50303 Paper No. 117 I Jw,e 2009 Understanding Violent Conflict in Indonesia: A Mixed Methods Approach* Patrick Barron World Bank Office Jakarta Sana Jaffrey World Bank Office Jakarta Blair Palmer AustraHan Na tiona l University jWorid Bank Office Jakarta Ashutosh Varshney Brown University ,. This paper provides an outline of the methodology being used for the Violent Conflict in Indonesia stud y. The research is funded through a grant from the World Bank's Post-Conflict Fund. Additional funds have been provided by USAID, through the lRO-SERASl program. Chris Wilson and Adrian Morel provided substantive inputs into the research design. Thanks to Bruno Boccara, Samuel Clark, Paul Francis, Markus Kostner, Stephen Miller, Dave McRae, Wi lliam Wallace, Alys Willman, SUS,111 Wong and Matthew Zurstrassen (World Bank), Yuhki Tajima (Universi ty of California, Riverside), Sidney Jones (International Crisis Group), and Supryoga Hadi (Bappenas) for comments on earlier drafts of this paper. The views in this paper are those of the au thors alone and not of the institutions to which they are affi liated or of any of the funding bodies. SOCIAL DEVELOPMENT WORKING PAPERS Paper No. 117/June 2009 Understanding Violent Conflict in Indonesia: A Mixed Methods Approach Patrick Barron World Bank Office Jakarta Sana Jaffrey World Bank Office Jakarta Blair Palmer Australian National University/World Bank Office Jakarta Ashutosh Varshney Brown University * * This paper provides an outline of the methodology being used for the Violent Conflict in Indonesia study. The research is funded through a grant from the World Bank's Post-Conflict Fund. Additional funds have been provided by USAID, through the IRD-SERASI program. Chris Wilson and Adrian Morel provided substantive inputs into the research design. Thanks to Bruno Boccara, Samuel Clark, Paul Francis, Markus Kostner, Stephen Miller, Dave McRae, William Wallace, Alys Willman, Susan Wong and Matthew Zurstrassen (World Bank), Yuhki Tajima (University of California, Riverside), Sidney Jones (International Crisis Group), and Supryoga Hadi (Bappenas) for comments on earlier drafts of this paper. The views in this paper are those of the authors alone and not of the institutions to which they are affiliated or of any of the funding bodies. This paper has not undergone the review accorded to official World Bank publications. The findings, interpretations and conclusions herein are those of the author(s) and do not necessarily reflect the views of the International Bank for Reconstruction and Development/ World Bank and its affiliated organizations, or its Executive Directors, or the governments they represent. To request copies of the paper or for more information on the series, please contact the Social Development Department Social Development The World Bank 1818 H Street, NW Washington, DC 20433 Fax: 202-522-3247 E-mail: socialdevelopment@worldbank.org Printed on Recycled Paper Table of Contents I. Introduction and Background..............................................................................................................................1 1. The Violent Conflict In Indonesia Study ........................................................................................................3 2. Research Topics..................................................................................................................................................4 II. An Overview of the Methodology.....................................................................................................................5 1. Understanding Variation ..................................................................................................................................5 2. Why Are We Using Mixed Methods? .............................................................................................................5 3. How Are We Using These Methods? ..............................................................................................................6 4. Quantitative Data: the ViCIS Newspaper Dataset ........................................................................................7 III. Patterns, Forms and Impacts of Violent Conflict ........................................................................................12 1. Aims and Research Questions........................................................................................................................12 IV. Routine Violence...............................................................................................................................................14 1. Aims and Research Questions........................................................................................................................14 2. Research Methods ............................................................................................................................................15 3. Case Selection and Fieldwork ........................................................................................................................16 4. Comparative Framework................................................................................................................................18 V. Escalation of Violence........................................................................................................................................19 1. Aims and Research Questions........................................................................................................................20 2. Research Methods, Case Selection and Fieldwork ......................................................................................21 VI. De-Escalation of Violence ...............................................................................................................................24 1. Aims and Questions ........................................................................................................................................25 2. Research Methods ............................................................................................................................................26 3. Case Selection ...................................................................................................................................................27 4. Fieldwork and Analysis ..................................................................................................................................30 5. Testing the Theory on Different Types of Cases..........................................................................................31 VII. Testing Our Hypotheses: Links to Other Quantitative Data...................................................................32 VIII. Audience and Outputs ..................................................................................................................................34 1. Audiences..........................................................................................................................................................34 2. Outputs..............................................................................................................................................................35 References .................................................................................................................................................................36 Annex A: Concepts and Definitions....................................................................................................................42 Annex B: Provinces and Estimated Distribution of Media Sources..............................................................43 i Annex C: Coding Template...................................................................................................................................44 Annex D: Explanation of Codes ...........................................................................................................................44 Annex D: Explanation of Codes ...........................................................................................................................45 Figures: Figure 1: Case Selection Principles for Studying Routine Violence ...........................................................................16 Figure 2: Three different trajectories of violence: large-scale rioting vs. medium-scale rioting vs. low violence ......22 Figure 3: Within-case analysis of conflict trajectories.................................................................................................22 Figure 4: Violent conflicts in Aceh: January 2005-January 2009................................................................................25 Figure 5a Different scale of de-escalation in two districts...........................................................................................27 Figure 5b Different speed of de-escalation in two districts .........................................................................................27 Figure 6: Different levels of routine postconflict violence ..........................................................................................29 Figure 7a: Rise in vigilantism in post-conflict period..................................................................................................29 Figure 7b: Rise in land conflicts in post-conflict period..............................................................................................29 Figure 8: Within-case analysis of de-escalation ..........................................................................................................30 ii I. Introduction and Background Violent conflict in Indonesia is in need of serious theoretical and policy attention. A new belief that conflict has de-escalated in Indonesia has crept into popular and policy circles. However, it is not clear whether the movement towards de-escalation is cyclical or permanent. Nor is it clear that newer forms of conflict will not erupt in Indonesia. Comparative theory and evidence indicate that violence often reappears in areas that previously had acute conflict. Theory also suggests that unless suitable institutions or policies are imaginatively devised and put in place, a multiethnic or multireligious society is vulnerable to the possibility of long-run violent conflict. A careful examination of Indonesia's recent history of conflict, and forms and patterns present today, is vital for ascertaining current risks. As the Indonesian government and society seek to consolidate the democratic gains of the past decade, understanding violent conflict is of upmost importance. By now, the literature on conflict in Indonesia is quite substantial and many elements of the story are reasonably clear.1 The fall of Suharto was accompanied by the outbreak of intense group violence in several parts of the country. As a result, and in dramatic contrast to studies of Indonesia during the late New Order when the literature emphasized order, stability and economic dynamism, conflict became an important concern in scholarly and policy circles. The literature that emerged has especially advanced our understanding of some large-scale conflicts ­ in Aceh, Kalimantan, Sulawesi and the Malukus. Yet there are limitations to the existing research on Indonesian conflict. Four are particularly worthy of note. First, the Indonesian materials have remained by and large unincorporated into the larger theoretical and methodological literature on conflict. The scholarship on ethnocommunal conflict has made enormous advances over the last ten years, but Indonesia plays virtually no part in this scholarly effervescence.2 Very little is known about Indonesia's conflict dynamics beyond a small circle of Indonesia specialists. Indonesia needs theory and, equally, conflict theory needs Indonesian materials. The conflict dynamics in Indonesia, among other things, are likely to have relevance for those multiethnic and/or multireligious societies that used to have authoritarian political orders and have of late gone through a democratic transition accompanied by considerable group violence. Nigeria, post-Communist Eastern and Central Europe, and Central America easily come to mind, but the list can be expanded. A creative engagement with theory and comparative experience nearly always illuminates uncharted dimensions of a problem, inaugurating newer ways of thinking and, in some instances, suggesting new policy and project interventions. Second, the emphasis in the literature has been virtually entirely on the colossal episodes of collective violence, especially in the Malukus, in Central Sulawesi, and the May 1998 riots in Jakarta, as well as the war in Aceh. This focus is understandable in light of how horrific these violent episodes were. However, it results in several serious limitations. The literature has more or less ignored routine acts of violence, such as fights over land or vigilante justice, which appear to be common in some parts of Indonesia. These have not been systematically studied despite their potential policy importance. If these forms of 1 Although, for the most part, the policy implications of the existing research are unclear. 2 For an overview, see Varshney (2007, 2008). 1 violence cumulatively have serious human security impacts, or if they are a precursor to larger outbreaks of unrest, an important part of the picture is missing.3 Third, the methods by which the large-scale cases of violence have been studied have led to incomplete explanations. One stream has focused on the structural conditions that lead to, or allow for, violence. Books by Bertrand (2004), Sidel (2006) and van Klinken (2007a) all take multiple case studies and look for commonalities to determine causal factors. These scholars may well be right about the causes of violence, but without a comparison with peaceful cases, they cannot, in principle, be sure that the causes of violence they have identified are indeed the right ones.4 For a causal theory to be right, it is not only important to identify what is common across the many episodes of violence, but it is also critical to demonstrate that the causal factors associated with violence are absent in peaceful cases.5 Studies based on the commonality of outcome (or unvarying values of the dependent variable) can certainly allow one to build a theory, but as King, Keohane and Verba (1994) have argued, such studies cannot give us an adequately verified theory. For that, we need variation in the research design.6 Finally, there is almost no systematic information available at all on the post-2003 years of conflict ­ its forms, causes, and trajectories.7 Varshney et. al. (2008) have put together a database for the United Nations Support Facility for Indonesian Recovery (UNSFIR) which records incidents of large-scale violence for the period 1990-2003; Barron and Sharpe (2005, 2008) have created a dataset on small-scale violence in Flores and parts of East Java for 2001-2003. Yet relatively little information exists on forms of conflict since 2003.8 These data are critical for understanding conflict and its impacts in Indonesia. With 3 There are some exceptions. Lynching is perhaps the best-studied of the forms of routine violence. See Welsh's (2008) analysis of lynching in four provinces; Vel's (2001) on Sumba, and Herriman's (2007) on the `witchdoctor' killings in East Java. The edited volume by Colombijn and Lindblad (2002) contains some research on `everyday' forms of violence. The World Bank's Conflict and Development program has also analyzed `local conflict' in Lampung (Barron and Madden 2004; Tajima 2004), and Flores and East Java (Clark 2005; Barron and Sharpe 2005, 2008). 4 One other type of research ­ large-n in inspiration ­ ought to be noted. Barron et. al. (2009) and Mancini (2005) use survey data to determine factors associated with conflict propensity. This sort of work does cover variation in the dependent variable, but as is true of large-n work in general, it is unable to identify the mechanisms through which the independent and dependent variables might be connected. 5 We say more on this later. See also Varshney (2007) and Aspinall (2008). 6 On the whole, a research design based on comparing similar episodes is useful in theory building, not in theory testing. Under one condition, however, theory testing is also possible through this method. If a theory is deterministic, not probabilistic, then even one case, let alone a few, where violence takes place in the absence of factors identified with violence, is enough to invalidate the theory. Karl Popper's famous example is relevant here: any number of white swans that we observe will not prove that all swans are white, but one black swan can prove that not all swans are white (Popper 2002). The Popperian observation, it should be noted, does not apply to probabilistic theories, which theories of violence, along with a lot of other social science arguments, tend to be. In a probabilistic scheme of things, one black swan could simply be an outlier.. 