Chapter 3, Module 1: Conflict-sensitive planning

Purpose of module

To help ensure that project / programme planning and implementation is conflict-sensitive, by incorporating conflict analysis into needs assessments and conflict sensitive design and planning. The module will help the reader to:

• identify the wider impact of planned activities on factors relevant to conflict
• ensure conflict is fully understood and linked to the needs assessment process
• ensure the needs assessment process is, in and of itself, sensitive to conflict.

Contents

1. What is conflict-sensitive planning?
2. Incorporating conflict analysis into the planning process
3. Steps for conflict-sensitive planning
4. Challenges
5. Endnotes

What is conflict-sensitive planning?

Planning has been defined as the process whereby certain problems are identified, their causal linkages analysed, and effective solutions developed, which can be implemented as a project or programme with objectives, activities and indicators. Conflict-sensitive planning brings in an additional ingredient — conflict analysis of the actors, causes, profile and dynamics in a given context — with the aim of ensuring that the project or programme does not inadvertently increase the likelihood of violent conflict, but rather serves to reduce potential or existing violent conflict.

Conflict-sensitive planning is called for in contexts involving all points along the conflict spectrum (from structural violence to violent conflict), regardless of whether the project or programme is for humanitarian aid, peacebuilding, or development; or whether the intention is to address conflict directly or simply to avoid indirectly exacerbating tensions. (See the Introduction to the Resource Pack for more information on the conflict spectrum). Conflict-sensitive planning is built on the elements identified during the conflict analysis in relation to profile, causes, actors and dynamics, and situates project planning within this analysis (see Chapter 2 for more information on conflict analysis).

Conflict-sensitive planning relates to both interventions that are defined through the conflict analysis, and to sensitising pre-defined interventions. A careful project strategy, taking account of each of these elements, can be the key to just and peaceful outcomes and more durable solutions. Because every activity is part of the conflict dynamic, whether focussed in, on or around a particular conflict, conflict sensitivity is relevant to projects and programmes that directly address conflict as well as those which seek simply to avoid indirectly exacerbating it. Thus, the conflict analysis needs to be integrated into the overall plan, and updated regularly. The overall plan can then be modified if necessary to reflect changes in the analysis.

Planning in a conflict-sensitive fashion is explored in more detail in section 3 below. Understanding the context involves taking the context (ie building on the conflict analysis triangle developed in Chapter 2), and situating the intervention within it.

2. Incorporating the conflict analysis into the assessment process

There are two ways of doing this: one is to link the conflict analysis (see Chapter 2) to the needs assessment (see Chapter 2 section 4), the other is to integrate the conflict analysis and the needs assessment into one tool. The advantages of having a stand-alone conflict analysis (linked) are that it explores the context in considerable depth, is easier to update, and avoids confusion that may be created by using one tool for two different purposes. The advantage of the second (integrated) approach is that it saves time and resources, and makes the processes of project design and conflict analysis more inter-related, as steps in the project design will raise questions regarding the context, which will in turn lead to further questions on project design. Box 1 gives examples of both approaches.

Whichever approach is adopted, the golden rule for understanding the dynamics of a problem is to involve all possible actors in the process of synthesising its key components.Stakeholder participation should not end with the identification of problems; stakeholders should also contribute to the analysis of the causes and consequences of the problems. Most agencies with a minimum degree of conflict sensitivity quote participatory processes as key to reducing conflict or even as part of the strategy for a constructive resolution of a conflict. Stakeholder-based analyses can provide a strong conflict transformation function (an optimalist approach to conflict analysis), although they are not in themselves conclusive as the information they provide is perception-based and not triangulated. (See Chapter 2 Box 10 and Chapter 3.3 Box 2).

3. Key steps to sensitising the planning process

3.1 The general approach

Planning a conflict-sensitive intervention requires careful and detailed exploration of the potential impacts, direct and indirect, (a) of the proposed activities on the actors, causes, profile and dynamics relating to conflict or potential conflict within the context, and (b) of the actors, causes, profile and dynamics on the proposed activities.

3.2 Step 1: Define intervention objective

Some organisations will use the conflict analysis to define their intervention, ie they begin with no pre-conceived ideas of what the intervention will be and use the analysis to decide on the objectives, by looking in particular at the scenarios generated by the analysis, and seeking to understand the possible key causes of conflict and how these may develop over time. For instance, if the analysis shows that water scarcity is expected to be a major source of conflict in the near future, improved water supply could be selected as the project objective.

