Transitional Justice and Theories of Change: Towards evaluation as understanding

This article has two goals. First, to make explicit the theories of change currently operative within transitional justice and, second, to critically engage with both these theories, and dominant theories in international development. As such, it seeks to replace a focus on results, attribution, and linearity with a privileging of process, contribution and complexity. Developing theories of change for transitional justice is challenging, as it is characterised by diverse interventions, complex and contested contexts, and the need to balance principles and pragmatism. Normative, linear and mechanism-based claims remain dominant, while the evidence base for transitional justice is still weak. This article looks at insights from adjacent ﬁelds, some of the challenges facing the development of theories of change within transitional justice, and evidence from impact studies and evaluations. In a ﬁnal section we propose an alternative, drawing on complexity theory and actor-oriented approaches, which suggest an important set of terms – systems, interaction, contingency, context, encounter, emergence, incrementalism – to inform what we term evaluation as understanding.


INTRODUCTION
There is a fundamental and existential problem with transitional justice: it does not really know what it is. In part due to a lack of what development practitioners term the 'theory of change', it is very difficult to delineate what and who transitional justice is for. Both a serious cause and consequence has been the expansion of the concept to incorporate a huge range of objectives and claims, from formal prosecutions to broader development goals, without sufficient critical reflection. Transitional justice is an over-burdened and underconceptualised idea. 1 A theory of change can be understood as setting out 'underlying assumptions about the relationships between desired outcomes and the way proposed interventions are expected to bring them about'. 2 In the field of transitional justice, few nongovernment organisations (NGOs), governments, inter-governmental organisations, or donors operate with an explicit theory of change. To the extent that there are implicit theories of change, they are often not clearly articulated, unsubstantiated by evidence, and mutually contradictory. As such, work on developing theories of change is needed to clarify the parameters of transitional justice, to ground expectations of achieving particular outcomes, and to achieve greater coherence within transitional justice interventions and between transitional justice and adjacent sectors. This article has two goals. First, to make explicit the theories of change currently operative within transitional justice. Second, to critically engage with both these theories, and dominant theories in international development, and explore the potential for theories of change to drive a practice that is evidence based, contextualised, processdriven and inclusive of concerned communities. The article champions evaluation as understanding, advocating a relativist approach, subjectively rooted in the change desired by stakeholders. 3 We seek to move beyond the conventional positivist (or more properly, post-positivist)-constructivist dichotomy in evaluation by seeking approaches that are relevant where programmes to be evaluated exhibit nonlinearity and other structural aspects of complexity.
Transitional justice presents considerable challenges to the development of theories of change. Addressing the past in transitions from conflict or repressive rule encompasses different mechanisms or approaches (criminal prosecutions, truthseeking, reparations, reform of law and institutions); highly contested and often fragile contexts ('transitions'); a rapid expansion of interventions, goals and expectations; and a relatively new field at the early stage of theory-building. 4 Scholarship has highlighted the tensions within transitional justice, for example between the goal of providing justice to victims and 'causal beliefs' about facilitating a transition to democracy, 5 and between 'irreconcilable goals' such as maintaining order and facilitating transformation. 6 Broad concepts such as justice, truth, reconciliation, peace and democracy, although contested, need some degree of clarity and focus, and the relationship between concepts also requires clearer theorisation. Given these challenges it is not surprising that there have been calls to pause and take stock. Mendeloff's plea to 'curb the enthusiasm' in relation to claims made about truth-seeking and -telling is illustrative: Claims about the peace-promoting effects of formal truth-telling mechanisms rest far more on faith than on sound logic or empirical evidence. The literature has done a poor job of specifying the logic of truth-telling arguments, defining and clarifying key concepts, operationalizing key variables, indicating the conditions under which proposed relationships hold, providing compelling empirical evidence to support core assumptions, and testing claims systematically against competing explanations. Assertions are frequently presented as empirical fact when they are merely untested hypotheses. In short, truth-telling advocates claim more about the power of truth-telling than logic or evidence dictates. 7 Despite these challenges, transitional justice advocates often make sweeping claims that the core components of their trade deliver significant change, e.g. that truth-telling contributes to reconciliation, that prosecutions act as a deterrent, that institutional reform can aid non-repetition of violations, and so on. These constitute a version of what Eyben et al. call an 'archetypes framework', 8 where change is implicitly thought to come about through some taken for granted conventional wisdom. Notwithstanding a significant growth in evaluation and theory-based work in the last decade, many theories of change in transitional justice remain normative and human rights-based, rather than empirically rooted, resulting in a tendency to present interventions as a self-evident good, rather than contested practices requiring justification and an evidence base.