7 One exception is the Potensi Desa (PODES) survey conducted by the Government's Bureau of Statistics. The 2005 survey contained a question on the incidence and impacts of conflict, for all Indonesian villages. While the 2002 PODES data has been used (Barron et. al. 2009), no-one has yet analyzed the 2005 data. Though the scale of the PODES (it is implemented in every village in Indonesia) is impressive, the fact that it collects data at a single point in time prevents analysis of how conflict evolves over time, and there may be reliability issues, given incentives for respondents (primarily Village Heads) to over- or under-report conflict. Nevertheless, where advisable, we will use the most recent PODES data to supplement our other data collection methods. 8 Aceh is the one exception. Here, the World Bank has been monitoring conflict incidents reported in local media since the tsunami (e.g. World Bank 2008). Some case evidence (e.g. van Klinken 2007b) and reports by the International Crisis Group also provide information. But these have not compared current conflict incidence and 2 the massive decentralization initiative, a whole host of new institutions have come into existence, altering the sites, group incentives and dynamics of conflict. It is important, and a priority of the Indonesian government, to build an empirical base that allows for consideration of conflict patterns and trends in the post-decentralization era. 1. THE VIOLENT CONFLICT IN INDONESIA STUDY The Violent Conflict in Indonesia Study (ViCIS) is a new World Bank project aimed at plugging gaps in the literature and adding to popular knowledge on violent conflict in Indonesia. The study aims to help the Government of Indonesia and others to formulate programs and policies to promote peaceful development and effective violence prevention. It seeks to bring a marriage of Indonesian conflict materials with the comparative theories of ethnic and communal conflict; it focuses attention on the widespread routine violence in Indonesia; it explores, within a broad comparative framework, how small clashes are transformed into large episodes of violence; and it investigates the process of de-escalation, asking whether Indonesian conflict de-escalation is likely to represent a permanent decline, or if there is evidence to the contrary. Finally, it aims to put together a comprehensive database of violence, updating and deepening the UNSFIR dataset (Varshney et. al. 2008) to include local conflict, violent crime and conflict since 2003, using around 100 newspapers as sources.9 Having started in mid-2008, the project will last for roughly three years. ViCIS builds upon and extends previous research conducted by the World Bank, other development agencies, and scholars on conflict in Indonesia. Since 2002, the Conflict and Development program of the World Bank has produced a number of studies on local conflict in Indonesia and its interaction with 10 development projects and processes. This research led to the formulation of a program of support to the Aceh peace process, and has also influenced the design and refinement of two large government programs funded in part through World Bank loans and credits: the KDP/PNPM community development program, which operates in every village in Indonesia, and the SPADA program which supports local governance and development in Indonesia's poorest and most disadvantaged areas. UNSFIR, with support from the United Nations Development Programme, created the conflict dataset discussed above. The new project will draw upon evidence, theory and methodological techniques developed in the prior work, as well as insights from the existing literature on conflict in Indonesia, and will complement this with fresh data collection and comparative analysis. patterns with those in earlier periods, making it difficult to know how serious violence is today compared with that of the immediate post-Suharto period. 9 See Annex A for definitions employed in the study. 10 This work has resulted in a large number of local case studies, including in "non-conflict areas" (Barron and Madden 2004), work on developing conflict typologies and conflict mapping techniques (Barron and Sharpe 2005, 2008), and evaluations of the impacts of projects on local conflict (Barron, Diprose, and Woolcock 2006). These and other papers are available at www.conflictanddevelopment.org 3 2. RESEARCH TOPICS The project will have four main components:11 Patterns, incidents and impacts of violent conflict ViCIS will provide quantitative evidence on the forms of violence prevalent in Indonesia, their impacts, and how these have varied over time. This will also allow us to identify the geographic distribution of violent conflict, its forms and impact in different areas, and will help answer questions about the extent to which it is concentrated in a limited number of areas or is distributed more widely. Routine violence A major focus of the project will be on `routine' forms of violent conflict, such as lynchings, land conflicts, and local political violence. ViCIS will help us map which forms are the most prominent in Indonesia, what their collective impacts are, and where they are concentrated. Qualitative work will focus on establishing why levels and impacts vary between areas, and on understanding the processes by which small-scale conflicts and disputes acquire violent forms. Escalation of violence The project will provide new comparative evidence on how small-scale violent conflicts escalate into larger outbreaks of mass violence, and why some areas have been prone to this while other have not. De-escalation of violence There has been little attention paid to forms and levels of violent conflict in the `postconflict' areas of Indonesia. ViCIS will provide insights into how forms of violence evolve after large-scale conflicts formally end, and why some areas become peaceful relatively quickly while in others sporadic violence continues. The analysis will help identify the extent to which these areas remain vulnerable to further outbreaks of violence, and the forms of intervention that can help ensure that peaceful conditions consolidate. 11 Fuller discussion of the research questions, and methods to be utilized, under each of these topics is given in Sections III-VI. 4 II. An Overview of the Methodology 1. UNDERSTANDING VARIATION The study will combine quantitative and qualitative approaches. The fundamental premise of our methodology is that figuring out why there are variations in the outcome of interest ­ namely, violence ­ is the one of best ways to understand the causes of violence.12 Suppose X represents violence and Y represents peace. With some exceptions, most explanations of violence have so far taken the following form: if transmigration (a), income differentials between two ethnic or religious communities (b), and demographic imbalances in the local military or police units (c) are present in X, they have been treated as the causes of violence. Methodological discussions of the last fifteen years, inspired by King, Keohane and Verba (1994), have by now clearly established that this sort of causal reasoning is fallacious. Factors (a), (b) and (c) can cause X, if one can show that they are not present in Y, which represents peaceful cases. Conversely, if (a), (b) and (c) are present in both X and Y, but another factor (d) is present only in Y, not in X, then (d) will be the cause of why X is different from Y.13 We cannot have confidence in our theory of violence if we study only the violent cases. Rather, it requires studying appropriately chosen cases of peace and violence.14 That is one of the key implications of the principle of variation for the study of conflict (Varshney 2007). We need to avoid `selection bias' in qualitative research. 2. WHY ARE WE USING MIXED METHODS? Methodological arguments in the social sciences are increasingly headed towards the view that both quantitative and qualitative approaches have distinct utilities and limitations, an exclusive use of either approach can unduly confine the scope of analysis, and ideally the two should be combined (Gerring 2007). Large-n datasets, for example, typically allow two kinds of analyses: (a) identification of broad patterns and trends, and (b) establishment of correlations between `independent' and `dependent' variables. On the whole, if not always, large-n datasets are unable to establish causality, whereas qualitative research, by systematically looking at which events led to violence (`process tracing'), allows us to separate causes and effects.15 Of course, there are conditions under which large-n datasets can move beyond correlations. They can allow us to assign causality, if good `instrumental variables' can be identified. However, even under such 12 Though admittedly it is not the only way. 13 Assuming all else that may be relevant is identical. 14 However, we ought to note that the causal factors we consider in our study of variation must be significant in an analytical, not mechanical, sense. Take an example. Suppose in a study of murder, (a) stands for men, (b) for hatred between them, (c) for a knife, and (d) for handcuffs. Further assume that (d) is present in Y (peace) but not in X (murder). If we mechanically follow the logic outlined above, the absence of handcuffs (factor d) could be viewed as the cause of violence. Rather, factor (d) should be viewed as the "cause" of violence only in the specific context where (a), (b), and (c) are also present. Thus, interventions to mitigate violence might need to address (a), (b), and (c), as well as (d). 15 This is particularly true for research on violence, where the direction of causality can be impossible to determine (see Barron et. al. 2009). 5 conditions, we need qualitative case studies. Instrumental variables can give us a good sense of causal effects (what is the effect of X on Y?), but not of causal mechanisms (how did X cause, or lead to, Y?).16 Following this reasoning, the objective of the quantitative approach in this project will be to generate a usable large-n dataset, building on and supplementing existing datasets, which will allow for identification of trends in conflict types, forms and their impacts. Based on the empirical results of the quantitative work, targeted qualitative studies will be carried out to determine the causes of the most frequent and high-impact conflicts, and of differing patterns of conflict escalation and de-escalation.17 It should be noted that there is another way to proceed. Sometimes, it is said that case studies are good at theory building, not for theory testing, for which large-n datasets may be required (Gerring 2007). Our use of case studies above is not conceptualized in this vein. As of now, we do not plan to move from case studies to datasets: rather we will proceed from datasets to case studies. Our contention above is that our large-n dataset will establish patterns and case studies will establish causes underlying such trends. For example, it is possible that large episodes of violence are concentrated in cities, not villages. If so, our case studies will be aimed at sorting out why this is so.18 3. HOW ARE WE USING THESE METHODS? Tracking trends: establishing patterns of variance The large-n database will enable us to observe patterns of variance in the incidence and impacts of conflict at multiple levels. Identification of such patterns is not only necessary to answer the research questions posed in this study but also has serious implications for policy-makers who need to identify areas and issues most vulnerable to violent conflict. First, we will consider spatial variation in the incidence and impacts of conflict. The design of the database will allow identification of patterns at the regional, provincial, district (rural kabupaten and urban kotamadya) and sub-district (kecamatan) levels. We will be able to ascertain which regions, provinces, districts and sub-districts have been the major sites of violence. Second, temporal variation in the incidence of conflict will be traced across the eleven-year period (1998- 2008). This will allow us to understand which regions, provinces, districts and sub-districts have moved from peace to violence (and to what degree) and vice versa. Third, we will analyze variation in conflict types (religious, ethnic, resource conflict, etc.) and forms (demonstrations, riots, lynching, etc) across regions and across time. We will thus know which forms and types of violence have been prevalent where, and how forms and types have changed in different parts of the country. 16 For details see Gerring (2007, pp. 43-48). See also George and Bennett (2005). 17 A second use of our dataset is also possible. We can test whether some existing theories in the larger literature ­ for example, the ethnolinguistic fractionalization (ELF) argument ­ are applicable to Indonesia. This is not the main thrust of our project, but if we are able to test some preexisting theories this way, we certainly will (see Section VII). 18 We are sure about this use of our dataset, but we remain open to the alternative methodological route. Our case materials will inevitably generate some theories of violence. If the elements or factors they identify as causes can be measured well, and if our dataset already has relevant information, we may also subject our theories to a large-n theory testing. Before figuring out what theories will emerge, it is hard to be certain about whether we will be able to use our large-n knowledge for theory testing. 6 Nested case studies: theory building While the dataset will be used to identify trends of conflict in Indonesia, causes or causal mechanisms will be established by conducting targeted case studies based on patterns detected in the data.19 The project will adopt two case study approaches to determine the causes of violence. The first, involves matched-case comparisons. Cases exhibiting different levels of conflict (high, medium and low) in the dataset, or showing different patterns of conflict escalation and de-escalation, will be selected after controlling for some factors that we will identify later to detect causes or causal pathways. Comparisons will be made at multiple levels (regional, provincial and district level) to allow for the identification of causal mechanisms. It is entirely possible that different causes or mechanisms are at work at different levels of the polity. This possibility cannot be theoretically ruled out. The second entails looking at within-case variation. Variation of violence within a single case will be studied across time.20 For example, if a district exhibits an overall trend of high-violence but is not uniformly violent across time, the case study will be used to establish the mechanisms through which violence occurs at specific times in that particular district. For example, we know that Ambon and Poso used to be peaceful until horrific violence rocked the two cities in 1999. Over the last three years, Ambon has become quite peaceful, but Poso has witnessed a recurring pattern of violence, though not at the same level as in the 1999-2002 period. Within-case analysis can help identify why violence levels may have changed within each district over time. Theory testing The main purpose of this study is to build theory in order to ascertain causal mechanisms. Process tracing based on case studies nested in the dataset will make possible. However, as discussed above, we will remain open to the idea that our dataset may allow some preexisting hypotheses to be tested on a large-n template. Such hypotheses may be based on Indonesia-specific scholarship, or that emerging from elsewhere in the world. Depending on how good our newspaper-based dataset turns out to be, theory testing in this manner is conceivable.21 4. QUANTITATIVE DATA: THE VICIS NEWSPAPER DATASET The main quantitative data source will be a comprehensive newspaper dataset that records all incidents of conflict (violent and non-violent) and violent crime reported in local newspapers for twenty-two provinces over the period 1998-2008.22 The decision to employ this methodology has been taken after 19 Further discussion of the qualitative components of the study is given in Sections IV-VI. 20 For the purposes of this study, we define a case as being a geographic area. For different types of analysis, cases will be at different levels ­ provinces, districts, sub-districts. Our choice for the unit of analysis will depend primarily on the level at which variation is observed. When we discuss within-case analysis, we primarily mean looking at temporal variations within a single geographic area. 21 See the discussion in Section VII. 22 For definitions of the concepts of conflict, violence, and crime, see Annex A. For a list of provinces, see Annex B. We include violent crime in the database, in addition to conflict (which is our primary focus), for three reasons. First, knowing whether there is a high level of deaths from violent crime is important in understanding the net human security impacts of violence. Studying violent conflict but neglecting violent crime would give a partial view of security impacts. Second, violent crimes sometimes play an important role in conflict escalation. Collecting data on violent crimes will thus allow us to study this aspect of escalation. Finally, there