Professional Documents
Culture Documents
23.10.19
Benutzung von GIS
o Diffusion von Politiken
o Politische und wirtschaftliche Ungleichheiten
o Migrationsbewegungen
o Intensitivitä t von Beziehungen wirschaftlich
o Historische Komponente
o Separatismus
o Klimawandel und Konflikte
o Mobilitä t, mange daran
o Wahlforschung (Gerrymandering)
Minard
o Hat eine Karte gemacht, die die Einmarschen von Napoleon nach Moscow
beschreibt, 19th cent
7 Arten von Information drin
Vektordaten
o Punkte
o Strecke
o Polygon
o Raster
13/11/2019
Finish 3a
Candidate localness and voter choice in the 2015 General Election in England-Evans et al
Said people the closer in distance to a candidate, the more likely they are to vote for
them
o This study looked closer at candidate-voter distance and other mitigating
factors
o Show that candidate-voter distance mattered in election, helps measure
voter information, candidate presence and marginality
Contiguity mattered, candidates living in non-neighbouring
constituencies had lower likelihood of vote than those who did
Introduction
o Distance-based models on neighborhood effect have been potentially
wrongly/under-specified because rely on measures of distance rather than
perceptions thereof, candidates location of origin/length of residence in
constituency
o Higher likeliehood of vote at lower distances in several studies in England
o Want to extend analysis to include models with a fuller specification of
perceptions of localness and consider how voter information and party
supply may moderate distance effects
o 2015 General UK election, English constituencies
Show that although effect size smaller than previous tests, candidate-
voter distance mattered
Candidate localness and voter knowledge
o Perceptual vs behavioral candidate localness
Perceptual uses opinion poll and survey data to gauge importance
voters attach to localness , no direct distance test
Can also account for birthplace/roots/length of time
somewhere
relies on voters demonstrating/stating importance of
candidate and rep attributes
behavioral infers importance of localness using ecological models of
vote share and indv voting behavior, testing localness and proximity
friends and neighbors mechanism
o vs. neighborhood effects which include predominant
political persuasion of the area in question
assumes conceptual justification from friends and neighbors
argument and empirical support from the perceptual approach
with survey data/artificial election scenarios
o contact with candidate is not necessary or sufficient for assuming knowledge
of their background by the voters , rather this derives from multiple sources
at a minimum in England during voting because the
address/constitutency of candidate is on the ballot
o incumbency matters but doesn’t necessarily trump distance effect
o perceived localness as control for distance
o need to control for selection by parties of local candidates to ensure high
visibility and contact
o predictions
controlling for party preference/incumbency, voters will be more
likely to vote for a closer geographic candidate
location based on territorial contiguity rather than distance, voters
will vote for someone living in their constituency or neighboring more
than non-neighboring
will vote for candidate with perceived higher localness and for whom
they have more information about or has contacted them
want to see if party strength account for the distance effect , which
indicates positioning of candidates in areas of previous strength
Data and Method
o Vote, voter knowledge and party pref collected in two-wave internet panel ,
597 foters in 532 of 533 english constituencies (not Buckingham because
speaker of house of commons)
o Asked 11 thermometer questions on 4 main parties to control for paryy
preference
Conservative, labour, liberal democrats, UKIP
o Ranked candidates on 4-point localness scale
Included “don’t know” option to understand if lack of awareness of
localness is worse for a candidate than being perceived as an outsider
nd
o 2 wave was vote results
o Used postal codes to map localness used fractional polynomial
transformation of distance effect in model to see if a log-transformed or
other power-transformed measure fit the data better, but linear was best
o Contiguity measures generated using sdep and spatial packages in R to
identify neighbouring and non-neighbouring constituencies
o Found
Candidates seen as more local do live closer to the voters, those seen
as “very local” have high levels of localness
But wide variation, and large lack of awareness for localness of
candidates
Looked at contact by leaflet
Weak neg correlation for conserve, labour, lib dems, weak pos
for UKIP
o ran series of alternative-specific multinomial probit models to predict vote
likelihood for each party using both case-specific variables like age/gender
and alternative-specific variables like distance
difference between mnp and asmp, asmp treats the choices made by
each respondent as obeservations, produes stacked data matrix
thereof
o IIA- irrelevant alternatives
Odds of preferring party A over B are not affected if new paties added
or old removed
Violated if party C is perfect substitute
Should restrict multinomial and conditional logit analysis to when
choices are perceived as distinct by each decision maker
o ASMP as precaution against IIA assumes error terms have joint multivariate
normal distribution
o Model 1a indicates distance matters even controlling for party/incumbency
o Model 1b, voters no more likely to cote for someone in their own
constituency vs the one next door
o Models 2a and 2b add localness variable to distance model
Clear relationship between perceived localness and distance, other
variables of localness to explain variance
Candidates ranked as fairly local or very are more likely to be voted
Not being local or hardly local isn’t more harmful than not knowing
localness
o Model 3a and 3b candidate profile/contact, marginality indices
Strong covariation between profile/contact/marginality washes out
significane of profile index , reduces explanotry power of candidate
profile index
Marginality significant esp in close races
Higher levels of contact increase probability of vote
o Incumbants have built in advantage, stronger campaigning of opponents
does have an effect though
Discussion
o Distance effect found in 2010 persists in 2015 even accounting for more
variables
o Voters prefer candidates in a more local environment, their constituency or
neighboring , contiguity
But weaker than distance
o Candidate origin/roots also worth investigating, probably significant, effects
voter knowledge
o Should explore gender/ethnicity/occupation/attractiveness , don’t think
they impact distance effect though
Politics, Time, Space, and Attitudes toward US-Mexico Border Security-Gravelle
how have attidues toward border security shifted 2006-2016 in response to
changes in the partisan political climate?