Given the growing emphasis on evaluation in transitional justice it seems clear that theories of change will also become more prevalent in the future. This article does not blindly advocate for the wider use of theories of change. Rather, the vision set out here is of an approach to theories of change that goes against the grain of the mainstream, both of current practice in transitional justice and of dominant approaches to theories of change in international development. In this article, we seek to use theories of change as a way of conceptualising transitional justice as a process, from conception to evaluation, with the aim of using such theories as a tool to drive processes that are bottom-up and accountable to concerned constituencies. The next section defines the concept of theory of change in more depth, identifies relevant theories of change that can be transposed from fields such as international development, and expands on the challenges in building theories of change for transitional justice. We then discuss impact studies and evaluations of transitional justice to identify the dominant current pathways to understanding change processes. In a final section we propose an alternative means of operationalising theories of change, drawing on complexity theory and actor-oriented approaches, which suggest an important set of terms -systems, interaction, contingency, context, encounter, emergence, incrementalism -to inform what we term evaluation as understanding.

THEORIES OF CHANGE IN TRANSITIONAL JUSTICE
This section clarifies the problem that the article seeks to address, through the lens of definitions of theories of change, mainstream theories from adjacent fields, and challenges facing the clarification of theories of change within transitional justice.

Defining Theories of Change
At its most simple, a theory of change can be understood as a testable hypothesis, in social science terms, or a justification for an action. As suggested by Aragón and Macedo's definition above, theories of change provide 'a way to describe the set of assumptions that explain both the mini-steps that lead to a long-term goal and the connections between. . . activities and the outcomes of an intervention'. 9 Developing a theory of change can be perceived as producing an output that describes how activities lead to outcomes, highlighting assumptions, justifications and pathways which 'unpack the black box of causality' 10 , or as a process, with an emphasis on conceptual thinking and ongoing reflection or cycles of learning designed to understand the relationship between activities and outcomes.
Among the lessons from the use of theories of change in international development are that they can have both positive and negative impacts. The breadth of understandings of theories of change in development range from a highly technical planning tool -an extension of the assumptions made in a log frame, for example -to a politicized approach to understanding how particular actions impact on power relations in a specific context. Much of the enthusiasm for theories of change has come from what might be called the 'results agenda': a desire to demonstrate that donor funds in international development are delivering change and 'value for money'. Whilst such approaches might seek to use the language of participation, they are often driven in a top-down way and prioritize donor agendas. However, theories of change can also be envisaged as a process-based, problem-solving, reflective and participatory approach in which a change model is developed and continuously revisited, inclusive of a range of stakeholder perspectives. 11 Theories of change can also facilitate a move beyond a focus on projects and programmes to a reflective and flexible understanding of how change takes place in a given context and community, and what role organisations and interventions could play in supporting such change. As Valters reports: 'This definition helps tackle a recurrent problem with Theories of Change -that organisations imply that change in a society revolves around them and their programme, rather than around a range of interrelated contextual factors, of which their programme is a part'. 12 In its final section this article champions a process-based approach to theories of change, evaluation as understanding, in which outputs and outcomes are not predetermined but rather characterised by contingency, emergence, and incrementalism. What follows below is a discussion of more conventional, largely output-based approaches to theories of change.

Borrowing Theories of Change
The table below sets out six core theories of change which derive from fields such as international development, but which can be applied to transitional justice. The theories in the table are often pursued in combination.  Causal change theories (if x happens then y will follow, because of z).
Change is rarely linear, unidirectional or simple, and attribution and cause and effect are often difficult to clearly pin-point.
There are numerous examples of this approach in transitional justice: that truth leads to reconciliation, prosecutions act as a deterrent, etc. Actor-centred theories focus on the actions and behaviour change required in individuals (elites, victims, perpetrators, spoilers, members of the public).
More process oriented, but can prioritise actor behaviour and relationships at the expense of changes in the lived reality of stakeholders and target groups.
Actor-centred transitional justice theories include elite-pacting and approaches that focus on victim-survivor mobilisation.
Coalitions or collectives theories (networks or partnerships).
More process oriented and global in perspective, but can prioritise actor behaviour and relationships at the expense of changes in the lived reality of stakeholders and target groups.
The most high-profile coalition theory is the boomerang model within transnational advocacy networks. 14 Policy window theories (windows of opportunity when change is possible). Individually and in combination these theories of change provide useful insights, as well as significant limitations, in understanding how change is perceived and how interventions might drive change. What follows builds on the theories in this table to highlight a set of challenges confronting the development of theories of change in transitional justice.