is a methodological reason, in that newspaper reports often make it difficult to determine whether a violent incident was 7 considering the limitations of several other options. Household surveys are weak at measuring conflict incidence and impacts, as they tend to record perceptions of conflict and have a tendency to underreport because (violent) conflict is a generally rare event that does not affect all in a community. Key informant interviews, as used by the PODES survey, create perverse incentives to under- or over-report conflict depending on the expectations about how the survey results will influence policy decisions and resource allocations (Barron et. al. 2009). Furthermore, survey methods rely on the memory of respondents and are hence less reliable for recording the details of older incidents of violence, making it difficult to create time series data. A comparison of police, hospital and NGO sources with newspaper data also reveals that there is systematic under-reporting of violent impacts, especially fatalities, as police and hospital data only include cases that are reported to the police or victims who are admitted to hospitals.23 Furthermore, these records do not contain the level of detail that would allow for a distinction to be made between incidents of conflict and those of forms of violent crime such as assault and arson. In contrast, the newspaper monitoring methodology has been shown to be effective in both high and low conflict regions (Barron and Sharpe 2005). UNSFIR showed that building a national dataset recording conflicts reported in newspapers was both possible and useful. Since August 2005, the local newspaper methodology has been used to monitor the Aceh peace process. Existing datasets have expanded our understanding of violence in Indonesia. Yet gaps remain. The new ViCIS conflict dataset will build on prior efforts in a number of ways. First, the dataset will expand on current spatial and temporal coverage. The UNSFIR dataset contains data on conflict in fourteen provinces for 1990-2003. Data collected since then through World Bank studies and pilots have developed more comprehensive methodologies for recording conflict data but have focused on a smaller set of regions for shorter time periods.24 The new ViCIS dataset will expand coverage by collecting data from 22 provinces, which cover 341 of Indonesia's 457 districts and 86% of Indonesia's population, and by collecting data from 1998 to 2008 (BPS 2007a, 2007b). This will enable us to trace trends since UNSFIR across and within more regions. Second, a larger set of sources of data will be used. The first iteration of the UNSFIR dataset collected data from national Indonesian newspapers. UNSFIR-II utilized provincial papers when it became clear that smaller conflict incidents were seriously underreported in national newspapers (Varshney et. al. 2008). Further studies, primarily based on sub-provincial level newspapers in lower conflict provinces (Barron and Sharpe 2005; Welsh 2008), demonstrate that provincial papers still miss certain forms of conflict: incidents of routine violence, such as lynching and land disputes, are reported in district level newspapers, but often not in the provincial media. An emerging conclusion about the Indonesian newspapers as sources of conflict data is that at different levels of coverage, newspapers differ in their the result of a conflict or a crime. Attempting to capture only violent conflict could mean excluding these incidents from the database, even though it may turn out that many such unclear incidents were in fact conflict. As a result, we wish to include all violence (including what appears to be crime). 23 A pilot conducted in Maluku and North Maluku compared deaths reported by newspapers, the police and healthcare providers between January and June, 2005. It found that newspapers reported 24 deaths, police recorded only twelve, UN Incident Tracking found 17 deaths, and the Maluku Interfaith Association recorded only four deaths. Hospital records recorded only one death in Ambon, compared with the seven reported in newspapers (for Ambon). The level of under-reporting was most pronounced outside of the provincial capital (Sharpe 2005). 24 These include the KDP and Community Negotiation dataset (for Flores and parts of East Java; 2001-2003 ­ see Barron and Sharpe 2008), a newspaper conflict monitoring pilot in Maluku and North Maluku provinces in 2005 (Sharpe 2005), and the Aceh Conflict Monitoring Updates (2005-ongoing ­ see, for example, World Bank 2008). 8 perception of which conflicts are newsworthy. While large episodes of violence are reported by the national newspapers, lynchings are better covered in the district level newspapers (Varshney 2008). Our study builds on these findings by using an estimated 57 district-level newspapers in addition to 42 provincial papers. This will provide a more accurate count of conflict incidents and their violent impacts.25 Third, the new dataset will improve on prior efforts by expanding the range of incidents included, and by developing a more detailed coding system that allows for more extensive disaggregation of data. UNSFIR-II, the most ambitious project to compile quantitative data on conflict in Indonesia to date, focused almost exclusively on large incidents of communal violence. The new dataset will also include local violent incidents between individuals and forms of violent crime. The coding system, developed for our proposed database, expands the analytical categories by allowing disaggregation of both violent and non-violent incidents, by conflict types (resource, administrative, religious, ethnic and political, etc.) and conflict forms (demonstrations, riots and group clashes, etc.). A broader classification of actors, interventions and impacts will enable us to capture detailed information about conflict dynamics in Indonesia.26 Phases of the newspaper study While newspapers appear be the best source of data for mapping conflict patterns and trends in Indonesia, they are not without weaknesses. The study anticipates the following problems and strategies by breaking the data compilation process into three stages. 1. Addressing biases: media assessments Before selecting specific newspapers for data collection in each province, exhaustive media assessments will be carried out to profile existing provincial and district newspapers. The staff of newspapers will be interviewed to assess the following: · Coverage. Even if we use provincial and district level newspapers, it is likely that the reporting coverage will be uneven across districts and sub-districts. By compiling information about areas where each newspaper has permanent offices, permanent reporting staff and free-lance reporters, we will be able to: (a) select newspapers with the best regional coverage for data collection; (b) supplant a weaker-coverage paper with others in that region; and (c) identify the strong and weaker sections of our data, even if a newspaper with limited coverage is selected. · Accuracy of reporting. Accurate reporting, especially as it pertains to assessment of impacts (deaths, injuries and property damage), is crucial for the validity of our data. It is likely that some newspapers do not emphasize accurate collection of facts prior to publication of incident reports. Gathering information about newspapers' sources of information and their policy on fact-checking, will enable us to select newspapers with high standards of reporting and to establish how accurate our data is likely to be. 25 For the estimated distribution of newspapers across provinces, see Annex B. 26 Bertrand (2008) has argued that expanding the range of incidents included will lead to a lack of analytic clarity, because the forces driving large-scale violence will probably differ from those leading to smaller-scale unrest. However, the coding categories employed mean that it will be possible to disaggregate different types of violence (large-scale communal, localized, violent crime, etc.). This will allow for consideration of the different causal factors and processes that lead to different outcomes; it will also allow for Bertrand's hypothesis to be tested. 9 · Reporting biases and censorship. There is widespread consensus that the press in post-Suharto Indonesia is relatively free. However, previous research shows that self-censorship in editorial policy due to the SARA legacy of the New Order or to prevent conflict escalation,27 and `envelope' journalism, where newspapers are sponsored by certain local groups or individuals and become advocates of those parties, still exist (Barron and Sharpe 2005). Assessing the institutional and personal biases in conflict reporting is vital for ascertaining the accuracy of the dataset. Extensive interviews with newspaper staff, eliciting responses about not only their own reporting standards but about the reputations of other papers in an area, will help us evaluate the strengths and weaknesses of our data and how it can be analyzed.28 · Completion of archives. Aside from conducting interviews to gauge the accuracy of reporting, the media assessment process will also obtain an accurate count of all existing archives in every province. By aggregating this information from the field, we will be able to: (a) select newspaper with the most complete archives for the eleven-year period of the study, and (b) identify potential gaps in our data to help strategize how to overcome them. 2. Data collection and coding Data collection and coding will be the most crucial parts of the quantitative study. Field teams will collect reports of conflict and violent crime incidents from local and provincial news sources and send this raw data to Jakarta. A standardized coding template will be completed for each reported incident and information about location, date, conflict type, form, actors involved, and violent impacts will be coded.29 The coded data will be subsequently entered into a searchable database. To ensure accuracy and uniformity of collection and coding, the following measures have been taken: · Training. A team of researchers has gone through a six-day training program to learn the concepts of violent and non-violent conflict and violent crime, as defined in this study, and how to map and collect data. A separate training has been conducted for the coding staff to teach them the nuances of the coding system developed for the ViCIS dataset. · Quality control. Given that the definitions of conflict and crime used in this study are complex, there are bound to be errors in the selection of articles in the field. We have developed systematic quality control procedures that will enable us to monitor the number of mistakes being made in the field in real time, so these can be corrected on an ongoing basis and additional training can be provided if necessary. Procedures will also allow us to identify the newspapers and articles that were subjected to quality control in the field so that they can be re-checked for the reliability of the quality control procedures themselves. Ten percent of coding templates will be randomly checked for accuracy and uniformity. 3. Testing the accuracy of compiled data After the compilation of the database, the final accuracy of our data will be tested by: · Cross-checking the data with the PODES survey; · Comparing collected data with the UNSFIR-II database for the period 1998-2003; 27 On ethnocommunal issues, the New Order government had a so-called SARA policy. SARA was an acronym for ethnic (suku), religious (agama), racial (ras), and inter-group (antar-golongan) differences. These differences were not to be discussed in the public realm. 28 For example, if a newspaper acts as a mouthpiece for a politician, it cannot be used for an accurate count of conflicts related to elections. However, it but may still be useful for reports on lynching, etc. 29 For the coding template and an explanation of the codes used, see Annexes C and D. 10 · Checking and augmenting the newspaper data with other sources such as police reports, court documents, and NGO reports; · Comparing the quantitative newspaper data with results of the qualitative case studies; and · Presenting results to peers for regular feedback. 11 III. Patterns, Forms and Impacts of Violent Conflict A primary focus of the study is to describe the overall patterns and trends of violence in Indonesia today, and over the past eleven years. This will address several of the key gaps in our understanding of violence in Indonesia. The newspaper database will provide the most comprehensive quantitative description to date of patterns of violence across regions, and trends in violence over time. 1. AIMS AND RESEARCH QUESTIONS In this component of the study we will seek to answer a number of questions:30 1. What have been the cumulative impacts of violent conflict in post-Suharto Indonesia?31 · Patterns of temporal variation in impacts; · Patterns of spatial variation in impacts (in particular, is there a concentration of violence in a small number of regions?) 2. Which types of violent conflict have had the largest impacts? · Patterns of temporal variation in the types that have been the most frequent or had the highest impacts; · Patterns of spatial variation in the types that have been most severe (in particular, which types of violence are severe in which places?) 3. Which types of violence have the largest impact per incident (i.e. the most fatalities per incident)? · Patterns of temporal variation in the most deadly types; · Patterns of spatial variation in the most deadly types (in particular, which types of routine violence are deadly in which places?) 4. How does urban and rural violence differ? Do they have different impacts? Are their forms different? Are they equally deadly? 5. Which actors are most likely to be involved in violent conflicts, and in particular in the deadliest ones? Are there variations over time and space? 6. Which weapons are used most frequently in violent conflicts, and in deadliest conflicts? Are there variations over time and space? 7. What are the variations in non-violent conflict types and incidence? How is non-violent conflict related to violent conflict? For example, if there is more non-violent conflict, is there generally more violent conflict also, or is it the other way around? 8. What are the gender dimensions of violence? For example, what is the role of women as victims, and as actors, in violence? Are women special victims of specific types of violence?32 30 The following list is not exhaustive. The database will also provide a host of other information (see template and codes, attached in Annexes C and D). Besides being of use to the current study, the dataset will be available for use by other researchers. 31 As noted earlier, the database provides information on immediate impacts of violence, such as deaths, injuries, and property damage, not on longer-term economic and political effects, which may be studied through other means. 32 Newspaper databases can not generally reveal much about forms of violence such as rape and domestic violence, due to both under-reporting and editorial priorities. Although we can not thus expect comprehensive or accurate data on violence against women from this database, it will provide some information on gender aspects of violence, which may be followed up through qualitative work. 12 9. Who are the victims in the most violent or most deadly types of violence? How does this vary over time and space? 10. Which cleavages most commonly drive violent incidents ­ ethnic, religious, tribal, political, or indigenous/local? · Which cleavages are present in the deadliest incidents? · Does Indonesia have what might be called a `master cleavage' like the Hindu-Muslim cleavage in India, the Malay-Chinese cleavage in Malaysia, the Sinhala-Tamil cleavage in Sri Lanka, or the racial cleavage in the United States?33 33 A `master cleavage' is one which, for a whole variety of historical reasons, is a primary, if not the only, determinant of politics and violence. 