that observed attitudes and behaviours match the prescribed norms and mandatory requirements of the intervention during the implementation period would fall admittedly short of its transformational aspirations.A transformational objective requires a theory of change that articulates the processes and conditions that would be necessary to stimulate change that persists over time.Both theories of change are articulated with this requirement in mind.
A theory around attitudes CDD interventions often attempt to stimulate attitudinal change and norm transformation.Given the democratic principles on which the CDD approach is based, the desire to shift attitudes and behaviours such that they reflect these principles is undeniably a part of the objective if improvements in governance are prioritised.CDD interventions that aim to improve governance typically have a 'sensitisation' or 'social preparation' period during which implementers provide information to potential participants and stakeholders about the scope and objective of the intervention and the democratic principles and procedures that shape the intervention.Assuming that the provision of information is critical and that the hypothetical CDD intervention includes a truly participatory decision-making process, the theory of change articulated in Figure 1 below describes a simplified potential pathway of change through which attitudinal change may occur.
The theory of change states that if information about norms is provided and this information is supported or corroborated by a respected figure of influence or authority and the recipient understands this information and engages directly or indirectly in processes that are based on these norms and he or she believes that these processes are likely to occur in the future, and the recipient derives some non-trivial benefit from the process and believes this benefit is at least in part due to the norms inherent in the process, then the recipient may internalize the norms, which in turn may shape his or her attitudes.This theory of change assumes that if norm transformation is the objective of the intervention, then these norms must be explicitly and clearly articulated to the intended participants.This information must be presented in clear and accessible ways that go beyond translation of words across languages to focus on the translation of sometimes abstract concepts and ideas in ways that are linguistically and culturally resonant.The theory makes no assumptions about who provides this information but does require that the information be corroborated by an influential, legitimate source.Corroboration precedes understanding only because it may either facilitate or reinforce understanding of the information while increasing its credibility.
The behavioural block in the theory of change -the participatory experience -is perhaps the most aspirational component.The basic idea is that the participatory experience of the CDD intervention provides individuals and groups with the opportunity to not just hear about the norms but to 'see' them in action i.e. programme participants actually practice transparency, accountability, equity etc. during the programme.An important assumption is that the CDD experience actually adheres to (or at least is perceived to adhere to) and reiterates the norms and principles that it claims to be important.If this is true and the participant derives some value from the experience then the participant may attribute value to the norms that govern that experience.
For the participant to derive value from the experience at least one of the following conditions must be true; the participatory experience must: (i) be considered at least as meaningful or effective as other decisionmaking and collaborative strategies (ii) have net positive returns (i.e.benefits outweigh the costs); (iiia) be different / distinct from other popular decision-making processes (i.e. it holds up a mirror and the contrast is striking); (iiib) be similar to other popular decision-making processes (i.e. it resonates and is easily assimilated) (iv) be perceived as transferrable to other spheres of engagement that are important for the individual or his or her group.The positive benefit derived from the experience could be material such as greater access to healthcare; practical such as improved farming skills; social such as being recognised as an agent of change or psychological such as a sense of belonging or pride from having worked constructively with others towards a common goal.Finally, the individual must be able to make a connection between the benefits he or she enjoys and the value that is attributed to the participatory experience and, by extension, the norms that govern it.In other words, the participant must think that increased access to healthcare results from the democratic elections that determined aid resources would be used to improve healthcare or from the new budgeting and transparency practice adopted by the personnel managing the facilities.The participant must also believe there are opportunities to benefit (materially, practically, socially or psychologically) from future processes that are similar but not connected to the programme.Some steps in the theory of change may be optional.For instance, that corroboration of the information is important is a testable hypothesis.Likewise, individuals who do not participate directly in the participatory experience may adjust their attitudes in the direction of the promoted norms (hence the shaded box in the figure above).For instance, individuals may change their attitudes towards norms because of personal experience (the most direct pathway) or because, without having participated in the intervention, they are influenced by an authority figure, opinion leader or members of a salient group to which they belong (indirect pathway).This influence matters if the individuals face social sanctions or the threat of some loss by not adopting these attitudes.Even perceived (as opposed to real) threat may be sufficient to influence individuals to publicly display attitudes that they do not privately accept or endorse because they think everyone else does.This means that the theory of change should be further elaborated to include multiple pathways with different causal mechanisms, if these are plausibly important for achieving the specified outcome.These pathways need not be linear as depicted in the diagram.