Challenges to Building Theories of Change for Transitional Justice
The discussion below identifies four core challenges facing the development of theories of change within transitional justice. These challenges are: links to very broad, Transitional Justice and Theories of Change 285 macro-level goals; the relationship between levels of impact; the lack of theorising about transitional justice mechanisms and the changes they seek to effect, as well as on the connections between mechanisms; and prioritisation and sequencing within complex, 'holistic' interventions. First, the goals of transitional justice at a macro or state level are often framed to encompass outcomes such as reconciliation, peace-building, and strengthening the rule of law and democracy. 18 The pathways linking transitional justice to wider processes of social change remain largely implicit -whilst truth is presumed to be a tool for peace-building or strengthening democracy, for example, the mechanisms to deliver these changes are under-developed. The literature seeking to link transitional justice to strengthening the rule of law and democracy is illustrative of this first challenge. Cross-national quantitative studies present contradictory findings about the links between transitional justice and democracy, and provide little by way of clear policy guidance. 19 Qualitative research and inquiries using mixed methods have led the way in moving beyond correlation to causation. This research has begun to disaggregate the components of democracy that transitional justice is most likely to have an impact upon (the rule of law, security forces, and participation), and identify potential pathways to impact (promotion of legal and policy reforms, empowerment of marginalised actors, and (de)legitimisation of individuals, groups, institutions and ideologies); 20 call for assessment of the effects of mechanisms within and between fields / sectors, and for more complex rather than linear causal dynamics, including the possibility of negative effects; 21 and emphasise context (national, regional, international) and timing (the passage of time, sequencing of interventions, national and world time) as vital to the evaluation of transitional justice processes. 22 While literatures linking transitional justice to macro-level change processes are becoming more sophisticated, they remain largely uncritical of liberal and Western definitional criteria e.g. of democracy, hampered by the challenge of mapping causal relationships and integrating transitional justice interventions into broader processes of change, and beset by the fundamental question of whether effective transitional justice is the cause or effect of broader goals such as strengthened democracy.
Second, transitional justice demands change at multiple levels (individual, community, national, etc.), but is imprecise in defining which are important to any particular process or mechanism, or how change at different levels is connected. 23  , 193-96. 23 'It is perfectly plausible that the same policy may have positive macro-level effects but negative microlevel effects, the potential of amnesty legislation to lead to such an outcome has been widely suggested': Macdonald, supra n1 at 2, 59.
transitional justice processes comprise national institutional mechanisms, such as truth commissions and trials, which unfold largely in capital cities, while having elements -such as public hearings or testimony collection -that take place in local spaces. These can be complemented by other mechanisms, such as symbolic or material reparations, that have highly visible local manifestations. Multiple levels of activity are often characterised by profound differences, and tensions, in the institutional spaces used, the actors involved, and the moral repertoires which carry legitimacy. Wilson talks of 'relational discontinuities' 24 in South Africa between the Truth and Reconciliation Commission (TRC) and community-level social processes. These discontinuities existed between the TRC and local justice structures, elites and the masses, as well as normatively between reconciliation and human rights on the one hand and revenge as the dominant local response framework on the other. But research also illustrates that outreach and forms of translation can form connections and more positive forms of diversity, if not uniformity, across different levels.
Jeffrey's work on the Court Support Network of the War Crimes Chamber of the Court of Bosnia and Herzegovina, analyses 'the spaces through which judicial processes operate and the spatial imaginaries they bring into being'. 25 He argues that in this setting outreach 'brought new spaces of justice into being through education programmes, transnational connections and innovative forms of political participation'. 26 The spaces of justice as a result were plural -legal and non-legal, produced by the state and non-state actors, supporting the Court but also establishing their own practices of justice, and conceptualizing justice in different ways (less a legal process of retribution and more a 'shared experience' focusing on restorative notions of deliberation and reconciliation). 