13 IV. Routine Violence Routine violence is one of the most glaringly neglected aspects of the current scholarship.34 We define routine violence as: "frequently occurring forms of violence (such as the beating of suspected thieves, inter-village brawls, or fights over a plot of land) which are not part of a large or widespread conflict, and where the impacts of single incidents are typically low (less than five deaths)". Such incidents involve local actors struggling over local issues, rather than large-scale mobilization by identity characteristics (such as ethnicity, religion or region). Routine conflict does not have to be violent; it can take both non-violent and violent forms. Examples include demonstrations, protests, petitions and group mobilization. On the whole, such expressions of grievance are quite healthy for a polity. Freedom of expression in a pluralistic society is inevitably accompanied by such legitimate modes of politics. These non-violent forms should be separated from incidents of routine violence. Both non-violent and violent routine conflicts are important for understanding the dynamics of violence in Indonesia. There are several important justifications for the study of routine violence. First, small but frequent violence can exact a big toll. Although fatalities tend to be limited in each incident, the total number of those killed through routine violence can be large, if such episodes are common or frequent (Barron and Madden 2004; Barron et. al. 2009; Barron and Sharpe 2008; Welsh 2008; Tadjoeddin and Murshed 2007). Second, such small-scale violence has serious systemic consequences. If some regions of a country develop a tradition of lynching ­ a mob killing a suspected culprit instead of handing him over to the police or administration ­ it impedes the growth of the rule of law. Moreover, if the frequency of such acts is high, it deadens popular sensibilities, arguably creating greater acceptance of large-scale violence as well. Finally, sometimes small incidents initiate a process that leads to huge conflagrations. Often, if not always, the starting point of a big episode of violence is a small clash between two groups or individuals. If we develop a better understanding of why small acts of violence occur, especially if such violence is frequent and widespread, and of which people or groups are in conflict in these forms of violence, we can perhaps generate a policy-relevant theory that can identify the institutions and strategies relevant to minimizing the occurrence or limiting the effects of such violence. 1. AIMS AND RESEARCH QUESTIONS This component of the study will seek to answer a number of questions: 1. What are the overall impacts of routine violence in Indonesia? 2. Which types of routine violence have the greatest impacts across the country? 3. Which types of routine conflict most often become violent, and which types very rarely become violent? 4. How does routine violence vary across Indonesia? 34 For example, routine violence is not a focus of attention in the three most recent book-length works on group conflict in Indonesia (Bertrand 2004; Sidel 2006; van Klinken 2007a). 14 · Are there some provinces or districts which suffer much higher impacts than others?35 · Are these the same regions which also suffer from large-scale violence, or are these regions typically thought of as "low-conflict regions"? 5. How does violence vary between urban and rural contexts?36 6. How has routine violence varied over time? · Are some forms increasingly prevalent now? · Are others on the decrease? 7. Why do some areas experience high levels of routine violence while others do not? 8. Why do some areas experience particular forms of violence while others experience different forms? 9. Why do cases of routine conflict escalate into violence in some places (and at some times) and why not in other places and at other times? 2. RESEARCH METHODS All but the last three of these questions can be answered through the newspaper dataset. The previous section described how the ViCIS database will provide detailed information on the types and impacts of violence in Indonesia, and its geographic and temporal variations. However, the database cannot tell us why these patterns vary across regions and across time periods, and why some types of routine conflict very often become violent while other types rarely do.37 No understanding of routine violence in Indonesia would be complete without attempting to understand variation in time and space, and the transition from non-violent conflict to violence. Insights into these questions will likely have significant policy implications, as Indonesia seeks to strengthen its peaceful democratic system in ways most appropriate to each local context. A series of qualitative case studies will be used to understand these `why' questions. Case studies will each focus on a particular type of routine violence, and will be designed to understand variation in its impacts across regions (and, later in the study, over time within each conflict case). Case studies are needed for the reasons described above: large-n work can easily identify correlations but not causal mechanisms. Process-tracing within case studies will thus be used to identify causal mechanisms. Comparing a certain type of routine violence across locations will require obtaining information about local structural conditions (demographics, institutions, politics, and so on) that will not be provided by our newspaper database. Finally, we want to understand the transition from non-violence to violence, but non-violent episodes are not consistently reported by newspapers. Qualitative work is thus needed to explore why routine conflict takes different forms and has different impacts in different places. 35 Regional comparisons of violent impacts will be conducted in both absolute and per capita terms. Each has its own merits for analyzing violence. 36 Understanding differences between urban and rural patterns of violence is important for theory, and for policy ­ especially as Indonesia becomes more urban, and given the ease of mobilization and escalation in urban environments, where heterogeneous communities compete over limited resources and often rely heavily on ethnic networks. 37 Although comprehensive comparisons have not yet been carried out, several studies have indicated that patterns of routine violence do vary between regions. For instance, lynchings appear to be far more common on the island of Java than elsewhere in Indonesia (Varshney et. al. 2008; Welsh 2008). Similarly, Barron and Sharpe (2008) show how violent land conflict is much more prominent in East Nusa Tenggara province than in East Java. 15 3. CASE SELECTION AND FIELDWORK The first step is to select which types of routine violence will be studied. Our attention will focus on the three types of routine violence with the highest aggregate impacts across Indonesia. Based on the previous literature, we expect that types of routine violence such as lynchings, land conflicts, and gang fights may be among those chosen. However, a decision on which types to study will be made after the database produces a clearer picture of violence impacts and forms. The next step is to select where we will conduct case studies. For each of the three types to be studied, nested case study comparison will be used, selecting provinces, districts, sub-districts, and individual incidents. Figure 1 lays out the plan. The strategy allows for comparative analysis at a number of levels to help identify causal processes at each. Figure 1: Case Selection Principles for Studying Routine Violence High Low PROVINCES violence violence DISTRICTS / High Low High Low SUB-DISTRICTS violence violence violence violence V V NV V NV V NV CASES V V NV V V NOTE: V represents violent and NV non-violent. Provinces First, two provinces will be chosen for each type of routine violence we are studying, one with a high level of impacts and one with a low level.38 Efforts will be made to control for other factors as follows: · Levels of non-violent conflict. Provinces will be chosen that have similar levels of non-violent conflict of the type being studied. In the case of land conflicts, for example, this means that we would select two provinces with similar numbers of land conflicts, but where the number of violent land conflicts is much higher in one. In the case of lynching, such identification will not be possible because lynchings are by definition violent. Because of this, we will use proxy indicators for factors that tend to lead to lynchings, such as the presence of theft, as a basis for selecting the control areas. (We are aware that this may be particularly challenging, given that newspapers tend not to report theft if it does not have a violent impact or when it is small in scale. As such, we will attempt to incorporate other data sources, such as police crime data, and local knowledge, to drive case selection). 38 By impacts here, as we have already stated, we primarily mean deaths. For certain types of routine violence, other kinds of direct impacts (injuries or property destruction) may be more prominent, and hence become a basis for selection. Other indirect impacts--such as effects on the economy, psychological impacts, and so on--will be analyzed in the case studies. However, these cannot be used as a basis for case selection because we do not have quantitative data on them. 16 · Structural factors. Provinces selected should not differ substantially across structural factors (economic levels, education levels, etc.). Which factors are most important will be determined later, with consideration to the type of routine violence being studied and the likelihood of structural factors being pertinent to it. · Reporting levels. After conducting thorough assessments of media coverage in each province, we will be able to rate the coverage which our database provides of the news in each province (see discussion above). It would be misleading to compare rates of violence in a province with minimal coverage to rates of violence in a province with excellent coverage. Thus in order to be confident of selecting provinces with different levels of lynching violence, they should have similar levels of reporting.39 Districts or sub-districts Four districts or sub-districts40 will be chosen for each type of routine violence: two in the `high violence' province and two in the `low violence' province. As with the provincial selection, the districts/sub- districts will be chosen in such a way that controls for exogenous factors that might affect the (reported) incidence of routine violence. Where possible, we plan to choose neighboring districts to control for higher level factors that may drive variation. Comparative analysis of the four districts/sub-districts will help us tease out which of the factors leading to variation exist at the provincial level, and which at lower levels of geographic specification. Incidents Particular incidents of violent and non-violent conflict (V and NV in the figure above) will be chosen in each of the districts/sub-districts concerned. We will likely over-sample violent cases.41 The number of incidents to be studied will be decided later based on resource issues and the insights coming from early case studies. Case histories will be developed by the research teams. Within-case process tracing will help us understand why some became violent and others did not, and how these factors relate to the structural conditions present in each place. In analyzing each incident, the following research questions will be useful: · Why did the conflict become violent? · Why did the police or other actors not prevent this violence? · Did the police arrest the perpetrators? · How do locals perceive this case and the way it was handled? · What structural factors are connected to this case? · Who were the actors, who were the victims? 39 Media assessments conducted prior to newspaper selection will enable us to identify levels of district reporting. 40 Whether we choose to compare districts or sub-districts will depend on where intra-provincial variation is most marked. This decision will be taken later after we have analyzed the basic patterns of violence from the newspaper dataset. 41 Within the low violence district/sub-district in the low violence province, we will not do any case studies of specific incidents (violent or non-violent). However, fieldwork will be conducted to see what structural factors might be driving the lack of violence in this area. 17 4. COMPARATIVE FRAMEWORK The comparative framework outlined in Figure 1 above will allow for a number of controlled comparisons including the following: · Comparison between cases with similar outcomes (high levels of violence) but different districts/sub- districts conditions; · Comparison between cases with similar outcomes (high levels of violence) but different provincial conditions; · Comparisons between violent and non-violent cases within high violence areas; · Comparison of two districts/sub-districts (one high violence, one lower violence) within high or low violence provinces. Selection of provinces and districts/sub-districts will be guided by one additional consideration. If the database has shown that a number of provinces or districts/sub-districts contain a particular concentration of violence, then efforts will be made to focus several of the cases studies on those areas. That is, locations for more than one case study ­ perhaps, lynchings as well as land conflicts ­ will be selected from those regions. This will enable us to focus extra attention on these high-violence places, across the various types of routine violence. Later in the data collection process (when earlier years of data are available), further case studies may examine temporal variation in routine violence within a particular geographic area. 18 V. Escalation of Violence An extensive literature has emerged on large-scale violence in post-Suharto Indonesia. However, we still do not have a good theory for why the small sparks of localized violence and tensions erupt into the large fires of inter-group collective violence. Developing such a theory is important for understanding not only the deadly outbreaks of communal violence in the past, but also (a) the potential for small-scale conflict and routine violence elsewhere in the archipelago to escalate, and (b) the scope for intervention ­ by the government and/or civil society. If, with the aid of theory, we can understand how to prevent sparks from becoming fires, perhaps one can also hope for fewer and less deadly violent conflicts in the future. Intercommunal ethnic or religious violence in West and Central Kalimantan, the Malukus and Central Sulawesi, separatist conflicts in Aceh and Papua, and the Jakarta riots of May 1998 have received the greatest attention of Indonesia experts.42 Initially most analyses focused on individual cases with few attempts at cross-case analysis. More recently, three scholars (Bertrand 2004; Sidel 2006; van Klinken 2007a) have written books on the broader issue of violence in Indonesia, examining multiple conflicts. Cross-case analysis has been used to develop frameworks to understand why different forms of conflict arose in different places at different times, concentrating largely on similarities in structural conditions that predated the outbreak of widespread violence. This new work has undoubtedly enhanced our understanding of the specific conflicts and has pointed to general systemic factors, all largely a product of Indonesia's post-Suharto transition. Yet the books also have a number of weaknesses that need to be remedied. The greatest problem is methodological. None of these works is based on the idea of variation in research design. All have concentrated only on episodes of violence, mostly on large-scale episodes, and none systematically compares why violence occurred in some places, not others. Bertrand (2004) studies violence in East Timor, the Malukus and Kalimantan; Sidel (2006) focuses on the burning of Churches in Java in the early to mid-1990s, the violence in Jakarta, and intercommunal conflict in the Malukus and Sulawesi; and van Klinken (2007a) concentrates on riots in Kalimantan, the Malukus and Sulawesi. Methodologically, the choice of areas in all three books are examples of `selection bias'. A good theory requires showing that the factors identified as causal in making violence possible were missing in places that did not experience violence.43 As discussed earlier, if we do not study peace and violence together, we cannot conclusively show which factors were really causal in producing either.44 42 See, for example: van Klinken (2001) and Wilson (2008) on Maluku and North Maluku; Acciaioli (2001), Aragon (2001) and McRae (2008) on Central Sulawesi; McGibbon (2004) on Papua; van Klinken (2000), Davidson (2008) and Smith (2005) on West or Central Kalimantan; Aspinall (2006, forthcoming), Schulze (2004), Sukma (2004), Barron, Clark and Daud (2005), Reid (2006) on Aceh; and Siegel (1998), Purdey (2006) and Mietzner (2008) on the Jakarta riots. The reports of the International Crisis Group (ICG), available at www.crisisweb.org, have also illuminated many of the conflicts. A number of edited volumes have brought together pieces of conflicts, often drawing parallels with historical patterns of violence in Indonesia: see, Tornquist (2000), Anderson (2001); Wessel and Wimhofer (2001); Colombijn and Lindblad (2002); Husken and Jonge (2002); Anwar, Bouvier, Smith and Tol (2005) and Coppell (2006). 