Other organisations will have a pre-determined programme / project, and will use the conflict analysis to plan it in a conflict-sensitive way. For example their mandate may be to construct wells, and they will use the conflict analysis to determine where, when, how and for whom they will do this, usually through a process of prioritisation of causes and goals. Typical questions include:

• which issues (eg water, health care) or aspects of an issue (eg pastoralists competing for scarce access to water) appear to be most important?
• how does this choice relate to the context?

To be sensitive to an existing conflict, each question will have to be related to the conflict analysis (profile, actors, causes and dynamics) to see how the objective can be achieved in a way that will minimise unintended negative impacts on any of these four elements, and maximise positive impacts. This means trying to forecast the impact of the intervention. The scenarios developed in the conflict analysis (see Chapter 2 section 2.4) should help. The key is to consider the possible interaction between the proposed intervention and these different future contexts.

Either way, part of the design and strategy will be closely related to the logical framework analysis. Thus the logframe should include not only the overall objectives but also a description of the proposed contribution to improving the conflict situation (see Box 2).

In order to understand the limitations, and the potential areas where the intervention can have the greatest impact, four further questions need to be asked:

• what is my mandate?
• what is my capacity?
• what are other actors doing in this area?
• what is their capacity? (See next section)

At the planning stage, the framework of "control" versus "assist" versus "influence" can help individual organisations to understand the degree to which they can contribute to changes in any operational context. Organisations need to understand which factors and issues they control, which they can assist, and which they can influence. They need to be honest as to what an intervention can be expected to deliver — raising high expectations and failing to deliver can cause tensions and ultimately lead to conflict. Furthermore, understanding the joint impact of programming and interdependence with other actors will help outline where common approaches with complementary actors are necessary, or alternatively where a new intervention could be counterproductive to existing work.

3.3 Step 2: Define the intervention process

Having defined what the project objectives are, the intervention process itself must be designed in a conflict-sensitive fashion. The principles of transparency and accountability require developing a clear set of selection criteria for who, where, and when.

Who: Project beneficiaries, project staff, and operational partners

Experiences with Mary B. Anderson's Do No Harm framework1show the importance (and, sometimes, interaction) of carefully designed selection criteria for these three elements. Each of them can influence the causes, actors and dynamics of conflict in a positive or negative way. The identity of the persons or groups (ie their political affiliation, gender, caste, socio-economic profile, etc.) can have an important impact on the conflict. The selection criteria should therefore be directly derived from the conflict analysis and the project objective.

Project beneficiaries

The selection of beneficiaries must relate to both the needs assessment (or other form of assessment) and the conflict analysis:

• is the selection based on need (ie in terms of equity)? or should an entire community benefit, irrespective of differences in need (ie selection based on equality)?
• how does the selection relate to divisions within a community and what are the implications of that?

An equity-based approach, which by definition cannot favour inclusiveness, would normally require the use of selection criteria to determine who falls within the beneficiary group. This improves the transparency of the intervention. In some situations the community themselves decide who should be the beneficiaries, generating and implementing the selection criteria. Such processes may require carefully built-in safeguards to ensure equitable results — eg so that no one group is able to dominate and exclude other groups.

Project staff

Local and expatriate staff can both bring either benefits or disadvantages. Certain nationalities may be seen as biased because of political tensions between the host country and their country of origin — or may be generally well received if there is a history of friendship. Speaking local languages or dialects can be a key element of conflict sensitivity, but local staff may not always be perceived as neutral. Staffing can be a key element of conflict sensitivity (see Box 3). Situating the proposed intervention within the conflict analysis should reveal such sensitivities.

Success in mediation and intervention in disputes often hinges on the status of the intervener. Sometimes it may be useful to be able to speak the language of local politics; on other occasions someone totally unencumbered by local knowledge may be more effective. Foreigners, particularly in places with a colonial history, are clearly identified as outsiders; their "otherness" may be a severe handicap, or a great advantage, depending on the context. They need to know which.