27 An effective theory of change for transitional justice must be able to link different levels in societies in ways that extend beyond a vague hope that change will 'trickle down' from national, institutional interventions, and that nuances current theorising on transnational advocacy, boomerangs and cascades (see Table 1). Theorising is needed on how diverse actors mobilise resources, alliances and discourses across levels or spaces to achieve desired outcomes. Third, there is also a lack of theorising about transitional justice mechanisms, the changes they seek to effect, and the connections between mechanisms. For example, Cronin-Furman dismantles what might be termed an 'archetype' theory of change in transitional justice: the assumption that international criminal trials produce a general deterrence to other potential offenders. 28 Drawing on two bodies of social science theory, on criminal deterrence and the commission of mass atrocities, she argues that the work of the International Criminal Court (ICC) is not calibrated to produce a deterrent effect because of the lack of certainty and severity of sanction and the selection of offenders (commanders who order offences are less likely to be deterred than those who permit or fail to punish offences). Further, scholarship on the relationships or interactions between transitional justice mechanisms is scarce and contradictory. 29 Bisset argues that while the 'theoretical compatibility and complementarity of these mechanisms' has been widely affirmed, 30 there has been 'a failure to assess whether their actual modes of operation are compatible'. 31 A 'practical discord' between the investigations of truth commissions and trials remains, for example in the areas of access to information, the exchange and provision of evidence, and the role of witnesses. 32 Tensions exist between truth commissions and trials when they operate simultaneously, but also when they take place sequentially. In relation to Chile, Accatino and Collins have argued for the existence of various 'truth orders', and for truth as 'accumulation' rather than a one-off construction. 33 Chile has experienced three truth commissions -the Rettig Commission on death and disappearances; the Valech Commission on torture and political imprisonment; and the New Deal with Indigenous Peoples 34 -and, since 2000, around 1,000 former security service personnel have been placed under investigation or on trial for gross human rights abuses committed during the 1973-1990 military dictatorship. In this context, the truth commissions questioned previous judicial judgements, and over time their findings have in turn been questioned by further judicial proceedings. The questioning of administrative, and socially accepted, truths may be seen as necessary given judicial standards of proof and due process guarantees, but the fact that the truth is potentially permanently 'transitory' may be profoundly unsettling for victims and their families. 35 It is therefore clear that transitional justice theories of change need to build on, but go beyond, theories identifying policy windows or the role of coalitions and collectives (Table 1), to encompass the huge complexity implied by the use of diverse mechanisms, with multiple functions, interacting over long periods of time. Fourth, more holistic interventions and proposals, acknowledging that transitional justice must work at different levels, through diverse mechanisms and sectors, often resemble complex and ambitious programmes without a strategy. 'Do everything' is not a helpful instruction for policy makers or practitioners. Dimensions of influence and policy window theories (see Table 1) indicate the need to manage complexity and for patience and preparedness, so that political openings are exploited when they arise. Literature on prioritisation and sequencing within holistic interventions is emerging. Accatino and Collins suggest unsteady and non-linear progress towards truth and justice, where outcomes become a contested accumulation. 36 With a broader focus on Latin America, Dancy and Wiebelhaus-Brahm argue that neither timing nor sequencing of transitional justice interventions significantly alter the potential for democratic consolidation, with trials coming closest to being a necessary condition for successful consolidation. 37 Scholarship on the importance of seeing transitional justice as a process, rather than a short-term enterprise, highlights 'the possibility of a continuous return' as the past returns in new and unexpected guises 38 and the need for a 'deeper, broader, longer transitional justice'. 39 Insights into prioritisation and sequencing, and their role within theories of change, are likely to be highly contextual and as a result to resist easy generalisation.
This section outlines some problems that need to be resolved in developing theories of change for transitional justice. It illustrates both the positivistic, output focussed nature of much theory of change literature and practice, and the challenges associated with a lack of theorising within transitional justice. These challenges include: the need to operationalise concepts that are often vague in definition and contested in practice; analysing the relationship between intermediate goals, such as truth-telling and prosecution, and ultimate aims, such as democratic strengthening and peace; articulating how different levels of impact (individual, community, national) interact; understanding the impact and interaction of purportedly complementary mechanisms within transitional justice, and between transitional justice and related fields; and managing the prioritisation and sequencing of interventions in complex and evolving processes of change. In practice, such questions are in part answered by impact studies and evaluations, and the next section turns to this growing field as a route to understanding processes of change.