A theory of change around collective action
Increased social cohesion is among the most commonly desired outcomes of CDD interventions in conflict-affected contexts.Social cohesion is defined in several ways including as 'the glue that binds society together' (Colletta, Lim and Kelles-Viitanen 2001) and as 'local patterns of cooperation' (Fearon, Humphreys and Weinstein 2009).More generally, social cohesion is also described as the ' attitudes and behaviours that reflect community members' tendency to cooperate within and across groups' (King, Samii and Snilstveit 2010).Increasing social capital in conflictaffected contexts is particularly important due to assumptions that the social fabric of these areas is typically torn by violent conflict (Cliffe, Guggenheim and Kostner 2003).
The implicit expectation is that engaging in CDD-type interventions would improve social relations and reduce the likelihood of violent conflict in the future.The body of evidence around the effectiveness of the CDD approach as a means of improving social cohesion is limited.As previously mentioned, within the class of rigorous impact evaluations of CDD interventions in conflict-affected contexts, only one study has found positive impacts on social cohesion.An assessment of a CDR intervention in Northern Liberia found that communities that had participated in the CDR intervention were significantly more likely to contribute in a subsequent public goods game, especially when they played the game as a mixed gender group (Fearon, Humphreys and Weinstein 2009).This positive effect could be plausibly attributed to an increase in mobilisation capabilities (gleaned from the CDR experience) among mixed gender groups (Fearon, Humphreys and Weinstein forthcoming).The impact evaluation of the Community-Based Assistance for Reintegration of Conflict Victims programme in Aceh BRA-KDP programme in Aceh revealed a slight decrease in social cohesion as conflict victims that participated in the programme were less accepting of excombatants than conflict victims that had not participated in the programme.The authors postulate that this may be the result of a perceived lack of fairness in the disbursement of cash transfers to ex-combatants (Barron, Humphreys, Paler and Weinstein 2009).
As illustrated by the definitions and examples above, changes in social cohesion are often assessed using measures of collective action.The assumption is that communities or groups with higher levels of social capital or cohesion are better able to overcome collective action problems.Under some conditions, relationships based on trust and social networks increase the likelihood of collective action (Narayan and Pritchett 1999).Although the exact directionality of the causal relationship between social cohesion and development remains debatable, social cohesion or capital (often used interchangeably) is considered a critical element in the development of successful institutions and 'good' governance (Putnam and Nanetti 1994), institutional quality and economic growth (Easterly, Ritzen and Woolcock 2006) and overall development (Ritzen, Easterly and Woolcock 2000).However, for the purposes of this paper, collective action will be considered a social cohesion outcome.
Using the schema and language outlined in the previous section to develop a theory of change, this hypothetical CDD intervention again assumes a transformative function, a focus on process and social cohesion (as demonstrated through collective action) as the prioritized outcome.As with attitudinal change, by assigning a transformative function to the intervention, this theory attempts to capture a potential pathway for change that persists beyond the parameters of the programme.The theory of change is depicted in Figure 2 above.This theory of change states: If groups within a community (or across communities) come together (convene) and are able to identify and engage around a common need then they will identify and acquire critical information and resources that will enable them to construct credible commitment mechanisms that facilitate and enforce the action required to address the need which then allows for some benefit to be derived and subsequently for some value to be attributed to the process.
A number of assumptions need to be made for this theory of change to be plausible.First, there is an assumption that community members and community groups have an ex ante understanding of why they are convening and are willingly doing so.It assumes that the engagement processes are structured in ways that allow participants to not only reveal their perspectives on the common need and preferences over how the need may be addressed but also to share their knowledge, networks and resources.It is also assumed that some type of collaborative action is completed in an effort to address the problem and that this action results from the commitment/enforcement mechanism and the process of engagement that preceded it.For instance, community members may successfully work together to construct and manage a clean water source because committing to work together now establishes a pattern of trust and reciprocity for future interaction.