Operational partners

Developing partnerships is a challenging process and needs to be undertaken with care. Many of the issues should be revealed through the conflict analysis. Of particular concern is the understanding of who the actors are, and what their relationships are with other actors. For instance, it may be that a potential partner organisation has links to an armed group, or that personnel move regularly between the potential partner and the armed group. In some situations conflict protagonists have gained legitimacy through partnerships with international actors. Such knowledge can be acquired only by a thorough conflict analysis, focussing on the actors.

Equally important are the perceptions project participants have of the potential partners. They may be perceived as biased, insensitive to conflict, or to have links with potential or existing conflict protagonists. Whether or not these perceptions are grounded in reality, they are part of the operational context and should be taken into account in decisions about partnerships.

The capacity of potential partners is also important: what staff do they have? What is their mandate? What is their track record? What are the prospects for capacity building? CARE Sri Lanka have developed a partner assessment tool, leading to a process of partner capacity building (see Box 4). The decision to engage in such capacity development remains context-specific — in some situations an open and honest dialogue with potential partners on their conflict sensitivity could form part of a process of building this sensitivity. In other situations this may not be possible, and a decision may be taken not to engage in capacity development, not to engage with the partner at all, or to engage only in ways that build on what capacity the partner already has.

Assessment of the capacity of the intervening actors to impact conflict and peace is an essential part of assessing the potential for conflict and peace. Even if an institution has the responsibility for implementing a programme, it may not have the capacity. This is often true for governments or civil society in conflict-prone or -affected countries. International institutions often jump in to fill the gap, but they must be careful not to replace government capacity, creating a parallel system that impedes development of local capacity in the long term. A proactive strategy of capacity building may be the best option, even (in some cases) at the cost of low performance.

Staff concerned with implementation may not understand how the programmes/projects being implemented or supported impact on conflict and peace. They often see these issues as outside their operating situation, and overlook both their own potential for positive or negative impact, and the capacity of their organisation to make a significant difference. Involvement in the conflict analysis, programme development and monitoring and evaluation will help broaden their horizons, and perceptions.

Where: which geographic area to support

Determining the geographic area of support requires a full reflection of the relationship between the outcomes of the needs assessment process and the conflict analysis. It should also be developed with the input of all sectors of the community.

The selection can have a direct influence on a conflict context, for instance by exacerbating the violation of land rights, by providing (often unintended) support to certain military or economic interests, or by legitimising the political power of some groups or individuals. These are cases where an intervention can inadvertently exacerbate conflict (and even directly endanger the lives of the population) or miss the opportunity to mitigate it.

Remember too that most interventions will not benefit the entire population; there will inevitably be non-beneficiary people or communities located at varying proximity to the beneficiaries. The transparency of beneficiary selection has been addressed above, but it may be necessary to communicate this more widely, or even to broaden the selection, perhaps in coordination with other intervening organisations.

It is important to understand the geographic determination of the beneficiary community — does it fall along lines of division? Could it worsen an existing division — or create a new one? If the intervention specifically seeks to impact conflict, then this focus in itself will determine geographic locations for operation. These questions should all be addressed through the conflict analysis, ensuring that the intervention is targeted at the geographic level (eg national, district, local) appropriate to the context.

When: Timing and length of intervention

In conflict situations, time management is a core resource. Several "lessons learned" documents on conflict-related planning have highlighted the importance of timing in the injection of resources. Again, linking back to the conflict analysis is key.

Two elements of the analysis demand particular attention: conflict triggers and scenarios. Conflict triggers may be time-bound, such as an election or annual cycles of offensives linked to seasonal changes. Understanding triggers is important in deciding when to start and when to exit, and when contemplating any major changes in the intervention. Scenarios (see Chapter 2 Section 2.4) should be related to the proposed project timeline, and assessed for potential windows of opportunity or vulnerabilities. This may involve a reflection on the motivations behind the timeline — is the length of intervention being defined by organisational objectives, resource constraints, or by the needs of the context?

3.4 Step 3: Develop indicators

Conflict-sensitive indicators fall into three principal categories:

• Conflict indicators, developed during the conflict analysis stage, are used to monitor the progression of conflict factors against an appropriate baseline, and to provide targets against which to set contingency planning
• Project indicators monitor the efficiency, effectiveness, impact and sustainability of the project
• Interaction indicators, developed during the planning stage by taking the information gathered during the conflict analysis and applying it to the project planning process, are used to monitor and evaluate the interaction between the project and conflict factors by (a) measuring the impact the project is having on conflict (eg restricted access to safe drinking water no longer antagonises ethnic minority); and, obversely, (b) measuring the impact conflict factors are having on the project (eg number of staff approached by militants for financial "donations"; rising tensions make inter-group activities difficult to conduct).