IMPACT AND EVALUATION OF TRANSITIONAL JUSTICE INTERVENTIONS
This section looks at both quantitative and qualitative studies of the impacts of transitional justice, including evaluations. The heart of the dilemma here is that such studies have produced mixed, even contradictory, findings, used various methodologies, and at present there is neither a coherent evidence base nor clear policy guidance to support transitional justice interventions. 40 Quantitative studies of transitional justice are the most explicit about outcomes because they must create indicators onto which to map goals, including macro-level ambitions, and are thus less ambiguous about goals than many qualitative studies. Single context studies have included goals such as reconciliation, 41 and reducing retributive desires, 42 among their perceived aims. Large N studies, typically working across many tens of contexts, have made a choice of dependent variables, notably referencing human rights and democracy, 43 peace, 44 and repression of violations. 45 While these studies advance understanding of change processes, they have also been subject to significant critique. First, a narrow, Western view of key variables such as 'human rights' and 'democracy' is typically taken, with rights understood as an absence of gross personal integrity violations, 46 and democracy as political rights and/ or breadth of political participation. 47 Second, they remain vague about pathways to impact, or causal processes that link mechanism and outcomes. Where such pathways are identified they are typically linear and fail to link specific transitional justice interventions with the wider socio-political context of transitions, since a large N approach necessarily generalises across multiple contexts. As such, results are unlikely to provide models that coincide with the complexity in any one given transitional context. Third, even where correlation is demonstrated, causality remains elusive. Given the limits of purely quantitative analysis, it is likely that deeper causal insights into transitional change processes will require qualitative or mixed methods, and a new generation of such studies is indeed emerging. 48 At the heart of evaluation dilemmas is the question of who decides what the goals of the process are, which links to where power lies in the transitional process. Most national transitional justice processes still make no effort to measure their impact and as a result evaluation has been dominated either by academic studies seeking to measure impacts or by donors and operational agencies wanting to measure the impact of their own programmes to support transitional justice processes. The growth in volume and sophistication of evaluations is beginning to nuance both the methods used and the findings identified. 49 There are four main approaches to measuring the quality of transitional justice processes, each of which is informed by different methodologies. First, the most common approach is an impact assessment of particular mechanisms, typically truth commissions and trials, in terms of their impact on individuals. In many cases only the people who have directly interacted with a particular mechanism are studied, either those who testified and gave statements, those who were involved in the Commission, or NGO professionals, thereby overlooking wider societal impacts. 50 Second, process-based evaluative approaches measure the quality of the implementation of a process e.g. are methods of investigation appropriate to finding the truth, or is the intervention reaching its intended beneficiaries? 51 Third, outcome evaluation seeks to measure a process or mechanism against particular criteria, or in terms of desired goals. This has been attempted using multi-context quantitative approaches (as discussed above) against goals such as peace and democracy. 52 Other studies use mixed or qualitative methods to assess outcomes. 53 A final approach, formative assessment, 54 has sought to avoid such 'comparison to the ideal' 55 by targeting representative samples of victims, survivors, or the general population, to evaluate the quality of transitional justice against local perceptions of priorities and needs. 56 A limitation of all the above approaches is that, in the absence of a theory of change, they offer little insight into the causal drivers of the impact that is observed. Regression analyses with quantitative data can infer causality, but many such studies do not develop consistent theories of change that are supported by evidence. 57 Qualitative studies can capture the voices and opinions of stakeholders in ways that allow inference of casual mechanisms, but impact measurement is not always an explicit goal of such studies. The challenges facing evaluation overlap with many of those highlighted earlier in the article. For example, there is almost no longitudinal work which can track changes as transitional justice mechanisms unfold and demonstrate sustainable or evolving impacts. In very few cases have factors beyond the mechanisms of transitional justice been explored, to acknowledge that impacts can also result from the circumstances of transition -such as a clear military victory -the social and political environment, or from interventions originating in other sectors.
The challenges of evaluative approaches to transitional justice are demonstrated by studies assessing the impact of the South African TRC process, which reached highly divergent conclusions. For example, authors in Chapman and van der Merwe's edited volume used a range of qualitative and quantitative methodologies to evaluate the TRC process and notably its impact on victims and survivors. 58 In addition to condemning elements of the TRC's implementation, they conclude that it promised far more than it managed to deliver. The disillusionment of victims and survivors is further confirmed by Backer's longitudinal survey of 153 victims, which shows that approval of the amnesty process fell from 57.5% in 2002-3 to 20.4% in 2008. 59 Gibson, in contrast, draws on public opinion surveys, exploring the TRC's impact on all South Africans, to test the hypothesis that truth leads to reconciliation. 60 As such, the study prioritises reconciliation at the level of the general public as the phenomenon of interest, defined as interracial reconciliation, political tolerance, support for the principles of human rights and the legitimacy of political institutions. Gibson concludes that the TRC does appear to have moderated black and white views on apartheid and resistance to apartheid, bringing racial groups closer together, but it remains unclear -in the absence of any theory of change -to what extent this can be considered a result of the TRC rather than ongoing political changes. These studies demonstrate a basic but fundamental point: the result of an evaluation exercise will at least partly be determined by the question asked and how the evaluation is framed.
While new insights are emerging from impact and evaluation studies, these strands of inquiry provide no clear, or clearly theorised, direction of travel. They also currently provide insufficient evidence to support transitional justice interventions, or the expectations that should accompany such interventions. In the final section of this article an alternative route to developing clear and coherent theories of change is proposed.

A NEW APPROACH TO THEORIES OF CHANGE IN TRANSITIONAL
JUSTICE This article argues that using theories of change in transitional justice has significant potential to both understand and optimise impacts, but that realising this goal depends on adopting such theories in a critical and nuanced manner. The risk for transitional justice is that theories of change become another highly technical route fuelling a results agenda, 'the attribution obsession', 61 prescriptive responses and elite control. The vision set out here is of an approach to theories of change that goes against the grain, both of many such theories in fields such as international development and of current practice in transitional justice. Two elements are identified as central to this approach. The first is the issue of complexity and the use of theory to acknowledge and understand that transitional justice mechanisms are only one of many processes that shape transitional societies as complex systems. As a result, change is likely to be non-linear and multi-causal. The second is a focus on actororiented approaches which seeks to acknowledge that institutional or systemic change is only as relevant as its impacts on the everyday lives and needs of affected populations.