Perhaps the most crucial link in this theory of change is the one that lies between the exchange of information among participants in the intervention and the development of credible enforcement devices.The type of enforcement mechanism that is necessary is not captured by a development or action plan; it is captured by the socio-political resources (dynamics) that underlie and shape the interaction.Early collective action theorists posited that without coercion, individuals who are self-interested and rational would not collaborate even though doing so would help them to meet common goals (Olson 1965).However, collective action has been shown to be possible under conditions where social norms and sanctions shape incentives (Ostrom 1990).These devices can take a number of forms including mutual affection, pro-social orientation, external enforcement, reputational gains and longterm engagement (Mansuri and Rao 2013).The main idea is that in convening, engaging and exchanging information, one or more of these mechanisms will be directly or indirectly activated motivating participants to commit to working together and to fulfil that commitment by contributing to or engaging in some collaborative task.
To stimulate a change in participants' general proclivity for collective action, the theory assumes that some benefit has been derived from the overall engagement.As with the previous theory of change, this theory remains agnostic about the type of benefit that is derived -for some it may be a practical or material benefit, for others a social benefit and for others a psychological benefit or some combination thereof.What matters is that the benefit is perceived to sufficiently outweigh the costs or to be preferable even if it does not.The participants must also attribute the benefit to the engagement within the CDD intervention.
For this theory of change, in order for a proclivity for collective action to develop, participants must value the experience.In order for them to value the experience, they must believe that the experience and its attendant results are both relevant and replicable outside the confines of the intervention.The theory would require further elaboration on this point.However, perhaps one factor that may increase the likelihood of value being ascribed to collective action is replication.If participants are given repeated opportunities to collaborate, perhaps under slightly varying conditions, the commitment or enforcement mechanisms could become more common and accessible.So if the same groups work together to fix the water source, repair the school and establish a grain mill, they might be more willing to work on other projects together in the future.This possibility has immediate implications for the design of CDD interventions: repeated opportunities (rounds, phases etc.) may be important not just to address issues around the depth of needs to be met but to facilitate repeated interactions that may shape future collective action tendencies.

Caveats
There are a number of caveats to this section of the paper.First, these are two of a large universe of potential theories of change around the specific outcomes of attitudinal change and collective action.Other theories of change may highlight very different and equally (or more) plausible mechanisms and assumptions.This paper does not intend to present these theories of change as the best possible ones; it seeks only to offer them as modest starting points.Second, as mentioned before, these theories of change are linear and relatively simple.The authors suggest that these be read as the core of the theories from which a number of extensions, nested alternatives and elaborations may be developed.The intention is to distil the most fundamental aspects of potential theories of change and to illustrate the thought process and assumption(s) behind them.Third, the theories of change above are, in some ways, theoretically limited; they refer only to the perspective of the individual and the group.This was a deliberate attempt by the authors again to clarify relative starting points for more clearly developing CDD interventions.Given that most CDD interventions often hope for wide-scale social, political and economic change (eventually) at the systemic level, theories about institutions and systems are extremely relevant.Nevertheless, in an attempt to start with the simplest common denominator, this paper has applied theories that are most relevant to individual and group level interaction and change.Other conceptual frameworks that engage historic and economic institutionalism and complexity are both necessary and valid in understanding and (re)conceptualizing the community-driven development approach (Bennett and D'Onofrio 2014).

Towards Better Design
Having proposed schemas for clarifying and prioritizing the objectives of the CDD approach and having articulated potential theories of change around specific governance and social cohesion outcomes, this paper now turns towards clarifying the core design elements of a CDD intervention.In order to isolate the implications of a specific theory of change, the design frameworkwhat practically constitutes a CDD intervention -must be clarified.Although the CDD approach is normatively prescriptive, it is a strategy and no specific implementation model exists.There are no ex ante stipulations about how much of a particular activity is needed, at what point it is needed, or for what duration.The precise design of a given CDD intervention should be a function of contextual, theoretical and logistical factors.This paper proposes that design choices are also influenced by the objectives, outcomes and theories of change that policymakers and practitioners select, prioritise and articulate.