Chapter 2 (section 3) provides a description of conflict indicators, and Chapter 3.3 (step 2(c)) details the development of interaction indicators. Most organisations already employ programme or project indicators as a means of measuring the outputs and impact of their work against a baseline determined at the outset. Given the wealth of information available on project indicators 2, they are not examined here in any depth. However, as a conflict-sensitive project will have conflict-related outputs and impacts, these will need to be reflected in the project indicators. Project indicators should enable measurement of the various aspects (profiles, actors, causes and dynamics) of the context that may be affected by the project.

Interventions are commonly undertaken in partnership. It will be useful to have indicators that measure the impact of the actions of each partner. This is not simply to attribute credit, or blame, but rather to identify which approaches worked well, which did not work well, and why. Contribution programming is a way of attributing impact to different actors, and of understanding that no one actor alone is entirely responsible for a given situation. It is also a key concept in the area of conflict where real results are the consequence of the combined actions of different parties (see Module 3 for further details).

3.5 Step 4: Link project to scenarios and prepare contingency plans

In the absence of careful contingency planning, proactive programme implementers may react with potentially ill-conceived responses when quickly changing contextual environments throw up difficult circumstances. For example, if conflict dynamics rapidly deteriorate, an organisation may make a snap decision to evacuate, possibly leaving national staff at risk and beneficiaries suddenly without support.

Contingency plans define predetermined strategies for reacting to specific changes in the operational context. Put another way, if conflict dynamics deteriorate to a particular point, what actions will be required (see "Event" and "Response" columns in Box 5 below)? How will they be carried out? Who will undertake them? Within what timeframes? Contingency plans are designed using scenarios (see Chapter 2 section 2.4) in conjunction with conflict-sensitive indicators that monitor the evolution of a given conflict dynamic (see above).

A natural reaction to increased insecurity and violence is for implementers to move the project to the national capital or halt operations in the hope that things may soon improve. There is also an unfortunate tendency for contingency plans to focus on expatriate staff and neglect national staff and partners altogether. Conflict-sensitive contingency plans will need to include security for all situations and all people — staff, partners and beneficiaries. A well thought out conflict-sensitive contingency plan will allow for a continued level of engagement in a wide variety of difficult circumstances based on the organisation's detailed knowledge of the various profiles, causes, actors and dynamics. Should evacuation be required, a conflict-sensitive contingency plan will ensure a level of continued support and safety for staff who are not able to leave the region or country.

Contingency plans will allow for a level of stability and measured responses to difficult circumstances. The plans should allow for a degree of flexibility so that implementers may respond appropriately to circumstances as they arise, based on the detailed knowledge they have gained through the conflict analysis and careful monitoring of conflict-sensitive indicators.

During the latter part of the armed conflict in Sri Lanka, CARE developed a Risk Management Matrix to foresee possible events and plan adjustments in the project to mitigate the impact of the conflict dynamics on the project. Table 1 gives an adapted example of the format and the analysis of an actual risk.

Table 1: Risk Management Matrix

Event	Probability	Response	Risk Management
Description: Brief description of the event and its consequences on both project and population Level of risk as the seriousness of the consequences: Low / Medium / High	Description: Brief description of historical occurrences and of potential reasons or cases in which the event could take place Degree of probability: Unlikely (10%) Possible (20 to 40%), Probable (50% and up)	Description: Reaction from the project to adapt to the new or temporary circumstance	Description: Measures in place for early warning and for immediate response.
Example: Security situation in project areas deteriorates causing displacement, destruction and reduction of socio-economic conditions. Project development strategy can no longer be effective. (Natural disasters in the area could have similar consequences). Risk: Medium/ High	Example: This has happened before (most people have been displaced more than once). In the past, such events have always been temporary — and conditions have recently improved. The situation must be seriously considered for the safety of the staff and project participants. (Natural disasters are possible though not as probable). Possible: 30%	Example: The project refocuses on the secondary group of participants defined in the needs assessment (if in condition to benefit from the longer term development approach). The project implementation schedule is reviewed to allow for some of the staff to be temporarily diverted to the immediate, emergency work with previous participants.	Example: The security situation is reviewed every week to enhance the ability to predict possible changes in the security situation of project areas. Permanent emergency funds have been secured from headquarters in order to maintain responsibility towards donors and provide relief support as well.