How does spatial context (proximity to us-mexico border) shape attitudes toward
proposed border wall?
Find that time and space along with individual-level political attitudes key factors to
shaping attitudes toward us-mexico border security
Introduction
o Borders shape social phenomena within which they occur
o Debate over relevance of borders in time of globalization
o Study of American politics shows borders do matter
After 9/11 closed borders airspace disrupted trade
Secure fence act of 2006, built 700 miles of fencing along us-mex
border to protect against human traffickers and drug smugglers
o Us-mex border in 2016 election , executive order to build in 2017, used in
negotiations against DACA
o Want to fill gap on study of public support for specific border security
initiatives
Use data 20006-2016 to explore political factors like party id and
ideology, time, and space inclufence attitudes toward wall
Public Opinion and Border Issues
o 2000 survey showed 18% of americans in favor of doing away with Mexican
border, 42% with Canadian , post 9-11 found 69% wanted common border
security policy with Canada
o Eu studies, if closer to border, more in favor of eu
o Proximity to us canda border shapes attitudes towards us-canada relations,
cloer proximity increasing support among supporters of right of center
parties, decrease support for leftist
o Residents of us-mex border states, more likely to say immigration is most
important problem
Democrats near border more likely to support anti-immigrant
legislation than those far away , republicans support regardless
Proximity to us-mex border amplifies partisan divide on issue of
illegal immpgration
Dems close to border increases support for allowing
undocumented to stay, rep decreases
Theory and Hypothesis
o Out-group derogation, distinguish self from other group (rep vs dem)
o Most recent border policy was advance/identified as rep/right-wing admins
Dems see as pointless/pernicious
o Reps more likely to support more security/ conservatives, restrictionist
policy, higher defense spending
o H1: rep more likely to support us-mex border fence/wall than dems
o H2: ideological libs less likely to support us-mex border fence/wall than
conservatives
o Expects magnititude of cleavage to increase over time b.c pol polarization
o 2006 support from dems mixed, Clinton, Obama, biden voted for secure fence
, 2016 trumped stoked anti-im sentiment , anxieties about illegeal im
o H3: partisan differences between reps and dems in support a us-mex border
fence/wall will be greater when border security is more salient
o H4: Ideological differences between conservatives and libs in support a us-
mex border fence/wall will be greater when border security is more salient
o Expect effects of partisanship/ideology to vary overtime, interactive
relationship between party id and time and ideology and time
o Proximity to us-mex border, aspect of space, likely plays role in shaping
attitudes toward security issues
Space vs place
Space location or area on eaeth, place unique locales with
distinctive characteristics
o In border zone, militarization, checks, apprehensions lead to seeing border as
threat
Border research studies area as places not spaces
o indv near border more likely to represent it in terms of concrete aspects,
increase sense of threat , local media environment more cofused
o indv further from border more likely to represent it abstractly in regard to
polcieis about immigration
o H5: Individuals living closer to us-mex border, more likely to support border
fence/wall, further away less likely
Data and Methods
o Used surveys conducted over the years about attitudes toward
border/immigration
Asked if in favor of border/fence/wall
o Used data from lower 48 and DC
o from 2006-2015 46-49% americans supported, 43-48% opposed , 2016 only
36-40%
o took demographic variables into account, as well as wording of question
fence vs wall
o used zip codes to perform approximate or areal geocoding of respondents ,
assigned approx. lat-long coordinates, accounted for distance decay function ,
effects of proximity-distacne to salient geographic features diminish as
distance increases
o positive correlation between proximity to border and Hispanic
concentration, include measures of loval ethnic context as covariates in
regression analyses
o binary logistic regression as modeling approach, fit to the multiply imputed
data
Results and Discussion
o Framing question with wall vs fence didn’t influence support
o Support for increased border infra higher among men, older people, those
without college education
o Support lower among women, young americans, college ed
o No difference between ethnicities/races but Hispanics less likely to favor
construction of wall
o No effects for county-level concentration of Hispanic pop or change in
Hispanic concentration over time
o Predictors say that less support for border fence/wall as one moves fro the
east to the west of the us
o Rep more likely to support than dems, same with conservatives , support H1
and H2
o Partisan gap widened over time, 2016 dems became even less likely to
support wall , support H3, rep and dem identifiers became more polarized
but only more recently
o Moderates less likely to support fence/wall in 2016 than any other time ,
change little before , support H4
o Further from border, less support for fence/wall, closer, more
o Support H5
o Residents near border have more visual cues in physical environment, more
contact , increase feeling of threat and thus support, and vice versa
Conclusion
o American attitudes toward border more recently polarized
reps/conservatives stable over past decade in support, dems from
ambivalence to rejecting
o Proximity to border increases support for border fence or wall regardless if
dem or rep
Probs b/c visual cures and local media
What about those also near Canada esp in 100km zone, compared to those outside
those areas?