The CDD approach requires that certain processes occur in order for the intervention to be classified as truly community-driven.For the purposes of this paper, these processes will be referred to as CDD's ' core processes.'In other words, CDD must meet certain procedural criteria or facilitate a series of processes in order to ensure that communities are given the opportunity to exercise decision-making power over the use of resources.These core processes are standard; they are common to all CDD interventions regardless of their specific goals and objectives.If the objectives, outcomes, theories of change and contextual implications of a given CDD intervention are clearly articulated, practitioners are then able to use this information to adjust and shape the core processes into specific designs that are contextually appropriate and theoretically-motivated.
For the authors, the following definition of CDD best portrays the components of the approach that make it community-driven: the existence of a community decision-making mechanism to determine priorities and the mobilisation of resources for achieving those priorities (Guggenheim 2011).As stated previously, the basic premise of the CDD approach is that if individuals and groups are brought together to define common or shared goals and are provided with resources and support, they can work together to achieve those goals.For the purposes of this paper, a community-level decision-making mechanism and resources are taken as necessary and sufficient conditions for the CDD approach.This means that every CDD intervention must have these features, though they need not be limited to them.
What are the core processes?CDD's core processes are derived from the need to have a community-level decisionmaking mechanism and a way to commit and utilise resources.The authors have identified these processes as community definition, information dissemination, convening, deliberation, preference articulation, commitment and performance.The articulation of these core processes follows a relatively simple logic.First, there needs to be a ' community,' which will be endowed with decision-making power.The community may be a pre-existing unit of social or administrative organization or a relatively new unit created by the implementing agencies for the purposes of the intervention.That community needs information about the intervention, the opportunity to participate and the type of participation that is required.In order for collective decision-making to occur, members of the community must engage each other whether as individuals, sub-groups or as representatives of pre-existing civil society organizations (e.g.community or faith-based organizations) to discuss, prioritize and deliberate over their development needs.The members of the community would then need to reach a consensus about the way in which aid resources would be used.This requires that the community's preferences (over the use of resources) be revealed to the implementers and supporters of the intervention.With preferences articulated and resources made available, communities undertake varying degrees of engagement with various stakeholders to elaborate and commit to a strategy for effectively using the available resources.Finally, elaborating a strategy is insufficient; the actual execution and management of the plan (or its performance) is also a necessary process in which community members may be involved to varying degrees.
These are the most basic processes that one would expect to observe under the CDD approach.The core processes are not about specific activities per se; rather they are about a series of steps that must be achieved for the approach to be community-driven.The strategies and activities that are undertaken to ensure that these processes unfold are not articulated here.For instance, in some contexts, secret ballot elections may be the appropriate mechanism for prioritizing community preferences while in another context a series of focus group discussions followed by a general assembly may be practically and theoretically appropriate.This discussion does not focus on this level of design detail but on the primary building blocks that constitute a community-driven exercise.
In Figure 3 above, each box represents a core process and contains a list of factors that may affect that process.The list is not exhaustive but indicates some of the factors which, based on theories in social science, we would expect to influence whether, how and with what consequences the core process unfolds.For example, the definition of community boundaries invariably shapes the social dynamics within and between communities.If the implementers merge small preexisting units for programme purposes, the CDD ' community' could be bringing together groups of people who did not previously engage each other around development issues.This means there is the potential for new rivalries and alliances among subgroups as they compete for programme resources.Conversely, if the boundaries of the CDD communities match the boundaries that participants are used to, then old rivalries and tensions may emerge in the community decision-making process.In either case, the way that groups within a ' community' perceive themselves and each other has implications for whether they are willing to participate in the intervention, how and how much they will interact, whether and how much they will compete or collaborate with each other and the extent to which they can resolve issues.Which groups are brought together is a function of how the community is defined.
Awareness of the factors that may support, prohibit or alter the core processes is important, particularly for policymakers and practitioners because they indicate the critical decisions and trade-offs that must be made in designing CDD interventions.A clear elaboration of the core processes also demonstrates how expectations around key elements of design and implementation such as framing, timing, and the range of stakeholders may shift as a result of the interaction of the core processes with these factors.
