This problem set is due Monday 19 February2018 and must be submited via Faser.
NOTE that you need to upload two files to Faser:
1. A Word/PDF document where you answer the asignment's questions, and include also graphs as wel
as regresion outputs. This is the main document and should include everything you are asked to do in the
asignment.
2. Your R script. with al the commands and procedures you used to generate the results. Make sure that
your script. runs correctly (no eror mesages etc.). We wil replicate your analysis, and if your code
doesn’t execute this likely wil afect your mark.
For this asignment we wil explore some descriptive questions about ethnicity and conflict, using data
from Bosnia-Herzegovina, drawn from Weidmann, Nils B. 2011. “Violence ‘from Above’ or ‘from
Below’? The Role of Ethnicity in Bosnia’s Civil War”, Journal of Politics 73(4): 1178-1190.
We have two versions of this dataset:
The first version of the dataset is a simple text file caled “tablebosnia.txt”, where the fields are separated
by tabulators.
The second version of the data comes in thre diferent files, ps3bosnia.dbf”, “ps3bosnia.shp”, and
“ps3bosnia.shx”. These data include information on the geographical shapes and positions of the diferent
districts in Bosnia, in addition to the variables. This format is caled a shapefile, and is a common format
for GIS software such as ArcGIS. We wil use this version to create maps of the Bosnia data later.
You can find these files on Moodle.
The data include observations for 109 Bosnian municipalities. The variables in the data are listed below:
SP_ID Spatial polygon ID
MUNID Municipality ID code
name Name
TOTINCOME Income of municipality in 1991
TOTPOP91 Population of municipality in 1991
roadlength Length of roads in district
area Area of municpality
distserbia Distance from Serbia
distcroati Distance from Croatia
KILED Number of people kiled in war
MISING Number of people reported mising
pol_local Residential polarization index
urban91 Urbanization
strat91 Teritorial contestation index
roaddensity Road density
l.distserbia log10 distance to Serbia
l.distcroatia log10 distance to Croatia
borderdist distance to border
BORDER Border municipality
MOSLEMS91 Muslim population in 1991
SERBS91 Serb population in 1991
CROATS91 Croat population in 1991
MOSLEMS01 Muslim population in 2001
SERBS01 Serb population in 2001
CROATS01 Croat population in 2001
(1) Using the Bosnia 1991 population shares at the municipality level, compute a measure of ethnic
fractionalization for each municipality acording to the # ELF formula: ELF = 1-sum_i(s_i^2), where
sum_i is the sum over al groups, and s_i is the share of the respective group. Save your result in a new
variable caled ELF91. Draw a histogram of the distribution, and comment on what this tels us about the
pre-war ethnic population distribution.
(2) Now find the ELF for municipalities in 2001 using the appropriate population figures in the dataset.
Present and discuss the changes in ethnic fractionalization betwen 1991 and 2001. How has the overal
distribution changed overal for the countries, and what districts have changed the most?
(3) Are the population changes related to conflict experience?
(Hint: one way to do this is to create a plot of change in ELF by death rates, and use linear regresion to
estimate the “average” impact of the death rate on changes in ELF. Lab 4 provides some examples).
(4) Examine the variation in the death rate (i.e., casualties per capita) across municipalities, using linear
regresion and two measures developed by Weidmann (2011), “residential polarization” and “teritorial
contestation”, controling for per capita income. Comment on the results.
You wil need to create a per capita income measure yourself, dividing income by population. To facilitate
interpretation, here is how the two ethic geography measures were created. Se the cited articles for more
detailed discussions.
The residential polarization measure was developed by Weidmann (2011), with higher values indicating
teritories that two groups stake a claim to. The specific construction of the measure is as follows (p.
1083):
“I … measure the strategic importance of a unit to [an ethnic group] A by the multiplication of two
factors: First, the average of A’s shares in the unit’s neighbors, and second, the share of A in the
unit itself. … I ..[calculate] an aggregate indicator measuring the degre to which groups’ claims
clash .. I select the two highest importance scores for a unit across al ethnic groups and use the
product of these two scores as my indicator of teritorial ethnic contestation.
A short example ilustrates the logic behind the teritorial contestation indicator. The municipality
of Zvornik in Eastern Bosnia was one of the first locations to se violence betwen Serbs and
Muslims during the Bosnia war in earlyApril 1992 (Burg and Shoup 1999, 129). Zvornik has eight
neighboring municipalities in Bosnia (se Figure 2). Since some of these have high population
shares of Serbs (e.g., Bijelina and Sekovici), and Zvornik itself has a Serb share of almost 0.5, the
municipality is of high importance to the Serbs (importance score 0.19). At the same time, Zvornik
also borders some Muslim dominated municipalities (e.g., Kalesija and Bratunac) and has a high
share of Muslims (slightly more than 0.5). For that reason, it is likely to be claimed also by this
ethnic group (importance score 0.29). The multiplication of the Serb and Muslim importance scores
for Zvornik results in an ethnic contestation value of 0.056 and is among the ten highest scores in
the sample.”
Source: Weidmann, Nils B. 2011. “Violence `from above` or `from below`? The Role of Ethnicity in
Bosnia’s Civil War”. Journal of Politics 73(4): 1178-1190, available on Moodle:
https:/moodle.esex.ac.uk/pluginfile.php/709263/mod_resource/content/0/Weidmann2011_article.pdf.
Polarization is a measure suggested by Montalvo and Reynal-Querol (2005), reflecting how far the
observed distribution of ethnic groups approaches a case of bipolarity where two groups are of the same
size, , where s indicates a group share. The possible maximum value is 1, when
two groups have size of exactly 0.5 each.
Source: Reynal-Querol, Marta & José G. Montalvo. 2005. Ethnic polarization, potential conflict and civil
war. American Economic Review 95(3): 796-816. On Moodle at:
https:/moodle.esex.ac.uk/pluginfile.php/709264/mod_resource/content/0/MontvaloReynal-
Querol2005_article.pdf
(5) Create a map of the death rate in Bosnia, using the shape file data and instructions provided.
(6) Briefly (maximum of 750 words), discuss why diferent studies reach so divergent findings regarding
the role of ethnicity. Does this reflect weak theory, poor measures, a lack of any clear relationships, or a
combination of several of these factors?