43 Van Klinken (2007a) does develop a `vulnerability index' in order to compare the provinces of high violence to other provinces where large-scale violence did not break out. He identifies factors of rapid de-agrarianization and high dependence of the local economy on the state as important in differentiating high-violence and low-violence provinces. However, the focus of the book is not on establishing how these factors led to violence through a 19 Three more shortfalls are worth noting. These problems mark Bertrand (2004) and Sidel (2006), but not van Klinken (2007a).45 First, the comparative work has not fully considered the processes of escalation, which turned existent social tensions into smaller-scale acts of conflict to large-scale episodes of violence. Second, the explanations have largely been structural, and hence often rather deterministic, focusing on demographic shifts, economic balance, and changing access to political power, and have underplayed the importance of the processes of mobilization. Third, there has been an overriding emphasis on macro explanations for the outbreak of violence in certain localities. Bertrand (2004), for example, concentrates on differential group access to power in Jakarta, and their role in the Indonesian nation and polity, to explain why the Dayaks, Christians and Muslims rose up at certain points. This sort of approach can explain why violence gets clustered around certain periods (temporal variation), but it cannot help us understand why violence has geographically specific locations (spatial variation).46 For understanding the latter, we need to pay attention to micro or local factors in explaining violence (Aspinall 2008; Varshney 2002, 2008). Too often in the Indonesian literature, local-level conclusions have been drawn from national-level crises. Unless the local-national links are clearly established, such causal reasoning is flawed. A national-level crisis is, by definition, a constant for all localities, both violent and peaceful. It cannot explain both peace and violence. Logically, a constant cannot explain variation. 1. AIMS AND RESEARCH QUESTIONS The second component of the study aims to develop and empirically test causal explanations as to why small-scale conflicts and tensions escalate into large-scale violence in some cases and not in others. This can be used to increase our understanding of (a) why the large outbreaks of violence in Kalimantan, Sulawesi and the Malukus occurred47; (b) how they escalated; (c) what explains their timing, location, and the form they took; (d) the potential risk for escalation of local conflicts in other parts of Indonesia; and (e) potential areas for fruitful intervention by government, development agencies, civil society and the private sector to prevent future escalation of conflicts. Answers to these questions may also have applicability in conflict-prone areas in other countries. The research questions are as follows: comparison of dynamics in high-violence and low-violence provinces, but rather tracing the evolution of violence in the high violence provinces. 44 For a more detailed discussion, see Varshney (2008). 45 Van Klinken (2007a) does concentrate on escalation, processes and local dynamics, the three points we make below. However, his primary focus was not to isolate causal factors in order to develop policies for conflict mitigation, whereas this project does aim to do so. Van Klinken discussed one aspect of escalation for each of five big conflicts: Western Kalimantan, Central Kalimantan, Maluku, North Maluku and Central Sulawesi. As a result, we learn how Indonesian violence supports Tilly and his colleagues' conceptual categorization of the dynamics of contentious politics (McAdam et. al. 2001), and how elements of that theory can shed light on understanding violence in different provinces. But the lack of a comparative framework (even within high violence locations) mitigates against generating a broader understanding of why escalation occurs in some places and not in others. 46 Bertrand (2008) agrees with this point. 47 Van Klinken (2007a) has studied all three of these conflicts. However, as discussed above, he has focused on commonality of outcomes (large-scale violence), not outcome variation (violence and peace), as a way to build his argument. As a consequence, we cannot test his argument with his materials alone, even if the argument is right. 20 1. Given large, systemic and sudden shocks (such as the fall of the New Order and the financial crisis), how and why do local conflicts escalate into large conflicts in some places, and not in others?48 2. In the absence of large shocks, how and why do local conflicts escalate into large conflicts?49 3. How, and by whom, are groups mobilized? 4. In both contexts, what explains variation in incidence, timing and form? 5. What potential exists for the escalation of routine violence across Indonesia into large conflicts, and which areas are particularly vulnerable? 6. What policies and projects can help prevent future escalation? 2. RESEARCH METHODS, CASE SELECTION AND FIELDWORK Hypothesis generation Given that more has been written on large-scale violence in Indonesia than on local conflicts, a first step will be to commission a literature review of the existing studies. The studies of particular conflicts and some of the cross-case treatments cited above provide in-depth chronologies, including of the actors involved, violence triggers and hypothesizes causes. From these, along with other studies of large-scale violence, and especially riots, from around the world, we will distill hypotheses on potential factors and causal mechanisms that may be associated with conflict escalation. Case selection and comparative framework Qualitative fieldwork will then be employed. The hypotheses will be developed and tested in a number of ways. First, we plan to conduct structured, controlled and in-depth case comparisons of the sites of large-scale rioting (such as Ambon, Poso, Ternate, Sampit) with those that had very little violence (such as Manado, Palu and Yogyakarta). Controls will be specified later. Comparing the two will help elicit information on why violence escalated in some areas and not in others. Second, structured, controlled and in-depth comparisons will be made of the sites of high violence (e.g. Ambon, Poso) with those that had medium levels of violence (i.e., locations where violence rose but did not escalate beyond a point, such as Medan, Solo, Kupang, Lombok). Again, controls will be specified later. Figure 2 shows the types of comparisons that will be made. There are three hypothetical cases, consisting of large-scale rioting (such as in Ambon), medium-scale rioting (such as in Solo), and low violence (such as in Manado). The first kind of comparison involves comparing cases of large-scale rioting (the top line) with cases of low violence (the bottom line) for the first three months on the graph. The second kind of 48 We are interested in four different types of escalation here: (a) escalation from individual to group contention; (b) escalation from non-violent to violent conflict; (c) escalation in impacts; and (d) escalation of conflict forms and types, where less serious kinds of conflict change into more deadly ones (e.g. land to religious conflict). 49 This is particularly important to understand given where Indonesia is now. Economically, Indonesia has now fully recovered from the financial crisis (as measured in GDP per capita terms) and enjoys political stability not seen since the Suharto era. The recent global financial crisis could be a new shock, although current predictions are that Indonesia will continue to grow and poverty levels will continue to fall. 21 comparison involves the next two months of the graph (from month 3 to 5), where the top line keeps rising, while the middle line begins declining. Figure 2: Three different trajectories of violence: large-scale rioting vs. medium-scale rioting vs. low violence Large Scale Rioting Medium-scale Rioting Low Violence 60 50 Number of deaths 40 30 20 10 0 1 2 3 4 5 6 7 8 9 10 11 12 Months The analyses above will help identify the structural differences between areas that experience large-level conflict escalation and those that do not (or those experience escalation, but of a lesser extent). However, such analyses will not in and off themselves identify the triggers that led to conflict to escalate at a given point in time. A third form of analysis will thus examine varying conflict patterns within a case. This can help identify the turning points in conflict trajectories. For this, the newspaper dataset will be used to map patterns of conflict over time within a given geographic area. From this, points of heightened escalation can be identified (see the arrows in Figure 3). Fieldwork will focus on these points to see what was happening. Figure 3: Within-case analysis of conflict trajectories 45 40 35 Number of deaths 30 25 20 15 10 5 0 1 2 3 4 5 6 7 8 9 10 11 12 Months 22 Cross-area comparison of such `turning points' can then help us to ascertain the extent to which there are commonalities in the factors leading to heightened escalation. Finally, some of some hypotheses derived from the literature and fieldwork can be tested statistically. The conflict dataset we construct will contain information on conflict outcomes (over time). It can also be used to understand how different incidents oi conflict relate to each other ­ for example, if conflicts of a certain type (e.g. a lynching) at a given point of time tend to be associated with conflicts of a different type (e.g. an inter-group brawl) at a later point in time ­ 50 and will contain information on process variables (how escalation began, how it did, or did not, rise beyond a point; which institutions or organizations intervened to stop escalation, etc.). Other causal explanations may be more structural in nature, focusing on the social/demographic, economic or institutional conditions that tend to predict different patterns of conflict escalation. This data will not be captured from newspapers but will be taken from surveys conducted by the Indonesian Bureau of Statistics (BPS) such as PODES, SUSENAS, and various censuses. Section VII provides more information on how such statistical testing can be conducted. 50 Doing such `process analysis' is exceptionally difficult within quantitative datasets, even where information is collected continuously rather than at separated points in time. However, we think we will be able to do some such analysis. Separate conflict incidents within the database will be `linked' to other conflict incidents through the generation of a `Conflict ID'. This should allow for some process tracing analysis. 23 VI. De-Escalation of Violence In Indonesia and beyond, there has been little consideration of processes of de-escalation after episodes of large-scale violence have taken place, or of the conditions under which remaining tensions can re-escalate into new outbreaks of severe violence. The fourth component of the study will seek to develop theory explaining variations in the success of peace stabilization in areas that experienced massive unrest, the factors that explain the re-emergence of violent conflict in some areas and not in others, and why `postconflict' violence takes different forms in different areas. This will have implications for the design of postconflict programs and approaches in Indonesia and beyond. Areas that have experienced large-scale outbreaks of violence are prone to the resurgence of violence. Collier et al. (2003) have demonstrated that there is a significant chance of violent conflict re-emerging in areas where civil wars have formally ended, within five years of war termination. There are a number of reasons for this: signing a peace settlement does not necessary mean that conflicting parties, who may still see advantages in a future re- escalation of conflict, have fully `bought in' to peace; expectations over the benefits of peace may not be met; poor programs and policies in postconflict settings can create incentives for previously warring parties to pick up arms again.51 Conflicts also play a role in hardening identities and group cleavages, reconfiguring norms regarding the acceptability of violence in ways that take decades to overcome. Such factors, and others, can lead to the resumption of war in areas where peace agreements have been signed (Stedman, Rothchild and Cousens 2002). `Postconflict' areas can also experience new forms of violence (e.g. Rogers 2007; Chaudery and Suhrke 2008; Fortna 2008). In some cases, the human security impacts of such violence can be as great as those experienced during the initial period of war (Muggah 2009). In El Salvador, Guatemala and Nicaragua, for example, homicide rates are now higher than they were during the conflict period (Waiselfisz 2008). In Indonesia too, such `postconflict' forms of violence are also present. In Aceh, for example, the Helsinki peace agreement officially brought to an end a three-decade conflict between the Indonesian government and GAM, a rebel group. Yet while the peace process has by and large gone well, there have been rising levels of localized routine violence since the signing of the peace agreement (Figure 4). 51 See Muggah (2009) for a discussion. 24 Figure 4: Violent conflicts in Aceh: January 2005-January 200952 Source: Barron (2009) In Indonesia and elsewhere, relatively little is known about the forms of violence that emerge after peace settlements. This has negated from an understanding of the factors that lead to violence re-escalation. Most of the cross-country quantitative analyses, which aim to give causal explanations resumption of war, have implicitly treated violence as a binary variable: the lack of the reemergence of full-scale civil war is seen as a success. Patterns of postconflict violence are inherently important for understanding the potential for war or large-scale violence to restart. Yet, they have not been adequately incorporated into theories for why war resumes in some places and does not in others; Tilly's (1995) argument that understanding the `causes' of war and its reoccurrence are less important than developing deeper understandings of the nature of postwar violence has, to a large extent, not been taken up by researchers. Further, there has been relatively little study of how and why violence forms morph in postconflict settings, and how this negatively impacts on human security and stability (even without re-escalation to war or large-scale violence). Variations in the levels or forms of postconflict violence that have not escalated to full-scale civil war are not considered. In Indonesia, the treatments of the high conflict areas where violence has seemingly subsided (the Malukus, Kalimantan, Central Sulawesi, and now Aceh) have focused on initial actions taken to end the conflicts such as the Helsinki agreement (Aceh) and the Malino accords (Poso and Maluku). No-one to our knowledge has sought to systematically compare levels and forms of new violence and/or tensions since these conflicts peaked. Little data (quantitative or qualitative) has been collected to permit such comparisons. This lack of postconflict data makes it hard to analyze how conflict subsides, takes new forms, and (potentially) re-escalates. 1. AIMS AND QUESTIONS The fourth part of the study will look at issues relating to the de-escalation of conflict in areas that have been affected by large outbreaks of intercommunal violence in Indonesia. This involves mapping out levels of violence and/or tensions for a number of years after the larger conflict has ended and carrying 52 Note that the Helsinki MoU was signed in August 2005. 25 out more in-depth work to understand how these different patterns emerged. The results will have implications for understanding preemptive and responsive conflict resolution strategies in postconflict settings, allowing for the generation of policy recommendations applicable in both conflict and postconflict settings. The following questions will be addressed: 1. What patterns of conflict emerge after large-scale episodes of violence end, and how do these change over time? 2. What explains variation in the incidence and forms of postconflict violence?53 3. How do past patterns of violence shape the potential for conflict re-escalation down the line? 4. What is the relationship between patterns of postconflict routine violence and patterns of de- escalation? 5. Has the de-escalation of conflict in Indonesia acquired the properties of a permanent decline (a sustainable peace), or there is a real possibility of the reemergence of violence in some areas? 6. What policies and strategies can consolidate peace in both the short-run (the immediate period after agreements are signed or when conflict significantly de-escalates) and in the longer-run postconflict period? 7. To what extent are patterns of de-escalation similar in areas affected by intercommunal and separatist violence? 2. RESEARCH METHODS Quantitative and qualitative evidence will be used to answer these questions. The former will be used to identify patterns. This can then be used to select cases for more in-depth fieldwork. The quantitative dataset will be used to map violence by incidence, impact and form across those provinces that experienced high levels of communal violent conflict in the post-Suharto period: Maluku, North Maluku, West Kalimantan, Central Kalimantan, and Central Sulawesi.54 The key periods of concern are the years of high level violence and subsequent years. Importantly, both routine violence 55 and large-scale violence will be included (although they will be disaggregated). 