3.6 Step 5: Designing the project conclusion -sustainability, structural change and project evolution

Projects and programmes eventually come to an end. To ensure they remain conflict-sensitive throughout the project cycle, the project conclusion must be planned and executed in a sensitive manner. Broadly speaking there are three approaches available: (a) withdrawal at termination of project; (b) extension of project; and (c) following up the project with a new phase. To conflict-sensitise this stage of planning, exit strategies need to be flexible enough to address changes in conflict dynamics, but proactive enough to ensure beneficiaries" expectations are not unduly raised. Decisions should be well thought out, and should respond to dynamics identified through the conflict analysis and subsequent monitoring.

One approach to maintaining the balance between proactive planning and flexibility is to develop a menu of exit strategies that are reviewed throughout project implementation against the updated conflict analysis and regular monitoring. As the end of the intervention nears, the most appropriate exit strategy can be chosen. Early development of a variety of exit strategies will help to ensure that the eventual exit has been well thought out; is effectively communicated (along with other possible strategies and their associated triggers) at every stage of the project to minimise unrealistic expectations by staff and beneficiaries; and responds to the conflict dynamics identified through the updated conflict analysis. While the menu may create some ambiguity, this is a price worth paying to ensure that beneficiaries and others are prepared for the exit when it comes.

For projects that are intended to phase out completely, sustainable solutions need to be conflict sensitive. Most peacebuilding and development projects foresee a strategy of sustainable structures that stay behind to guarantee that the benefits of the intervention will remain in the long term. In order to make sustainability conflict sensitive, exit strategy planning needs to link back to the conflict analysis and scenarios, to see how they interact with these structures. Ensuring the sustainability of remaining structures and processes may require mainstreaming conflict sensitivity in local organisations and teaching conflict analysis skills to local partners.

For projects that intend to extend or adapt into new phases, it will be important to ensure that new interventions take into account any changes in conflict dynamics. Using the contingency plans outlined in Step 4 will help ensure that future approaches reflect not only the best-case scenarios but also the worst. The project should build on the successes of the previous project.

Whether the exit strategy is planned to be a complete phase-out or an adaptation into a new intervention, it is vitally important to plan for a proper conclusion of the initial intervention, including a comprehensive evaluation. In some cases the most negative impacts of interventions come not from their implementation, but rather from a poorly designed exit strategy (eg the impacts of an otherwise favourable intervention can be undermined when project staff, partners and suppliers suddenly find that their contracts will not be renewed). A well-planned exit strategy will not only help to seal the success of the initial project and leave a strong foundation for future interventions, but will also reduce the possibility of continuing existing activities — or designing new ones — that cause or exacerbate conflict.

Planning for the next intervention before the current intervention has completed its implementation phase is clearly not without its challenges. Nevertheless, thinking clearly and realistically about an exit strategy in a manner that balances being flexible with being proactive will help ensure the intervention is conflict-sensitive not only in its current phase, but beyond its anticipated lifespan.

Challenges

4.1 Relations with central and local authorities

Conflict sensitive planning and identification inevitably raise the question of conditionality. Central and local authorities that interact with the intervention may have policies or approaches that appear insensitive to conflict and risk, undermining the objectives of the intervention. As part of the planning phase, the intervening organisation may feel the need to effect changes in existing policies or practices of authorities to meet the minimal conditions required for a project or programme to succeed. Conflict-related conditions may include, for example, state willingness to support the independence of the justice system, or to pay appropriate salaries for security sector personnel.

To make the implementation of a programme conditional on the partners meeting these terms often requires a political commitment that is beyond the scope of planners. Finding allies and developing commitments from all levels of the intervening organisation will ensure that the issue of conditionality does not remain with planners, but can instead receive the attention and support of the entire organisation. The approach adopted by many organisations is to introduce the issue of initial conditions in the first stages of design, and later reflect the issue in evaluations and policy papers, in an attempt to make the dilemma more visible to political decision makers.