Complexity Theory and Transitional Justice
As has been seen in the development field complexity science offers the potential to generate greater insights into work on change processes, 62 and a means for providing an underlying theoretical framework for transitional justice as a tool for change. A crucial part of complexity theory is the idea of a system, characterised by interconnected and interdependent elements, and in which change arises from a multitude of relationships rather than from linear cause-effect chains. 63 Complexity also incorporates elements such as feedback, where the outcome of an activity can amplify or diminish change in a system, and emergence where the behaviour of a system emerges -often unpredictably -from the interaction of its constituent parts, such that the whole is different to the sum of the parts, and change is non-linear. 64 The emergent character of complex change is both driven and compounded by the influence of many different actors, with different perspectives and relationships, impacting upon social dynamics. This framework revisits the idea of holistic interventions, and challenges such as the prioritisation and sequencing of interventions. It acknowledges that a transitional justice mechanism will always be only one of many elements impacting on social actors in a transitional context. Perhaps the greatest test for transitional justice theory and practice offered by complexity theory is that the best 61 Duggan, supra n supra n 4 at 323. course of action will be highly context-dependent, challenging the very idea of a single practice relevant in all contexts and spaces. While there is relatively little discussion of complexity in transitional justice literature, 65 and a danger that indeterminacy in fact drives a retreat to technocratic solutions and toolkits, 66 three case studies drawn from secondary sources inform our call to embrace complexity: the ecological model of social reconstruction; incremental peacebuilding; and the restitutional assemblage. The first example is the ecological model of social reconstruction developed by Fletcher and Weinstein and Stover and Weinstein, 67 drawing on research in Rwanda and the former Yugoslavia. The authors question the role of individual criminal trials in furthering social repair and reconstruction in the context of communal violence. They argue that people use war crimes trials in such settings to reaffirm the collective innocence and the victimization of their particular group. By way of an alternative they propose that communal violence needs a communal response, which highlights collective guilt and responsibility for the violence. Such an approach would acknowledge and start to address the profound social breakdown experienced. An 'ecological' model of social reconstruction understands the community and society as a social system. All aspects (social, economic, political) of this system must be understood, as well as the fact that a change in one place produces change elsewhere within the system. Concretely, the authors propose a layered intervention that addresses individuals, families, communities and the state to ensure social reconstruction. Interventions include: state-level interventions; criminal trials (national or international); commissions of historical record (truth commissions); individual and/or family psychosocial support; externally-driven community interventions; and community-based responses, among others. 68 While the specific term is not used, this approach relates to complexity theory in a variety of ways. No single intervention seeks to address all of society's needs but each intervention is understood to have consequences on society as a social system, and synergies should be sought between levels and interventions to amplify positive effects. In specific contexts, the model illuminates the relationship between justice and social processes within communities affected by conflict.
A second case study draws on the work of Anna Lowenhaupt Tsing on friction as an ethnography of global connection. 69  Press, 2005), states: 'Cultures are continually co-produced in the interactions I call "frictions": the awkward, unequal, unstable, and creative qualities of interconnection across difference' (4). Her argument is that the universals of science, economic and social justice do not produce homogeneity but need to be understood through contingent practical encounters. 'Engaged universals' travel across distance and difference, mobilizing, changing and being changed (ibid.: 6-11).
rather than the outcomes of such processes. 70 Millar uses the term 'compound frictions' to describe the way in which 'a diversity of international peacebuilding interventions -each embodying different and even competing universal norms and paradigms -interact with one another in the minds and imaginations of local audiences to produce unpredictable expectations and experiences'. 71 Friction is 'nested' with complexity theory, echoing the distinction made by Rogers between complicated and complex interventions (see footnote 59), in arguing that complex systems cannot be understood simply by analysing their component parts, since interaction between these component parts produces outcomes that are unpredictable, generative and non-linear. Millar applies this conceptual framework to Sierra Leone, and the work of the Special Court for Sierra Leone, the Truth and Reconciliation Commission, DDR (disarmament, demobilisation and reintegration) and international NGOs, to facilitate insights into the locally experienced confusions and re-interpretations caused by compound frictions between multiple interventions carrying rival universalising ideas, when the interventions are imported into a particular setting. The implication of this approach is unsettling as it suggests that outcomes are unplannable: 'we cannot, in short, know what will happen, but we can know what might happen and from that set of possible future states we can choose a compass point by which to steer'. 72 Millar advocates 'incremental peacebuilding' 73 consisting of small calibrated actions, constant attention to shifting contexts, and slow steering towards desired outcomes. Finally, Reading's work on the Parramatta Girls Home, 74 a site of institutionalized female and indigenous containment in Australia, and the concept of 'restitutional assemblage' provides similar insights. She argues that restitution is more usefully conceived not as a 'one-off, discrete or bounded process towards a particular goal', but rather as 'an assemblage of practices that involve unfinished processes and interventions that operate across a number of domains'. 75 The assemblage of practices spans the material and economic, but also non-material realities (symbolic, affective/emotional, spiritual and cultural), and takes place across space/levels and time. Outcomes -inquiries, apologies, compensation campaigns, academic and cultural coverage, and more -are not finite but emergent, with new connections and priorities arising through processes of change. The restitutional assemblage suggests that everything does not and indeed should not be done simultaneously, but rather actions take place strategically over different time and space thresholds, recognizing that it is this ongoing process which is critical in terms of creating a sense of justice that is truly transformative.