53 This will include consideration of the different aid interventions, which may have contributed to conflict de- escalation (or which may have helped lead to new violence). This is particularly important within the World Bank given the emphasis on documenting lessons learned on different tools for postconflict areas (e.g. Kostner and Johnson forthcoming). Other work under the PCF grant involves evaluations to test the efficacy of reintegration programs in Poso and Aceh, conflict resolution training, psycho-social programs for traumatized conflict victims, and public information peace-building programs. Insights from these studies will help inform our analysis. 54 Aceh will be excluded from the initial analysis because of the different nature of conflict there. Likewise, our focus will not be on other areas that experienced only short-run violence, such as Jakarta. 55 In practice, distinguishing between large-scale ("big") violence and routine or small scale violence will not be simple. In many postconflict contexts, new forms of violence that emerge are often linked (directly or indirectly) to the initial conflict. However, our dataset contains a number of variables that should allow us to separate routine and large-scale violence, including conflict types, actors involved, and the conflict cleavage. 26 Such mappings will help us select areas for comparative study. However, in-depth fieldwork will be needed to understand causal factors that explain variation. For the escalation component of the study, many studies already exist, meaning that less fieldwork is necessary. In contrast, very little work on de- escalation has been done which focuses on recent years and the present day. Thus, it is likely that teams will need to spend a substantial amount of time in the field. 3. CASE SELECTION Cases will be selected to allow for a number of different comparisons to be made: · Variations in levels and temporal patterns of de-escalation · Variations in the extent to which new forms of violence are linked to old conflict-era forms · Variations in the types of postconflict violence that emerge Comparing different de-escalation patterns First, within the sample of five postconflict provinces, we will look for variation in levels of de-escalation. Such variation might be in aggregate levels (e.g. to a certain level of violence) or in patterns of de- escalation (e.g. in the speed at which violence has de-escalated). Figures 5a and 5b illustrate. In Figure 5a, district A has seen a de-escalation of violence as has district B. However, the scale of such de-escalation differs, being greater in district B. In Figure 5b, in contrast, both district A and district B end up with a similar level of violence. However, the speed of de-escalation differs between the two, as does the extent to which it is linear, being faster and more linear in district B. Figure 5a Figure 5b Different scale of de-escalation in two districts Different speed of de-escalation in two districts District A District B District A District B 60 60 50 50 Number of deaths Number of deaths 40 40 30 30 20 20 10 10 0 0 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Months Months Comparing such cases--through in-depth fieldwork and, to the extent possible, a review of the case literature--can help in the development of causal theories for why de-escalation patterns differ. Comparing levels of routine violence in postconflict settings We are also interested in the relationship between routine violence and forms of violence that were prevalent in the conflict period. Even where aggregate levels of de-escalation are similar across areas, varying forms of violence may emerge in different places. Some of these may have direct links to the forms of violence prevalent during the conflict period, while others may be new. 27 The four hypothetical scenarios laid out in Figures 6a-d illustrate. In each, the shaded space represents the trajectory of the original, large-scale, violence, and the red line represents routine, small-scale, violence. All four figures show steadily declining large-scale violence, but four different trajectories in the evolution of routine violence. In Figure 6a, routine violence continues, unaffected by the decline in large-scale violence. In Figure 6b, both large-scale and routine violence decline, at a similar rate. In Figure 6c, the two rates of decline are different, with routine violence taking longer to decline. In Figure 6d, large-scale violence declines over time, but routine violence increases in its stead. 28 Figure 6: Different levels of routine postconflict violence Figure 6a Figure 6b Large-scale violence Routine violence Large-scale violence Routine violence 35 35 30 30 Number of deaths Number of deaths 25 25 20 20 15 15 10 10 5 5 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12 Months after end of conflict Months after end of conflict Figure 6c Figure 6d Large-scale violence Routine violence Large-scale violence Routine violence 35 35 30 30 Number of deaths Number of deaths 25 25 20 20 15 15 10 10 5 5 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12 Months after end of conflict Months after end of conflict Comparing forms of routine violence in postconflict settings Variation in the forms of postconflict violence present will also be considered. Figures 7a and 7b break down further the different forms of postconflict violence observable in two areas. In each, levels of violence remain similar but violence has taken different forms. Figure 7a shows a rise in vigilantism, while Figure 7b shows a rise in land conflicts. Figure 7a: Figure 7b: Rise in vigilantism in post-conflict period Rise in land conflicts in post-conflict period Total Violence Large-scale violence Total Violence Large-scale violence Land conflicts Vigilantism Land conflicts Vigilantism 60 60 50 50 Number of deaths Number of deaths 40 40 30 30 20 20 10 10 0 0 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Months Months 29 Within case study analysis Much can also be learned by looking at variations within cases in levels and forms of violence. The quantitative dataset will allow us to look for sharp rises and falls in violence over the postconflict period. In the qualitative cases, time will be spent identifying reasons why such rapid changes occurred. Our within case analysis will focus on comparing periods of conflict escalation with those where conflict has re-escalated (after initial drops in violence). Figure 8 shows the pattern of violence over time in a hypothetical district. Comparative analysis of two periods (marked with arrows) within the same conflict can help in determining whether a recent period of violence re-escalation is a precursor to a future period of extended violence or whether the characteristics are different, with re-escalation a temporary blip. Within case analysis will also be conducted for periods of conflict de-escalation. Figure 8: Within-case analysis of de-escalation 45 40 35 Number of deaths 30 25 20 15 10 5 0 1 2 3 4 5 6 7 8 9 10 11 12 Months Finally, within case analysis will compare conditions after conflict de-escalation with those before the original period of violence. Are the factors which (we hypothesize) led to the initial violence missing or do they remain. (These will be identified in the component of the project on escalation). If these underlying factors remain, we can posit that violence may re-emerge again in the future? 4. FIELDWORK AND ANALYSIS Variation along these dimensions will be used to select a number of cases (to be decided) for comparative analysis. The level at which case comparisons will take place (province, district or sub-district) will be decided after we obtain information indicating the level at which variation is most marked (which will come from the database). Ideally, we will match cases at multiple levels (provincial, district, sub-district) to help tease out the factors at each level that contributed to differential trajectories. Qualitative fieldwork will then be needed. As with the work on routine violence and escalation, structured comparisons will be employed. After cases are selected, in-depth fieldwork will then carried out to try to tease out the sources of the different forms of variation. When there, they will focus on a number of issues which may include:56 56 Other focus areas will be identified once the quantitative data is in. This will allow us to develop hypotheses that can be tested during the fieldwork. Relevant literatures that may also help in hypothesis formulation include that on social movements (e.g. Tarrow 1994), sources of revolution (Wolf 1969; Scott 1976; Skopkal 1994); the security 30 · What are informants' (villagers, leaders, conflict actors, government, etc.) views on why violence has declined (or not)? · How have conflict actors reorganized themselves? Are their former networks still strong? To what extent are their identifies, and their positions in communities, still defined by the role they played in the conflict? · What role have government policies and actions played? How did local and national government respond to the conflict, either directly or in providing belligerents with positions of power or resources? · What role have aid and peacebuilding efforts played? How have they contributed to limiting/triggering renewed violence? · What preexisting social networks exist? What cross-cutting relationships and institutions existed before the period of large-scale violent conflict, how were they affected by the violence, and how have they been reconstituted (or not)? 5. TESTING THE THEORY ON DIFFERENT TYPES OF CASES The analysis above will help generate theory on why violence de-escalates in different ways and what conditions make for sustainable peace in postconflict areas. Our sampled provinces were all sites of large-scale intercommunal violence. To what extent do the same factors and dynamics that lead to de- escalation and sustainable peace in these areas lead to similar outcomes in areas that have experienced other types of extreme violence? This question has particular theoretical relevance. In much of the literature it is assumed that resumption of intercommunal violence has very different causes to those that lead to renewed violence against the state.57 This sounds plausible, but is it true? We plan to apply some of the hypotheses generated from the analysis above to the case of Aceh, where conflict between a secessionist movement and the Indonesian state led to over 15,000 deaths. The method and analytical steps will be similar to those above. The quantitative dataset will identify patterns of postconflict violence by area within Aceh. We will then look for variation in current violence levels, and in patterns of de-escalation. Fieldwork can help test whether the same factors led to de-escalation here as in the sites of intercommunal violence. dilemma (Posen 1993); the organization of violence and rebel movements (Kalyvas 2006; Weinstein 2007); institutional organization (e.g. North 1990; Fukuyama 2004); and broader work on the political economy of democratic transitions (e.g. Haggard and Kaufman 1995; Bates 2001). 57 On the difference between ethnic wars and other violent conflicts, see Horowitz (1985), Fearon (2004), and Doyle and Sambani (2006). 31 VII. Testing Our Hypotheses: Links to Other Quantitative Data As we have said above, we also remain open to the idea that the ViCIS study data can be used for econometric testing of hypotheses about what causes violent conflict in countries and what accounts for variations in violent conflict levels. A new literature on the micro-dynamics and foundations of civil war and local conflict has emerged in recent years. These studies look at the impacts of local social, economic and institutional factors in shaping violence propensity.58 One weakness of many existing studies is that the dependent variable (incidence or impacts of violent conflict) is poorly measured or collected at only one point in time. This is an artifact of the use of household or key informant surveys. The newspaper conflict dataset will provide a series of dependent variables (presence of violent conflict, levels of violent conflict, presence of particular forms of violence, etc) that can be used for econometric analysis. The dataset will have limited data on independent variables. However, these can be integrated from other existing datasets such as SUSENAS, PODES and the Governance and Decentralization Survey. As such, we expect that a final stage of the project will be to merge the newspaper conflict dataset with others survey datasets. This will allow for formal testing of some of the hypotheses developed from the case study work. At this point we cannot determine what hypotheses we would like to test, and hence what variables we would want to merge into the dataset. However, areas of hypotheses are likely to relate to three areas: (a) social/demographic causes of violence; (b) economic causes of violence; and (c) institutions and violence. Independent variables may include the following: Social/demographic factors · Percentage of unmarried men · Youth bulge: high or increasing proportion of youth (especially young men) · Ethnic heterogeneity ­ ELF, with temporal aspect (changes in ELF) · In-migration · Season and temperature Economic factors · Economic shocks: national and local · Changes in wages · Unemployment, changes in employment levels (especially for young men) · Changes in inflation · Changes in commodity prices Institutional factors · Perceptions of governance · Proximity to religious institutions · Representativeness of government institutions (for example percentage of civil servants by ethnicity) · District splitting (pemekaran) 58 See Kalyvas (2007) for a summary and critique. 32 · Changes in government after local elections · Remoteness from urban centers 33 VIII. Audience and Outputs 1. AUDIENCES The project aims to speak to a number of distinct audiences. First, ViCIS will provide data and insights of use to policy makers in Indonesia. These will feed into Bappenas's (the National Planning Agency) strategy for developing of conflict-affected disadvantaged areas. The study will be financed, in part, from funds from a joint World Bank-Bappenas grant from the Post-Conflict Fund (PCF), which aims to build knowledge of conflict in Indonesia, and to build the capacity of the state and non-government organizations to respond to it in effective ways. Counterparts from Bappenas will be involved in peer reviewing materials throughout the study to help ensure that the study contributes information useful to the government's Medium Term Development Plan 2010-2014. Bappenas will be responsible for disseminating results and ideas from the study to other Government ministries. Results will be fed into the drafting of the law on conflict management, which is scheduled to go to parliament in 2009 or 2010. With support from Bappenas, outputs from this study will be streamlined with the timeline of the drafting of legislation to inform lawmakers about conflict dynamics in every province and enable them to devise effective conflict management strategies. Results will also be disseminated widely to local governments to strengthen their role in conflict management and prevention. These are increasingly important to promote conflict sensitivity in participatory planning process within the policy context of Bappenas' support of the Musrenbang process. It is envisioned that funds for pilot project responses based on findings will be found from other trust funds, other donors, and government budget.59 Second, ViCIS will provide key information for development practitioners from aid agencies and national and international NGOs. The study will help these groups prioritize areas to work in, types of programs to finance and implement, and will give a broad sense of the extent to which "conflict programming" should remain a priority as Indonesia matures as a democracy. Findings will feed into the ongoing implementation of large-scale Bank-financed projects such as PNPM/KDP and SPADA that operate in many conflict-affected areas in Indonesia. Materials may be of use in building the capacity of facilitators and other project staff to understand, analyze and manage conflict in their areas. Third, the project will provide fresh insights to Indonesian and Indonesianist scholars and to the broader global community working, and writing, on conflict issues. The analytic approaches should help to contribute to answers to some of the key question on conflict in Indonesia today (as discussed earlier). ViCIS will provide key data and frameworks that will have applicability to understanding conflict, and how to deal with it, in other countries. In addition, our conflict data may be used by others to perform their own analyses. Materials will be published through international fora, and workshops will be held in Washington and elsewhere to disseminate findings. The project aims to help in the process of incorporating Indonesian materials into international debates and work on violent conflict. 59 Some donors, especially USAID, have expressed interest, as has Bappenas. 