The diagram also depicts the processes as conceptually linear though they need not be strictly sequential.In fact, in practice, they are often not.The loops in the diagram indicate that there are plausible feedback processes e.g. as information travels and adjustments to the intervention are madethere may be several stages of decision-making and preference articulation.The loops also indicate that several intermediate processes may mediate the links between each of the core processes.For example, after disseminating information, programme facilitators have to engage local norms, procedures and authorities around bringing people together and may have to adjust the format of the assembly based on pre-existing formal or informal patterns.This demonstrates the contextual and procedural adaptability of the core processes.

Core processes and design choices
The set of design options for CDD interventions is partially determined by the selected objectives, desired outcomes, the associated theories of change and the core CDD processes.The core processes are essential components of an intervention's design.In order to translate these core processes into concrete activities and inputs, practitioners must answer a battery of questions so that they can tailor the intervention to the context and the desired outcomes and objectives.One way to approach the design of an intervention is to envision a layered process in which one first elaborates the design implications of the core processes, or at least the questions that need to be answered, in order to identify preliminary design options.The next step would be to superimpose the specific objectives, outcomes and theories of change on this preexisting rubric, again asking a series of questions emerging from having specified what must be achieved (objective and outcomes) and the plausible ways in which it could be achieved (theory of change).Taken together, the answers to these questions help to shape the final design.
For instance, one must determine what a 'community' looks like.Will it be based on pre-existing units, on the practitioners' sense of what would be most logistically efficient or on something else?To answer this question, there are a series of contextual questions about the nature of pre-existing units and theoretical questions about the function that the 'community' will serve for the theory of change, the objectives and outcomes that must be answered.For instance, a practitioner must determine whether a particular configuration of a 'community' is useful because of the social capital it already possesses or because the community is the vehicle through which competing groups will be able to engage each other constructively and in so doing build social cohesion.The extent to which these functions can be fulfilled through the use of pre-existing structures or whether they require new configurations can then be contextually and theoretically justified.As a result, the parameters of a 'community' will not be solely based on considerations of logistical and implementation efficiency but also on the implications of sound theory applied to knowledge of context.

Conclusion
Given the limited positive evidence around the effectiveness of the Community-Driven Development (CDD) approach in conflictaffected areas, donors, policymakers and practitioners face the arduous task of determining whether, how and to what extent the approach and by extension, investment strategies and evaluations of the approach, must be altered.This paper is motivated by the calls for more modest aspirations or objectives, more defined outcomes and better articulated and more explicit theories of change for the CDD approach.It presents a starting point from which revised approaches to CDD can be constructed.
The paper articulated the need for clarity around the objectives, outcomes, theories of change and core processes of the CDD approach, based on an overview of the state of evidence of CDD's effectiveness in conflict-affected contexts.Within this paper, the authors attempted to demonstrate ways to clarify and articulate the objective or purpose of a given CDD intervention, propose (reduced form) theories of change for two outcomes -attitudinal change and collective action -and a set of considerations around CDD ' core processes' and associated questions that could assist practitioners to identify design options and trade-offs.
Taking these ideas further requires engagement with the specificities of a given context and the set of problems that a CDD intervention might seek to address therein.The type of conceptual work that this paper outlines is useful in highlighting key assumptions about causality and driving greater precision in thinking about why and how a CDD approach might be deployed in a given context.The next steps are to translate these ideas into practice with more explicit problem analysis and programme design, such that future rigorous impact evaluations can test these more precise hypotheses in hopes of ultimately delivering better interventions to conflict-affected populations.

Notes
1 The Beyond Critique project is a research study, supported by DFID Research and Evidence Division funding.For further details see the project inception paper (Bennett and D'Onofrio 2014), which lays out major theoretical frames of reference of relevance for CDD, and the final working paper (Bennett and D'Onofrio 2015), which provides further analyses of relevant theories of change for CDD and examines some implications for policy, practice and learning agendas.

Figure 1 :
Figure 1: A Theory of Change around Attitudinal Change.

Figure 2 :
Figure 2: Theory of Change Around Collective Action.