While Fletcher, Stover and Weinstein help us to understand that transitional justice is itself complex and operates in countries and communities which are also complex social and political systems, Millar and Reading advance this analysis by combining complexity, or an 'holistic' approach, with guidance about prioritization and sequencing (see Table 1). Providing links to the actor-oriented approach, discussed below, each of the approaches set out above must be informed by the needs and views of the local population.
Towards Actor-oriented Evaluation Approaches in Transitional Justice Theories of change driven by concerned stakeholders -for example victims or affected populations -can create evaluation methodologies that are actor-oriented and holistic (see Table 1), in contrast to the programme-oriented approach of mechanistic, log-frame led evaluation. The key principles on which this approach is based are that understandings of justice and human rights emerge from everyday lives and the political struggles around them, not necessarily through alignment with national or international laws, 76 and initial goals are defined by concerned stakeholders who themselves drive the theories of change upon which evaluation is based. Such actors may also have unique insights that professionals lack. This approach differs dramatically from the way in which transitional justice is traditionally planned, challenging a purposive approach, based on fixed goals and targets, with one that is purposeful, based on agile measures of success, adaptable to changing situations (mirroring Millar's incrementalism). This perspective embraces evaluation as understanding, rooted in the stakeholders' values and the change they seek, 77 and responsive evaluation, 78 which implies a renegotiation of the relationships between the production of evaluative knowledge and the constituencies with which it engages. Because change is necessarily non-linear, accountability and attribution cannot be measured in a positivistic cause-effect chain, but rather the contribution of various factors to a particular change must be gauged subjectively by stakeholders. 79 Responsive evaluation lends itself to a longitudinal approach that can provide a feedback mechanism and reflexive monitoring to inform an unfolding process, thereby becoming not only a measurement tool but a potential approach to steering transitional justice. 80 Among the important elements of these approaches are an acknowledgement of multiple stakeholders, wide-ranging participation, a renegotiating of power relationships, and the importance of process and multiple, open ended goals. Examples of such responsive and constructivist evaluations in transitional justice are few. However, some work has been done, typically framed in the language of a 'victim-centred' approach, that echoes the idea of a transitional justice defined subjectively by concerned actors. 81 Participatory approaches and action research are methods that offer empowerment as a potential impact of evaluation. Two evaluative approaches that engage with these concepts, outcome mapping and the Most Significant Change (MSC) method, will be introduced below, with reference to their use -and potential use -in transitional justice. An example of a method based on actor-oriented theories of change, and that is implicitly constructivist in approach, is outcome mapping. 82 This approach seeks to identify changes in attitudes, knowledge, and behaviours as perceived by the direct beneficiaries of an action. Outcomes, and an understanding of how they emerge, are harvested from the actors most affected. A fundamental principle of the approach is to acknowledge complexity and seek to measure the extent of an intervention's contribution towards outcomes rather than attributing change exclusively to the intervention. While outcome mapping as a fully-fledged project management strategy in transitional justice remains rare, increasingly agencies are using the technique of outcome harvesting to access subjective perspectives on project impacts from concerned communities. 83 Outcome mapping has been used in transitional justice work by the International Coalition of Sites of Conscience (ICSC), a global network of initiatives using memory of violations to advance human rights and social justice. 84 A set of monitoring and evaluation questions focuses on how social actors move from memory to action and what role the Coalition plays in contributing to that process. 85 Evaluation questions address new behaviour, relationships, actions, policies or practices of individuals, groups, communities, organisations or institutions, through a qualitative engagement with the relevant social actors. This approach permits such actors -typically community members, victims, or youth -to describe the outcomes sought and to develop a theory of change in advance of a project, and to report on change during and after the memory action to allow a longitudinal measurement of impact. The ICSC experience of the use of outcome mapping is that it resonates with and reinforces participatory and community-based methodologies. The Global Initiative for Justice, Truth, and Reconciliation, of which ICSC is a part, has seen this in a context in Africa where conflict is ongoing, where it has sought to support documentation of violations with a long-term view of informing formal transitional justice mechanisms: 86 local partners who are collecting data have championed short-term projects such as local truth-telling and community dialogues which advance local peacebuilding as well as long term justice goals. Outcome mapping cannot of course ensure that different actors share the same goals of any process, and this limits the choice of engaged actors to those who share broad aims or depends upon the interactive process of discussion of such goals producing a set of priorities that is shared.