34 Finally, the study also aims to build the capacity of local research organizations to conduct empirical policy-focused research. Large portions of the fieldwork will be implemented by partners. Local research institutes will assist in the quantitative data collection, while much of the qualitative data collection will be contracted to Indonesian social scientists (political scientists, historians, anthropologists), selected from local universities in sites chosen for case studies wherever possible. This will also help ensure capacity- building activities target organizations outside of Java. 2. OUTPUTS Given the diverse audiences for the project, outputs will take a number of different forms. Policy briefing notes Regular briefing notes will be produced summarizing data and analysis. Key target audience is the government. Briefing notes will also be produced for an international audience and will be disseminated through the World Bank's Crime, Conflict and Violence team in Washington. Working papers, journal articles and book Working papers will be produced for each of the four research topics. There will be an emphasis on getting data and early analysis out quickly to help maximize usefulness for government and practitioners in Indonesia. At a later point, more conceptual and theoretical papers will be produced aimed at a wider audience. A book will bring together the findings from the study. Conflict dataset The project will produce a comprehensive dataset on violence in Indonesia. This will be on-line and freely available for conflict researchers, government, NGOs, etc. It is planned that the dataset will be maintained in partnership with Bappenas to ensure data is kept up-to-date when the project ends. Workshops and capacity building The project aims not only to produce good data and theory, but to build the capacity of local groups in Indonesia. Regular workshops will be held to disseminate and discuss data. Over time, it is hoped that funds will be available for targeted capacity building of local government and civil society groups to interpret the data and findings, and to plan responses. 35 References Acciaioli, Greg (2001). "Ground of Conflict, Idioms of Harmony: Custom, Religion, and Nationalism in Violence Avoidance in Lindu Plain, Central Sulawesi." Indonesia 72 (October): 81-114. Anderson, Benedict R. O'G. (ed.) (2001). Violence and the State in Suharto's Indonesia. Ithaca: Southeast Asia Program, Cornell University. Anwar, Dewi Fortuna, Hélène Bouvier, Glenn Smith, and Roger Tol (eds.) (2005). Violent Internal Conflicts in Asia Pacific: Histories, Political Economies and Policies. Jakarta: KITLV. Aragon, Lorraine V. (2001). "Communal Violence in Poso, Central Sulawesi: Where People Eat Fish and Fish Eat People." Indonesia 72 (October): 45-74. Aspinall, Edward (2006). "Violence and Identity Formation in Aceh Under Indonesian Rule", in Anthony Reid (ed.), Verandah of Violence: the Background to the Aceh Problem. Singapore: Singapore University Press, pp. 149-178. Aspinall, Edward (2008). "Ethnic and religious violence in Indonesia: a review essay." Australian Journal of International Affairs 62(4): 558-572 Aspinall, Edward (forthcoming). Islam and Nation: Separatist Rebellion in Aceh, Indonesia. Stanford: Stanford University Press. Barron, Patrick (2009). "The Limits of DDR: Reintegration Lessons from Aceh", in Small Arms Survey Yearbook 2009. Oxford/Geneva: Oxford University Press. Barron, Patrick, Samuel Clark, and Muslahuddin Daud (2005). Conflict and Recovery in Aceh: An Assessment of Conflict Dynamics and Options for Support to the Peace Process. Jakarta: World Bank. Barron, Patrick, Rachael Diprose, and Michael Woolcock (2006). "Local Conflict and Community Development in Indonésia: Assessing the Impact of the Kecamatan Development Program." Indonesian Social Development Paper No. 10. Jakarta: World Bank. Barron, Patrick, Kai Kaiser, and Menno Pradhan (2009). "Understanding Variations in Local Conflict: Evidence and Implications from Indonesia." World Development. 37(3): 698-713. Barron, Patrick, and David Madden (2004). "Violence and Conflict Resolution in Non-Conflict Regions: The Case of Lampung, Indonesia." Indonesian Social Development Paper No. 2. Jakarta.: World Bank. Barron, Patrick, and Joanne Sharpe (2005). "Counting Conflicts: Using Newspaper Reports to Understand Violence in Indonesia." Conflict Prevention and Reconstruction Paper No. 25. Washington, D.C.: World Bank. Barron, Patrick, and Joanne Sharpe (2008). "Local Conflict in Post-Suharto Indonesia: Understanding Variations in Violence Levels and Forms through Local Newspapers." Journal of East Asian Studies 8(3): 395-423. 36 Bates, Robert H. (2001). Prosperity and Violence: The Political Economy of Development. London: W.W. Norton. Bertrand, Jacques (2004). Nationalism and Ethnic Violence in Indonesia. Cambridge: Cambridge University Press. Bertand, Jacques (2008). "Ethnic Conflicts in Indonesia: National Models, Critical Junctures and the Timing of Violence." Journal of East Asian Studies 8(3): 425-450. BPS (2007a). Daftar Name Provinsi/Kabupaten/Kota Menurut Dasar Hukum Pembentukan Wilayah, 2007. Available at BPS (2007b). Proyeksi Penduduk Indonesia, Tahun 2005-2014. Available at Chaudhary, Torunn and Astri Suhrke (2008). "Postwar violence". Unpublished background paper, Small Arms Survey, Geneva. Clark, Samuel (ed.) (2005). "More Than Ownership: Ten Case Stdueis of Land and Natural Resource Conflicts from East Java and NTT." Indonesian Social Development Paper No. 4. Jakarta: World Bank. Collier, Paul, Lani Elliot, Havard Hegre, Anke Hoeffler, Marta Reynal-Querol, and Nicholas Sambanis (2003). Breaking the Conflict Trap: Civil War and Development Policy. Washington, DC: World Bank/Oxford University Press. Colombijn, Freek, and Thomas Lindblad (eds.) (2002). Roots Of Violence in Indonesia. Singapore: Institute of Southeast Asian Studies. Coppel, Charles A. (ed.) (2006). Violent Conflicts in Indonesia. London: Routledge Curzon. Davidson, Jamie S. (2008). From Rebellion to Riots: Collective Violence on Indonesian Borneo. Madison, WI: University of Wisconsin Press. Doyle, Michael W., and Nicholas Sambanis (2006). Making War and Building Peace. Princeton, NJ: Princeton University Press. Fearon, James D. (2004). "Why do some civil wars last so much longer than others?" Journal of Peace Research 41: 275-301. Fortna, Virginia Page (2008). Does Peacekeeping Work? Shaping Belligerents' Choices After Civil War. Princeton, NJ: Princeton University Press. Fukuyama, Francis (2004). Statebuilding: Governance and World Order in the Twenty-First Century. London: Profile Books. George, Alexander L. and Andrew Bennett (2005). Case Studies and Theory Development in the Social Sciences. Cambridge, MA.: MIT Press. 37 Gerring, John (2007) "Case Study: What is it? What is it Good For?" in C. Boix and S. Stokes, eds. Oxford Handbook of Comparative Politics. Oxford: Oxford University Press. Haggard, Stephan, and Richard R. Kaufman (1995). The Political Economy of Democratic Transitions. Princeton, NJ: Princeton University Press. Herriman, Nicholas (2007). "A Din of Whispers: Community, State Control, and Violence in Indonesia. Unpublished Ph.D. dissertation, University of Western Australia. Horowitz, Donald (1985). Ethnic Groups in Conflict. Berkeley, CA.: University of California Press. Hüsken, Frans, and Huub de Jonge (eds.) (2002). Violence and Vengeance: Discontent and conflict in New Order Indonesia. Saarbrücken: Verlag für Entwicklungspolitik. Kalyvas, Stathis N. (2006). The Logic of Violence in Civil War. Cambridge: Cambridge University Press. Kalyvas, Stathis N. (2008). "Promises and pitfalls of an emerging research program: the microdynamics of civil war", in Stathis N. Kalyvas, Ian Shapiro, and Tarek Masoud (eds.) Order, Conflict, and Violence. Cambridge: Cambridge University Press. King, Gary, Robert O. Keohane, and Sidney Verba (2004). Designing Social Enquiry. New York: Princeton University Press. van Klinken, Gerry (2000). "Ethnic fascism in Borneo," Inside Indonesia 64(June-July). van Klinken, Gerry (2001). "The Maluku Wars of 1999: Bringing Society Back In," Indonesia 71: 1-26. van Klinken, Gerry (2007a). Communal Violence and Democratization in Indonesia: Small Town Wars. London: Routledge. van Klinken, Gerry (ed.) (2007b). Renegotiating Boundaries: Local Politics in post-Soeharto Indonesia. Leiden: KITLV Press. Kostner, Markus, and Kelly Johnson (forthcoming). "Cash Transfer Programs in Emergency Situations: A Guidance Note." World Bank Guidance Note. Mancini, Luca (2005). "Horizontal inequalities and communal violence; evidence from Indonesian districts." CRISE Working Paper No. 22. Oxford: Center for Research on Inequality, Human Security and Ethnicity, University of Oxford. McAdam, Doug, Sidney Tarrow, and Charles Tilly (2001). Dynamics of Contention. Cambridge: Cambridge University Press. McGibbon, Rodd (2004). "Plural Society in Peril: Migration, Economic Change and the Papua Conflict." Policy Studies No. 13. Washington, D.C.: East-West Center. McRae, Dave (2008). The Escalation and Decline of Violent Conflict in Poso, Central Sulawesi, 1998-2007. Unpublished PhD dissertation, Australian National University. 38 Mietzner, Marcus (2008). Military Politics, Islam and the State in Indonesia: From Turbulent Transition to Democratic Consolidation. Singapore: ISEAS. Moser, Caroline and Dennis Rodgers (2005). "Change, Violence and Insecurity in Non-Conflict Situations." ODI Working Paper No. 245. London: ODI. Muggah, Robert (2009). "Post-conflict armed violence and security promotion", in Small Arms Survey Yearbook 2009. Oxford/Geneva: Oxford University Press. North, Douglass C. (1990). Institutional Change and Economic Performance. New York: Cambridge University Press. Popper, Karl (2002). Conjectures and Refutations. London: Routledge Classics. Posen, Barry (1993). "The Security Dilemma and Ethnic Conflict", in Michael Brown (ed.) Ethnic Conflict and International Security. Princeton, NJ: Princeton University Press. Purdey, Jemma (2006). Anti-Chinese Violence in Indonesia, 1996-1999 . Singapore: ASAA SEA Publications Series, Singapore University Press. Reid, Anthony (ed.) (2006). Verandah of Violence: the Background to the Aceh Problem. Singapore: Singapore University Press. Rodgers, Dennis (2003). "Dying for It: Gangs, Violence, and Social Change in Urban Nicaragua, 1997- 2002." LSE-DESTIN Development Research Center Crisis States Programme Working Paper No. 35. London: DRC. Rodgers, Dennis (2007). "Slum Wars of the 21st Century: The New Geography of Conflict in Central America." Crisis States Research Center Working Paper No. 10. London: LSE. Schulze, Kirsten E. (2004). "The Free Aceh Movement (GAM): Anatomy of a Separatist Organization." Policy Study No. 2. Washington, DC: East-West Center. Scott, James C. (1976). The Moral Economy of the Peasant: Rebellion and Subsistence in Southeast Asia. New Haven and London: Yale University Press. Sharpe, Joanne (2005). "Using Newspapers to Monitor Conflict: Evidence from Maluku and North Maluku, Indonesia". Mimeo. Jakarta: World Bank. Sidel, John T. (2006). Riots, Pogroms, Jihad: Religious Violence in Indonesia. Ithica, N.Y.: Cornell University Press. Siegel, James T. (1998). "Early Thoughts on the Violence of May 13 and 14, 1998 in Jakarta." Indonesia 66: 75-108. Skopkal, Theda (1994). Social Revolutions in the Modern World. Cambridge: Cambridge University Press. 39 Smith, Claire Q. (2005). "The Roots of Violence and Prospects for Reconciliation: a Case Study of Ethnic Conflict in Central Kalimantan, Indonesia." Conflict Prevention and Reconstruction Paper No. 23. Washington D.C.: World Bank. Stedman, Stephen John, Donald Rothchild, and Elizabeth M. Cousens (eds.) (2002). Ending Civil Wars: The Implementation of Peace Agreements. Boulder, CO. and London: Lynne Reinner. Sukma, Rizal (2004). "Security Operations in Aceh: Goals, Consequences, and Lessons." Policy Studies No. 3. Washington, D.C.: East-West Center. Tadjoeddin, Mohammad Zulfan, and Syed Mansoob Murshed (2007). "Socio-Economic Determinants of Everyday Violence in Indonesia: An Empirical Investigation of Javanese Districts, 1994-2003" Journal of Peace Research 44(6): 689-709. Tajima, Yuhki (2004). "Mobilizing for Violence: The Escalation and Limitation of Identity Conflicts: the Case of Lampung, Indonesia." Indonesian Social Development Paper No. 3. Jakarta: World Bank. Tarrow, Sidney G. (1994). Power in Movement: Social Movements, Collective Action and Politics. Cambridge: Cambridge University Press. Tilly, Charles (1995). "Democracy is a Lake", in George Reid Andrews and Herrick Chapman (eds.) The Social Construction of Democracy, 1870-1990. New York: New York University Press, pp. 365-387. Törnquist, Olle (ed.) (2000). Political Violence: Indonesia and India in comparative perspective. Oslo: University of Oslo, Center for Development and the E Varshney, Ashutosh (2002). Ethnic Conflict and Civic Life: Hindus and Muslims in India. New Haven: Yale University Press. Varshney, Ashutosh (2007). "Ethnicity and Ethnic Conflict", in Carles Boix and Susan Stokes (eds.), Handbook of Comparative Politics. Oxford: Oxford University Press. Varshney, Ashutosh (2008), "Analyzing Collective Violence in Indonesia: An Overview." Journal of East Asian Studies 8(3): 341-360. Varshney, Ashutosh, Rizal Panggabean, and Mohammad Zulfan Tadjoeddin (2008). Creating Datasets in Information-Poor Environments: Patterns of Collective Violence in Indonesia, 1990­2003. Journal of East Asian Studies 8(3): 361-394. Vel, Jacqueline (2001). Tribal Battle in Remote Island: Crisis and Violence in Sumba (Eastern Indonesia). Indonesia 72:141-58. Waiselfisz, Julio Jacobo (2008). Mapa da Violencia: Os Jovens da America Latina: 2008. Brasilia: Institute Sangari. Weinstein, Jeremy M. (2007). Inside Rebellion: The Politics of Insurgent Violence. Cambridge: Cambridge University Press. 40 Welsh, Bridget (2008). "Local and National: Keroyakan Mobbing in Indonesia." Journal of East Asian Studies 8(3): 473-504. Wessel, Ingrid, and Georgia Wimhöfer (eds.) (2001). Violence in Indonesia. Hamburg: Abera. Wilson, Chris (2008). Ethno-Religious Violence in Indonesia: From Soil to God. London: Routledge. Wolf, Eric (1969). Peasant Wars of the Twentieth Century. New York: Harper & Row. World Bank (2008). Aceh Conflict Monitoring Update: October 1st-November 30th. Banda Aceh: World Bank. 41 Annex A: Concepts and Definitions Conflict · "A struggle over values and claims to secure status, power, and resources, a struggle in which the main aims of opponents are to neutralize, injure, or eliminate rivals" (Coser 1956). Violent Conflict · Includes all forms of conflict between individuals and groups that involve any form of physical action that has concrete consequences such as deaths, injuries and destruction of property, or which has a violent form. Non-Violent · Includes all incidents that involve: (1) the formation and mobilization of a Conflict group and (2) where the group travels to a location and delivers a demand for action or response. Violent Crime · Includes all violent criminal incidents that have not been coded as conflict. These include those illegal activities that involve any form of physical action that has concrete impacts such as deaths, injuries and destruction of property or that have a violent form. Crime differs from conflict in that there is no issue over which the parties are in dispute. Escalation · The following will be considered in the study of escalation: Change from non-violent conflict to violent conflict. Increase in scale, i.e. number of participants, impacts, number of incidents. Escalation within incidents as well as across incidents. Escalation into large high profile conflicts. Escalation on a much smaller scale: from a demonstration to a few small riots with no deaths. De-escalation · The following will be considered in the study of escalation: Change from violent to non-violent form. Decrease in scale, i.e. number of participants, impacts, number of incidents. Mutual resolution of underlying issues between parties. Routine · Frequently occurring forms of violence, which have low per-incident impacts Violence (maximum five deaths) and which are not part of a large or widespread conflict. Such incidents involve local actors struggling over local issues, rather than large-scale mobilization by identity characteristics (such as ethnicity, religion, or region). 