Threatening Events and Anti-refugee Violence…Jä ckle and Kö nig
Violence against refugees in Germany 2015-2016
Test if 3 events prompt violence
o Violence/crime committed by refugees
o Police raids against Islamic fundamentalists
o Terrorist attacks in Germany/neighboring countries
Investigates whether violence is mobilized through public statements by politicians
of anti-imm parties
o Results suggest yes
o Statements seem to be prompted by preceding threatening events
Terrorist attacks in neighboring countries most significant
Introduction
o Large gap between xenophobic sentiment (high) and violent actions (low)
o Structural factors help explain spatial/temporal distribution of violence
o Social disorganization but not econ deprivation leads to more xenophobic
violence
o Occurrence of xeno violence indicates likelihood of more to follow esp
nearby
o Mass media can have an impact, as well as publicly visible events like terror
attacks
o Types of events have potential to associate refugees with threat to people or
social cohesion
o Multi-level logistic regression along with model specifications, district-days
on lowest level, districts on second, german laender on highest level
Attacks on Refugees in Germany in the Past Years
o 890,000 refugees arrived in 2015 in Germany
o Attacks on refugees x5 increase 2015-16
o Chronik fluechtlingsfeindlicher Vorfaelle
Personal injuries
Arson attacks on refugee accommodations
Other attacks on accommodations
Hostile demonstrations against refugees
o Group first 3 types together, ignore demos
o Major increase in level of violence against refugees during 2015 and 2016,
berlin, hamburg, munich stand out
Also eastern Germany, esp Saxony
Theoretical Assumptions and hypotheses
o Violence against refugees proportionally rare
o Shed light on why violence against refugees takes place more often at certain
times/places
o Outgroups can be seen as threats if seen as competitors for scarce econ
resources or cultural/identity based motives
o Size of outgroup/increase can impact
But contact theory says repreated interaction with out-group
members reduces fear increases tolerance
o Terrorist attacks, even far away lead to more negative opinion on immigrants
and support of restrictive immigrant policy
Existential threat prompts need for feeling of control, thus emphasize
important ingroup identity
Retribution
o H1: When a threatening event takes place, the probability of attacks on
refugees temporarily rises
Terrorist attacks, criminal acts by refugees, police raids against
islamists
Proximity as strong indicator of perceived social releance
o Terrorism in neighboring countries likely to have
similar effect as domestic
o H2: When politicians of anti-immigration parties depict refugees as
threatening in media reporting, the probability of attacks on refugees
temporarily increases
Data and Method
o Binary dependent variable from Chronik… indicates whether on a given day
between 1.1.2015-31.12.2016 one of the following 3 types of attacks on
refugees took place in 1/402 german districts
Personal industries
Arson attacks on refugee accommodations
Other attacks on accommodations
293,862 observations, 4,098 attakcks
o Used Spiegel online to find info about independent variables attacks
1,269 articles screen for relevant events, 861 from Die Welt
Also added warnings of terror attacks in addition to actual
attacks
o Day with attack score is increased by 1, reduced by ¾ every succeeding day
o 1,008 articles from search refugee+AfD/CSU
o Days coded with 1 if there is an article, 86 days, reduce by ¾ the next day
o Structural variables
Unemployment rates, education attainment, election turnout , people
with immigrant bacgrkound , number of official recipients of asylum
benefits , party variable 2013/14 elections
Controls for number of inhabitants, number of males, summer period,
and work day
o Multilevel logistic model for analysis
Empirical Analysis
o Effect for police raids against Islamic fundamentalists is negligible
o Incidents of violence and crime from refugees and terror attacks in
neighboring countries strongly spark attacks on refugees in following days ,
but not german terror attacks at all
Attacks in Germany complicated, also usually smaller
o Statements by politicians encourages as well
Public statements by anti-imm parties matter the most, followed by
more attacks on refugees, statements induced in part by preceding
incidents of violence and crimes by refugees
o Distance of closest attack/demo in preceding period has a highly significant
negative effect, cumulated number of attacks in a district highly significant
positive effect
o Unemployment on district level unexpected negative significant effect