The Most Significant Change (MSC) technique is a further actor-oriented evaluation approach used for evaluating complex interventions, representing a qualitative participant-driven approach, focusing on the human impact of intervention. 87 It is often characterised as a transformative evaluation approach. In essence, MSC involves the generation of stories by various stakeholders involved in an intervention, where this could reference a NGO-implemented project, a particular transitional justice mechanism, or an entire transitional justice process. These are stories of significant changes in people's lives caused by the intervention, the most significant of which are selected by the stakeholders and used to facilitate in-depth discussions. These discussions bring to the stakeholders' attention the impacts of the intervention that have the most significant effects on the lives of those it targets. MSC has begun to be used by a wide range of development actors, 88 but has also been used in a number of transitional justice contexts. 89 Examples of MSC in transitional justice contexts have largely focused on local approaches, rather than national processes, since emphasis resonates with the goals of agencies which have led such interventions. For example, a conflict resolution project in the Democratic Republic of Congo was evaluated through the collection of 125 MSC stories. 90 These were analysed initially in terms of 'domains of change', which included ex-combatant integration, justice, community development and mobilisation, while additional analysis allowed the identification of indicators defined by members of concerned communities. This enabled the evaluation to demonstrate for example that communitybased 'peace courts' established by the project were highly valued and considered effective in resolving conflict. Beyond this, the MSC process itself has created a route of communication between communities and local authorities, and between communities and the INGO. In a challenge to claims that a lack of rigor deters donors from supporting its use, the US Department of State's Bureau of Democracy, Human Rights, and Labor (DRL) has used MSC in the human rights work it funds. 91 The approach permits DRL to 'understand intended and unintended changes, both positive and negative' and 'question which outcomes we were valuing most and how we were determining significance'. 92 MSC enabled DRL to interrogate power relations around how 'most significant ' was determined and to ensure that incremental successes, such as building acceptance or creating movements and alliances, were not neglected in the light of 'bigger wins' in areas such as policy reform.
What techniques such as outcome mapping and MSC facilitate is both the bottom-up development of theories of change by those most impacted by the change sought, and a direct, subjective measurement of that change. They are currently used largely by agencies seeking to evaluate their own programmes but have the potential to measure the effectiveness of transitional justice mechanisms at various levels. Such approaches face their own challenges, notably the willingness of donors and states to allow locally driven, more open ended outcomes, and downward rather than upward forms of accountability. While both approaches can be resource intensive, further experimentation and innovation has the potential to streamline methodologies and steer an incremental, responsive transitional justice.

CONCLUSIONS
Transitional justice faces numerous challenges when it comes to establishing theories of change, as it is characterised by diverse interventions, complex and contested contexts, and the need to balance principles and pragmatism. Normative claims, implicit or archetypal theories of change, alongside linear and mechanism-based arguments, remain dominant, while the evidence base for transitional justice is still weak. The result is invariably excessive expectations and a practice characterised by managing disappointment. This article argues that the development of theories of change could play a role in remedying this situation. This argument is made against the grain of both mainstream development theories of change, and mainstream thinking in transitional justice, in that it seeks to replace a focus on results, attribution, and linearity with a privileging of process, contribution and complexity.
This article looks at the potential to draw insights from adjacent fields, some of the challenges facing the development of theories of change within transitional justice, and evidence from impact studies and evaluations. While new insights are emerging, these strands of inquiry provide no clear, or clearly theorised, direction of travel. As such, this article argues for an approach to theories of change in transitional justice which unpacks and challenges the beliefs, assumptions and hypotheses about how change happens in transition. Drawing on complexity theory and actor-oriented approaches, it makes the case for theories of change which are evidence-based, stakeholder-led, and complexity-focussed. We are left with an important set of terms -systems, interaction, contingency, context, encounter, emergence, incrementalism -to inform what we have labelled evaluation as understanding. If expectations are co-constructed, and managed through reflective processes, the practice of transitional justice would become less fraught with conflict and disappointment, and more energised by new forms of legitimacy and accountability.