42 Annex B: Provinces and Estimated Distribution of Media Sources Province No of Districts Provincial Papers Sub-Provincial Papers Aceh 23 2 2 Papua 20 2 2 West Papua 9 2 2 Central Sulawesi 10 2 2 Maluku 8 2 2 North Maluku 8 2 1 Jakarta 6 - 3 Central Java 35 2 4 Lampung 10 2 3 East Nusa Tenggara 19 2 3 East Java 38 2 4 North Sumatra 26 2 4 West Kalimantan 13 2 3 West Nusa Tenggara 9 2 3 North Sulawesi 13 2 2 Yogyakarta 5 2 2 Central Kalimantan 14 2 2 West Java 26 2 4 South Sulawesi 23 2 2 Riau 11 2 3 Bengkulu 9 2 2 Banten 6 2 2 Total 341 42 57 43 Annex C: Coding Template 44 Annex D: Explanation of Codes SECTION 1 Coder Initials Enter your initials in this field. Incident ID · Assign each incident bundle a unique incident ID from the list of IDs provided by the Project Manager. · For two or more events to be considered a single Incident and have the same Incident ID they must be driven by the same issue, involve the same actors and occur on the same date. Source Info · Enter the initials of newspapers (from the provided list) and date of every article attached to incident bundle. SECTION 2 Actors · Enter minimum one actor per side and maximum two actors per side. · If more than 2 actors involved on a side, or if three sides are involved in one incident, write this in the `Coding Issues' Section. · If there is a clear pelaku and a clear korban, enter the pelaku in Side 1 and the korban in Side 2. · Each set of actors, or `sides', should be coded for General, Type, Total and Women. General · At `General', enter the code for the best description of the form of participant to the incident: 1. Individual 2. Group 3. Institution Type · At `Type', enter the code for the best description of who the actors are affiliated with, or acting on behalf of: 1. Individual, group of individuals with no clear affiliation 2. Group (warga, massa) with no clear affiliation 3. Militias (ethnic, religious, martial arts) 4. IDPs 5. Government ­ elected or bureaucracy (civil servant or government body, e.g. courts, Department of Heath) 6. Foreign aid organization/NGO (including employees) 7. Indonesian NGO (including employees) 8. Private company, contractor, shop 9. Political party 10. Ormas ­ secular societal group such as youth organization, PP 11. Religious leader or group (Church, Islamic group, santri group, NU, JI). Includes preman-like groups which claim to be religious groups such as FPI. 12. TNI (Indonesian armed forces) 13. Brimob (Police special forces) 14. Police 15. Separatist group (GAM before MoU, RMS, OPM) 16. Trade union/workers group (formal or informal) 17. Electoral institution (ie Panwas, KPUD, KPU) 18. Students 88A UNCLEAR 99A OTHER (Explain) 45 Total · Enter the total number of actors involved for every actor. Follow the following guidelines: o IF ranges are mentioned, estimate numbers using the following rules: Beberapa-2 Belasan-11 Puluhan-20 Ratusan-100 Ribuan-1000 Range (e.g 2-4)-lower number. o IF multiple sources report different totals, use the LOWEST estimate (UNLESS the lowest is zero). o IF revised numbers are published by newer sources, use the revised numbers. · IF you have estimated a total from a range, OR if the difference between totals reported by multiple sources is more than 5, tick the box next to that Total. Women? · Tick this box if women were prominent in this set of actors. Cleavage? · Tick this box if you feel there may be a cleavage associated with this conflict. (See definition of cleavages). If you tick this box, then enter codes for Generic and Local Cleavages. Generic · There is space for maximum 2 generic cleavages. If you detect more than 2, write Cleavage this in the Coding Issues section. · Enter the code for the generic cleavage you think is present between the two sides of actors: 1. Ethnic 2. Inter-Religious (Choose between several options, Muslim Christian, Muslim Hindu, Hindu Christian and Other). 3. Intra-Religious 4. Migrant-Local 5. Separatist 99A OTHER (Explain) Local Cleavage · Each region may develop a set of locally relevant cleavage codes. These might be further specification of the generic codes (i.e. which ethnicities are in conflict, or which geographic regions), or they may be different cleavages to those in the generic list (in which case `other' would be selected as the generic cleavage). Local cleavage codes will vary from region to region. List of local cleavages from your province are attached at the end of your coding key. Intervention · If some parties came to the scene of the conflict and attempted to stop the events in this incident, this intervention must be noted. · There is space for entering up to 3 interventions. If you detect more than 3, write this in the Coding Issues section. For each intervention, you must identify, the intervener, the result of the intervention, the conflict form (1 or 2) in which the intervention happened and if the interveners were also the actors. 46 Intervener · For every intervention enter the code for the interveners from the following list: 1. Individual, group of individuals with no clear affiliation 2. Group (warga, massa) with no clear affiliation 3. Militias (ethnic, religious, martial arts) 4. IDPs 5. Government - official, formal or informal (includes Village Heads, Camat, member of parliament) 6. Government ­ bureaucracy (civil servant or government body, e.g. courts, Department of Heath) 7. Foreign aid organization/NGO (including employees) 8. Indonesian NGO (including employees) 9. Private company, contractor, shop 10. Political party 11. Individual political figure 12. Supporters of political candidate(s) or individual political leader. 13. Ormas ­ secular societal group such as youth organization, PP 14. Religious leader or group (Church, Islamic group, santri group, NU, JI). Includes preman-like groups which claim to be religious groups such as FPI. 15. TNI (Indonesian armed forces) 16. Brimob (Police special forces) 17. Police 18. Separatist group (GAM before MoU, RMS, OPM) 19. Trade union/workers group 20. Electoral institution (ie Panwas, KPUD, KPU) 21. Students 77A NONE 88A UNCLEAR 99A OTHER (Explain) Result · For every intervention, enter the code for the result of the attempted intervention: 1. Were contacted but did not come 2. Were contacted but came late 3. Came to the scene but did not intervene 4. Came to the scene, tried to intervene but were unsuccessful 5. Came, intervened successfully to stop the violence, made no arrests 6. Came, intervened successfully and arrested parties to the conflict 7. Came and took into custody the victim of violence 8. Came and aggravated the violence 88A UNCLEAR 99A OTHER (Explain) Conflict Form 1 · If two conflict forms were present in this incident, enter the number of conflict or 2? form in which this intervention took place. (1 or 2?) Intervener · Tick this box if the affiliation of the intervener is the same as one of the actors. affiliated with (e.g., if the conflict is between FPI and Ahmediyya and the religious leader who an actor? tries to intervene is also a member of the Ahmediyya but was not present at the scene before). Violent Impacts · Before writing any numbers in the impacts boxes, read all sources that report this incident and all follow-up articles (attached) to ensure there isn't any more specific information. · In Violent Impacts, you should write information about both human and building impacts. 47 Totals (Deaths, · For every type of impact, enter the total number following the following Injuries, guidelines: Kidnappings, o If it is clear that there were NO impacts, write 77A(NONE) in those fields. Sexual Assaults o IF no impacts are mentioned in any of the sources but you think there and Building may have been some write 88A(UNCLEAR) in those fields. Impacts) o IF impacts are mentioned in ranges estimate numbers using the following rules: Beberapa-2 Belasan-11 Puluhan-20 Ratusan-100 Ribuan-1000 Range (e.g., 2-4)-lower number. o IF multiple sources report different totals, use the LOWEST estimate (UNLESS the lowest is zero). o IF revised numbers are published by newer sources, use the most recent numbers. · IF you have estimated a total from a range, OR if the difference between totals reported by multiple sources is more than 5, tick the (?) box next to the Total of that impact. Female Victims · If it is not clear how many victims were male or female write 88A in corresponding fields. · If you know for a fact that there were no male/female victims, write 77A in those fields. · If you know the breakdown of the male-female impacts, enter them here. · If you estimated the breakdown from a range or if the difference in breakdowns reported by multiple sources was more than 5, tick the ? box next to the male- female breakdown. Breakdown · If it is not clear how many buildings were damaged or destroyed write 88A in (damaged- corresponding fields. destroyed · If you know for a fact that there were no damaged buildings (only destroyed) or buildings) that there were no destroyed buildings (only damaged) , write 77A in those fields. · If you know the breakdown of the damaged-destroyed buildings, enter them here. · If you estimated the breakdown from a range or if the difference in breakdowns reported by multiple sources was more than 5, tick the ? box next to the male- female breakdown. Building Type · Tick all the buildings types that have been damaged or destroyed in this incident. Other Impacts · Use this space to describe any other violent impacts that you could not record in the other fields (e.g burning of cars etc) 48 Weapons · There is space for up to two weapons. If you found that more than two weapons were used, write this in the Coding Issues section. · Enter the code of weapons used from the following list: 1. Club/rock 2. Knife/Spear/other sharp, cutting weapon 3. Small Firearm. 4. Large Firearm. 5. Firearm(but size not clear) 6. Molotov/grenade/bomb 7. Homemade weapon. 8. Fire 77A NONE 88A UNCLEAR 99A OTHER (Explain) Section 3 Conflict · Tick this box if the incident is Conflict (violent OR non-violent) Incident? Conflict Types · For every incident classified as conflict you will enter minimum one Conflict Type and maximum two. · If you think more than two conflict types are present, write this in the Coding Issues section. · To assign a Conflict Type, enter the number of the Type and the letter of the corresponding Conflict sub-type (e.g. a Resource Issue over common Land will be coded as `1A', Resource issue over private Natural Resources will be coded `1D' · If a subtype is not listed in our codes, choose the `OTHER' sub-type and explain it in the space provided. (Write code 199A, 299A...etc) · If the Conflict Type does not fit in any of our types, enter 99A. · If the Conflict Type is entirely unclear, Enter 88A. 1. Resource issues: ownership, access, and use A) Land (common): owned communally or designated public use by the state or owned by individuals/families, private companies or the state. B) Natural resources. C) Man-made resources. D) Markets, routes, customers, etc. E) Access to jobs, markets, routes and services. F) Commodity Prices G) Pollution, environmental damage, noise 99A) Other (explain) 49 2. Administrative issues A) Tender related issues. B) Corruption or misuse of government funds not related with tender issues. C) Public services (quality of a public service, e.g. a education, healthcare utilities, this includes services provided by government and private institutions) D) Other program implementation issues for government funds, including funding priorities and complaints about implementation or unaddressed needs, including subsidies. E) Corruption or misuse of funds for non-government aid and development programs F) Non-government aid or development projects that are not related to corruption of funds. G) Labor-related (industrial action, complaints over pay, conditions) H) Splitting of regions (pemekaran wilayah) or geographic border I) Arrest or other law enforcement issue or legal procedure. 99A) Other (explain) 3. Political issues and contests over position, influence and power A) Government position a national level (e.g. National elections, calls for resignation of national MPs, President or Vice President) B) Government position at provincial level (e.g. Provincial elections, calls for resignation of elected officials) C) Government position at district level (Pilkada and legislative elections etc) D) Government position at sub-district level (e.g. dispute over Camat appointment) E) Government position at village level (e.g. Pilkades, BPD elections) F) Position/influence/power within a political party G) Position/influence/power of appointed government workers outside A-E above, including civil-service and bureaucratic appointments H) Struggle for independence/separation from NKRI, or special autonomy which govern the province's relationship with the central government (eg Aceh/Papua). I) International issue. 99A) Other (explain) 4. Identity issues A) Ethnic issues (eg disputes over cultural attributes of migrants, name of district, monument etc) B) Religious issues (ie attacks on Church for holding service or on gambling parlour, JI attacks ie Bali bomb) C) Intra-religious (eg attacks on Ahmadiyah, enforcement of shariah) D) Migration-related E) Migration and Ethnicity-related (use if the issue involves BOTH Migration AND Ethnicity) F) Geographic (ie inter-village fight) G) Gender related 99A) Other (explain) 50 5. Moral Issues A) Humiliation/loss of face/offense B) Accident C) Debt D) Theft E) Damage to property F) Sexual indiscretion G) Murder/Assault H) Gambling/Alcohol/Narcotics I) Witchcraft 99A) Other (explain) Conflict Forms · For every incident classified as conflict, enter minimum one and maximum two conflict forms. · If you think more than two conflict forms were present in this incident than write this in the Coding Issues section. · Enter the code for the conflict forms present from the following list: Non-violent 1. Demonstration 2. Formal Delegation (delegasi) 3. Formal complaint or protest (formally aired through media, or by submitting complaint to relevant authority) 4. Dispute (squabble between parties) 5. (non-violent) Threat 6. Blockade 7. Strike (stop work) 8. Court case (group takes a challenge to court) Violent 9. Riot: Group attacking and damaging property 10. Riot: Group (over 10, massa) attacking people 11. Group clash (over 10 or use of Indonesian terms ­ kelompok, massa) 12. Fight (small groups or individuals ­ beberapa orang) 13. Smaller Group Attack (many against one or two ie Vigilantism, beating of political opponents) 14. Terror-style attacks, where an individual or small group attacks people or buildings (i.e. attacks which are intended to cause fear, such as a bomb thrown into a crowd, or the Aceh grenade attacks) 15. Vandalism, arson 16. Assault / sexual assault (one-sided) 17. Sweeping/Forcible entry 18. Kidnapping 88A UNCLEAR 99A OTHER (Explain) Violence? · Tick this box if you recorded violent impacts for this conflict incident AND/OR you chose at least one violent Conflict Form 51 Organization in · In case of Riots (Conflict Form 9 or 10) tick this box if there was any sign of Riots? organization, such as coordination, funding, leading by any individuals and / or organizations. · Enter the code of actor responsible for organizing the riot from the following list: 1. Individual, group of individuals with no clear affiliation 2. Group (warga, massa) with no clear affiliation 3. Militias (ethnic, religious, martial arts) 4. IDPs 5. Government - official, formal or informal (includes Village Heads, camat, member of parliament) 6. Government ­ bureaucracy (civil servant or government body, e.g. courts, Department of Heath) 7. Foreign aid organization/NGO (including employees) 8. Indonesian NGO (including employees) 9. Private company, contractor, shop 10. Political party 11. Individual political figure 12. Supporters of political candidate(s) or individual political leader. 13. Ormas ­ secular societal group such as youth organization, PP 14. Religious leader or group (Church, Islamic group, santri group, NU, JI). Includes preman-like groups which claim to be religious groups such as FPI. 15. TNI (Indonesian armed forces) 16. Brimob (Police special forces) 17. Police 18. Separatist group (GAM before MoU, RMS, OPM) 19. Trade union/workers group 20. Electoral institution (ie Panwas, KPUD, KPU) 21. Students 88A UNCLEAR 99A OTHER · Specify which form of organization occurred in the notes section. In field provided, indicate the type of actor involved in organizing the incident. Election Conflicts · Tick the box provided if the conflict concerned an election. To specify the level of election, choose the appropriate Conflict Type 3. Related to · Tick this box if the conflict incident you are coding is related to a crime Crime? committed previously. Individual issue · Tick this box if the incident was a purely individual dispute and not concerning a group issue. Tick this box if "the dispute concerns the `status, power and resources' of a central individual actor or small number of individuals and not those of the wider community". Change from · Tick this box if the issue changed from individual to a group issue within the same individual to incident. group issue? Related to other · Tick this box if the conflict is related to another conflict in the same province. conflict Conflict ID · TO BE COMPLETED BY THE CODING TEAM LEADER after the template has been coded. SECTION 4 Pure Crime? · Tick this box if the incident is a Pure Crime (make sure you haven't entered any information in Section 3) 52 Crime Form · For every incident classified as pure crime, enter minimum one and maximum two crime forms. · If you think more than two crime forms are present, write this in the Coding Issues section. · Enter the code for the crime form from the following list: 1. Murder/Manslaughter 2. Assault 3. Rape, Sexual Assault 4. Robbery 5. Deprivation of liberty (ie kidnapping) 6. Domestic violence 88A UNCLEAR 99A OTHER SECTION 5 Incident Write a brief summary, 2-3 sentences, so that the reviewer can understand what Summary: happened without needing to look at the article. Try not to repeat too much information which is already coded on the template but do the following: · Identify specific actors · Distinguish between sides · Explain sequence of events · Other important details Coding issues: · Note any problems you had choosing appropriate codes, or coding issues which arose, so that they can be reviewed. 53