THE FLETCHER SCHOOL OF LAW AND DIPLOMACY
New Wars and Regional Conflict
Formations
A Case study of Conflict Trends from 1989-2012 in the East Africa:
Rwanda, Uganda, Burundi and the Democratic Republic of the Congo
Nachama Rosen
05/07/2013
Contents
1. Introduction .............................................................................................................................................. 3
2. Research Question .................................................................................................................................... 3
3. The Data: Sources, Projections and Limitations ....................................................................................... 4
5. Methodology ............................................................................................................................................. 5
6. Discussion.................................................................................................................................................. 6
References: ................................................................................................................................................... 7
Appendix A: Annotated Citations.................................................................................................................. 8
Appendix 2: Dataset Comparison Table ........................................................................................................ 9
Appendix 3: Poster ........................................................................................................................................ 9
2
1. Introduction
The ‘new wars’ theory, made popular by Mary Kaldor, argues that conflicts today have a
multitude of transnational connections so that the distinctions between internal and external, and
local and global, are difficult to sustain1. Additionally, the discourse on regional conflict systems
has started to gain momentum in the post-Cold War era as scholars have started to analyze
conflicts through a regional lens that takes into account the considerable speed with which
security threats can travel across space and borders.2,3 In regards to conflicts in Africa, these
dynamics have been referred to as “regional conflict formations”, where conflicts become
interconnected due to their geographic proximity, making them that much more complex.4 This
process of regionalization of armed conflict is facilitated by the individual structural conditions
of states, which in turn has prompted the arguments in-favor of multi-dimensional peacekeeping
operations (weak states, porous borders, failing security infrastructure, regional instability,
migration, and marginalized identities).5 Additionally, a byproduct of this proposed paradigm
shift in how conflicts are fought in both the ‘new wars’ and the ‘regionalization of conflict’
arguments is that states are now more likely to act via proxy in the conflicts of other states. As
Rubin et al argue, Africa constitutes a complex interconnected conflict system and consequently
there is a need to formulate “sets of policies that address multiple arenas and sources of conflict
within a given region in an integrated way.”6
2. Research Question
This project will look seek to answer to questions:
1. Is there evidence, geospatially, of conflicts become regionalized in Africa? This paper
will use the case study of the east African countries of Rwanda, Uganda, Burundi and the
Democratic Republic of the Congo (DRC).
2. What are the conflict datasets available? How do conflict data sets compare to each other?
Three conflict data sets will be compared.
1
Alex Bellamy and Paul Williams, Understanding Peacekeeping: Polity, 2010). p38
Barry Buzan and Ole Wæver, Regions and Powers: The Structure of International Security, Vol. 91: Cambridge
University Press, 2003).
3
David A. Lake and Patrick M. Morgan, "Building Security in the New World of Regional Orders," Regional
Orders.Building Security in a New World (1997).
4
Barnett Rubin, Andrea Armstrong, and Gloria Ntegeye, "Draft Discussion Paper I: Conceptual Overview of the
Origin, Structure, and Dynamics of Regional Conflict Formations" (Nairobi, Kenya, 2001).
5
J. Giroux, D. Lanz, and D. Sguaitamatti, "The Tormented Triangle: The Regionalisation of Conflict in Sudan, Chad
and the Central African Republic," (2010).
6
Rubin, Armstrong, and Ntegeye, Draft Discussion Paper I: Conceptual Overview of the Origin, Structure, and
Dynamics of Regional Conflict Formations
2
3
3. The Data: Sources, Projections and Limitations
(For full citations please see bibliography)
Conflict Data:

Armed Conflict Location and Event Dataset (ACLED) for years 1998-2012

Uppsala Conflict Data Program's Georeferenced Event Dataset (UCDP-GED) for years
1989-1997

Social Conflict in Africa Database (SCAD) – not used for map analysis, just for dataset
comparison
Administrative Boundaries & Geographic Features:

Global Administrative Areas (GADM)

ESRI Data Maps 10

GfK Macon Company
Population:

Center for International Earth Science Information Network (CIESIN) Gridded
Population of the World, 2000
Projection: WGS UTM Zone 35N
Limitations: Two different datasets, using two different definitions of conflict were combined
for this analysis. One of the reasons for this is because the UCDP-GED dataset has events
starting from 1989, which is pre-Rwanda Genocide, while ACLED only has events starting in
1997. The Rwandan Genocide was a large turning point in the region in terms of the
regionalization of the conflict, with the huge refugee flows from Rwanda flowing into the DRC
in 1994, and a Rwanda assisted coup of the government of the DRC in 1996/97. Without those
years in the dataset, the resulting analysis would be severely lacking. This analysis used each
row in the UCDP-GED dataset as one event. As the table comparing the different datasets will
show, using the UCDP-GED dataset in this way probably underestimated the number of conflicts
events occurring. Additionally, both the ACLED and the UCDP-GED datasets seemed to lack a
large number of events occurring along the southern DRC where it borders Angola and Zambia.
At several points both Angola and Zambia had crossed the border and were actively fighting in
the DRC. Additionally, very few hotspots showed up in that area during analysis. One of the
reasons this might occur is that event data is mined from media sources, so event data is
dependent on what media sources decide to cover – or deem most important to cover at that time.
4
5. Methodology
Preparatory Steps:
1) ACLED, UCDP-GED and SCAD were researched and analyzed to determine how to best
combine the datasets (see data limitations and discussion section for further information
on this process). ACLED and UCDP-GED data sets were used in this analysis.
2) After dragging all layers into Arc GIS they must be properly projected into WGS UTM
Zone 35N
3) GADAM boundaries for DRC, Uganda, Rwanda and Burundi should be merged in order
to run kernel density analysis in the whole region
4) UCDP-GED Data Prep:
a. UCDP data should be clipped to the new merged boundary database, and a new
shape file created of these data points. This creates a new shapefile of just the
conflict events within the boundaries of interest.
b. Using ‘select by attribute’ tool select conflict events between 1989-1994 and
create new layer from selected attributes; then export that layer as a shapefile to
create a new file. Do the same for the years 1995-1997.
5) ACLED Data Prep:
a. Since ACLED shapefiles were missing 2012 conflict data, excel files of the data
was used. ACLED has a separate excel file for each country. Merge all 4 country
datasets in excel and then import it into ArcGIS. After the file is pulled into
ArcGIS, convert it into a shapefile.
b.
Using ‘select by attribute’ tool select conflict events between 1998-2002 and
create new layer from selected attributes; then export that layer as a shapefile to
create a new file. Do the same for the years 2003-2006, 2007-2010, and 20112012.
Analysis:
The ultimate goal of the analysis is to show how the density of conflict changed overtime. In
order to do any analysis, the conflict event shapefiles will need to be changed into raster files.
Additionally, calculations were done to determine estimated population possibly effected by
conflict in current time period of 2010-2012
5
1) Using the kernel density tool, calculate the conflict density for each time period. This
operation calculates the number of events within a specific radius creating a density of
events. The cell size chosen was 1000meters and the radius was 75 kilometers.
2) After the density is calculated, use the raster calculator to subtract the different time
periods from each other. Five change-over time raster were calculated.
3) Each change over time is then classified into standard deviations, where negative
numbers mean a decrease in conflict and positive change means conflict is increasing.
4) The spatial analyst tool, ‘zonal statistics’ was used to calculate the estimated population
possibly effected by the conflict in the period from 2010-2012
a. First, the raster depicting conflict density for the period 2010-2012 was
reclassified into 2 categories – high and low conflict.
b. Using ‘zonal statistics’ the CIESIN population count data from 2000 and the
raster of 2010-2012 conflict density was input into the tool, using the ‘sum’
function to calculate the estimated population possible effected in areas of high
conflict.
6. Discussion
The maps on the poster clearly show a regionalization of conflict between the Rwanda, Burundi,
Uganda and the DRC overtime, visually depicting the timeline of events that have occurred in
the 4 countries between 1989 and 2012. Please see timeline on poster for historical events
occurring in each of the four countries. The estimated population possibly effected by the
conflict in this region between the years 2010 - 2012 is 20 million. While there are limitations to
the validity of this analysis (as discussed above), this project a) provides a method to do this type
of analysis, b) depicts a trend of regionalization of some of the conflicts in this region, supported
by socio-political historical events, and c) a comparison do three different datasets to help further
researchers better understand the conflict data available for analysis.