4.2 Information networks

Conflict sensitive programming requires networks to provide information on the context, and the interaction of the intervention with that context. Many organisations have existing information networks, and these may require further expansion to capture the sensitive and context-specific information required to monitor relevant changes and impacts. The principle of transparency may need qualification in this situation, as information sources sometimes require protection, and the information they provide might need to be treated with confidentiality. This is a context-specific question, but clearly sources should not be put at risk. A particularly innovative information network has been created by Al Quraish in Sri Lanka involving "Peace Birds" (see Box 5).

4.3 Relations with donors

This planning module has predominantly used an implementing agency's perspective in describing the various elements involved in conflict-sensitising a planning process. Donors of course do planning too. The following paragraphs address the unique challenges donors face in conflict-sensitising their planning processes, and outline some of the issues governments and NGOs need to consider in interacting with their donors on issues of conflict sensitivity.

Many donors rely on a tender process where the donor plans the project — sometimes down to specific details like how many days one particular type of technical advisor will spend supporting the intervention — and organisations bid by submitting an implementation plan and associated budget.

This planning module has argued that a project plan must be tied to some sort of conflict analysis. If donors do not undertake a conflict analysis, do not tie the project plan to the analysis, but instead design the plan based on an assessment that is then imposed on bidders, then ensuring a conflict-sensitive intervention will be highly problematic.

Further, the time required by most donors to conceive of an initial intervention idea, design it, secure the necessary internal funding, and then proceed through all steps of the bidding process means that any initial assumptions about conflict dynamics are often outdated by the time project implementation finally begins. Even if the initial project planning incorporated key elements of conflict sensitivity and was developed in a timely manner, many tenders necessitate a level of inflexibility that is at odds with conflict sensitivity: forced partnerships, restricted timing, specific location, and detailed specifics about the intervention itself.

Donor-funded projects that do not use a tender process also face challenges related to conflict sensitivity. While DFID (UK) has recently untied its aid, some donors still require that funding favour goods and services from their respective countries. The clearest example of tied aid and its potential for negatively impacting conflict is monetisation. Recipients are offered a product from the donor's country in lieu of cash. The recipient then sells the commodity in the country in which they operate and uses the revenue to fund an approved project. Monetisation can undermine local production and distribution networks and fuel corruption, patronage and other root causes of conflict.

Whether or not aid is tied, donors that wish to conflict-sensitise their funding relationships could request applicants to include conflict analyses with their proposals in addition to the gender and environmental impact assessments most currently demand. A conflict sensitive end-of-project evaluation should also be required. In both cases, donors should provide the resources required to support these additional components.

More broadly, donors can conflict-sensitise their funding relationships by conducting their own broad conflict analysis and then evaluating projects on the basis of how they fit into the conflict dynamic (in addition to the regular criteria). The conflict assessment required from funding applicants should be seen as complementary to the conflict assessment conducted by the donor agency itself, and cases of contradiction should be seen as opportunities to learn more about the complexity of the conflict dynamic. Box 6 illustrates a tool used by one agency to analyse incoming proposals against the likelihood of their negatively impacting on conflict dynamics (the agency,CARE, is usually considered a generalist international development NGO, but in this case was acting as a donor, being responsible for funding grants and evaluating project proposals submitted by other organisations).

Accepting funding from some donors may imply political support or an affiliation that could jeopardise the implementer's conflict sensitivity. Both peacebuilding and human rights work are premised on the political independence of the implementing agency. This has boosted the involvement of civil society during the past twenty years, often structured into NGOs, unions and religious groups. The limited number of donors and the multiplication of sophisticated fundraising actors — and thus increased competition for scarce funding resources — have led to the emergence of a real quandary for planning the development of such activities: should implementers remain small but independent, or should they align with public funding priorities and grow?

Some tools, such as the Clingendael Conflict and Policy Assessment Framework (see Chapter 2, Annex 1, item 9), advise using a cost-benefit analysis to define the true cost of a particular objective in conflict prevention in relation to other objectives. Such analytical tools help decision-making not only regarding project activities at the implementation level, but also on wider programmatic issues, such as aligning with external funding objectives, at all levels of an organisation.

Box 7 summarises some of the measures adopted by organisations to secure operational capacity and independence.

5. Endnotes

1 Mary B. Anderson (ed), "Essays from field experiences", Cambridge Mass: CDA, Inc: 2000.

2 See, for example, Frances Rubin, A Basic Guide to Evaluation for Development Workers, Oxford: Oxfam, 1995