To see a chart comparing the three data sets (ACLED, UCDP-GED and SCAD) please see
uploaded pdf (chart was difficult to insert on a word document).
6
References:
BBC News Africa “Democratic Republic of Congo profile”, http://www.bbc.co.uk/news/worldafrica-13286306, Accessed 5/6/13
BBC News Africa “Burundi Profile”, http://www.bbc.co.uk/news/world-africa-13087604,
Accessed 5/6/13
BBC News Africa “Rwanda Profile”, http://www.bbc.co.uk/news/world-africa-14093322,
Accessed 5/6/13
BBC News Africa “Uganda Profile”, http://www.bbc.co.uk/news/world-africa-14112446,
Accessed 5/6/13
Center for International Earth Science Information Network (CIESIN)/Columbia University,
International Food Policy Research Institute (IFPRI), The World Bank, and Centro Internacional
de Agricultura Tropical (CIAT). 2011. Global Rural-Urban Mapping Project, Version 1
(GRUMPv1): Population Count Grid. Palisades, NY: NASA Socioeconomic Data and
Applications Center (SEDAC). http://sedac.ciesin.columbia.edu/data/set/grump-v1-populationcount. Accessed 5/7/13
Global Administrative Areas Database 2012, www.gadm.org Accessed 5/1/13
Raleigh, Clionadh, Andrew Linke, Håvard Hegre and Joakim Karlsen. 2010. Introducing
ACLED-Armed Conflict Location and Event Data. Journal of Peace Research 47(5) 1-10.
http://www.acleddata.com/index.php/data Accessed 5/1/13
Cullen S. Hendrix and Idean Salehyan. "Social Conflict in Africa Database (SCAD)."
www.scaddata.org Accessed 5/1/13
Sundberg, Ralph, Mathilda Lindgren and Ausra Padskocimaite, 2010, “UCDP GED Codebook
and dataset version 1.5-2011”, Department of Peace and Conflict Research, Uppsala University.
http://pcr.uu.se/research/ucdp/datasets/ucdp_ged/ Accessed 5/1/13
7
Appendix A: Annotated Citations
1. Eck, K. (2012). In data we trust? A comparison of UCDP GED and ACLED conflict
events datasets. Cooperation and Conflict, 47(1), 124-141.
In recent years, several large-scale data-collection projects have produced georeferenced,
disaggregated events-level conflict data which can aid researchers in studying the microlevel
dynamics of civil war. This article describes the differences between the two leading conflict
events datasets, the Uppsala Conflict Data Program Georeferenced Events Dataset (UCDP GED)
and the Armed Conflict Location Events Dataset (ACLED), including their relative strengths and
weaknesses. It is important to note, that this article was written by one of the producers of the
UCDP dataset – and so while reading the article one can tell it is biased against the use of
ACLED data.
One critique of ACLED data is that ACLED’s unit of analysis is the event (violent or nonviolent), yet researchers rarely theorize about events per se, but rather the production and targets
of violence. ACLED’s approach means that an event like the massacre at Srebrenica is given the
same weight in the data as a sniper attack in Sarajevo. It is conceptually problematic for many
theories of civil war when no distinction is made in the intensity and nature of violence. The
authors also discuss how the geospatial precision of each event is coded. For ACLED it is given
a value 1-3 and for UCDP it is given 1-7. Geospatial precision is explained as how precise the
coders can determine where an even occurred (i.e. news sources give a specific village name,
only a district, region, etc).
2. Raleigh, Clionadh and Havard Hegre. 2009. Population Size, Concentration and Civil
War: A Geographically Disaggregated Analysis. Political Geography 28, 4:224-238
This paper asks the question ‘Why do larger countries have more armed conflict?’ The authors
argue that their analysis indicates that the risk of conflict events increases with local population
size, and is particularly large in highly populous locations within a country. They claim that
Conflict events cluster particularly strongly in larger population concentrations that are distant
from the capital, such as the Eastern provinces of the Democratic Republic of Congo.
8
3. Dulic T (2010) Geocoding Bosnian violence. In: Paper presented at the International
Studies Association Annual Convention, New Orleans, LA, 16 February
In his analysis of Bosnia, Dulic finds that news reports provide insufficiently detailed
information to georeference data to the sub-municipal level which he argues is most theoretically
relevant in the Bosnian context. He also finds that journalists frequently fail to distinguish
between military and civilian victims of war, which has implications for both the study of armed
conflict and for one-sided violence. This is just something to keep in mind when doing any
analysis with these types of data sets, specifically surrounding conflicts.
4. Stephenne, Nathalie, Clementine Burnley, and Daniele Ehrlich. "Analyzing spatial
drivers in quantitative conflict studies: The potential and challenges of geographic
information systems." International Studies Review 11, no. 3 (2009): 502-522.
This paper provides a mini literature review of conflict studies, from conceptual research to
quantitative research. The authors also investigate spatial components in conflict and security,
the existing types of information/data and the quantitative methods used. They then go into
discussing methodological concerns when using these spatial drivers.
Appendix 2: Dataset Comparison Table
Please see attached document
Appendix 3: Poster
Please see attached document
9