Greater economic deprivation in a distrct lowers probability of attacks
Contradicts theory of competition
o Contact theory confirmed
o More fatalities increases probability of attack
o No localized radiating effect
o Effect from negative cueing/terror in neighboring countries is strongest for
days, for terror attacks/warnings in Germany 1 day
Conclusion
o Variable for threatening events consistently influential-terro
attacks/warnnings, gravity of attack not much impact
o Media statements by politicians about refugees lead to increase of attacks on
refs in following days
Do 3b for 27.11
The Legacy of Historical Conflict: Evidence from Africa-Besley
Variation between and within countries, compare legacy of recorded conflicts in
Africa with the precolonial period (1400-1700)
o Show historical conflict correlated with more postcolonial conflict, lower
levels of trust, higher sense of ethnic identity, weaker national identity
o Historical conflict negativel correlated with patterns of development
Introduction
o Conflicts most prevalent in poor weakly inst. Countries
o Conflict prev high in Africa 8.5% of years since 1950 conflict, vs 5% in the
world
o Conflict in af not only post-col phenomenon
o Before col, Africa divided into tribes/kingdoms
o Indirect rule colonial period
But created new borders
o Located 91 conflicts in af between 1400-1700
Persistent conflict after could be do to above, or intereaction btwn
geography, natural resources, and patterns of settlements, no link
with weak inst. At country level
o Use data on conflicts subnat level between 1997-2010 to show historical
conflict leads to more recent conflict and lower econ devel
o Precolonial continuity in form of political organization
Political Violence
o 4 main typical hypotheses for why afr conflict prone
Natural resource dependent
Weak and poorly functioning pol inst
Ethnic fragmentation/polarization
Ethnic polarization more positively correlated with conf
Endemic poverty
Reduces opportunity cost of fighting
o Trend over time to move away from small group identities to national
Historical Legacies
o Hist conf have determinants in soc, pol, geo, eco context of regions at the time
of conf
Geography /land can be reason for conf, nat. resources
o Measure physical geo at grid cell level, geo of pol org in old kingdoms of afr
o Conf can affect ev of eco, pol, soc outcomes
Distrust btwn social groups, poorer regions, choice of pol inst
Data and Measurement
o Hist conf in afr from conflict catalogue of Brecke, from 3 main sources
32 or more deaths
o Identified modern country of hist confli and specific geo location
o Between-country analy
Variable =prev of confl btwn 1400-1700, # of years
Ex: mean 5.13 years, range 0-91 years, stand. Dev 15.17 years
Dummy variable=1 if country had any violent confl
o 3546 country-grid cells across 49 countries in Africa, 1 if confl 0 is no
GIS code each confl w/ modern country
Size of dots=# of conflicts at each location if more than 1
o Match data with info from Armed Confl Location Event Data
Gives GIS codes for afr conflict 1997-2010
o Post-indep conflict mean of 2.4 years between countries who had had hist
confl
o Hist confli occurred in places more densely pop in 1400
o Hist confl more likely in capital city modern /where cities were in 1400
Conflict and Political Development in the History of Africa
o Civil wars common
o Process of territorial consolidation led to emergence of afr kingdoms
23
Found hard to consolidate power over wide areas, borders
porous, conflict
o Conflicts x2 likely to occur in grids of hist king
Between Country Evidence
o yj = α + βcj + γxj + εj ,
where yj is the outcome of interest in country j, α is the intercept, cj is
the historical conflict variable, and xj are other controls whichwe
describe aswe go.
o Each additional year of hist conflc=.12 years (2months) of confl post-col (10
years more of postcol confl)
o Dummy variable sfor yellow fever and ruggedness of terrain?
o Add in repression as measurement, purges
Countries with hist of confl more likely to have
o Slave trade variable/pop dens 1400
Slavery does not disrupt corr with war preve btwn 1400-1700
o Angola and Ethopia as outliers
o Countrywith hist conflict 5% lower gdp per cap in 2000
o No corr with expropriation risk/strength of checks/balances
o Afrobaramoter
Attitudinal questions
2008, 18 countries , 25,397 respondents
Not administered in countries w/ ongoing conflict
Asked about trust between groups, self-ascribed identiy
Pos and sig corr between only a sense of ethnic id and hist confl in
country
Within Country Evidence
o Look at level of devel and nat pol inst
o Corr between hist confl and more recent conflict/levels of econ devel
o Our core empirical specification is: yj _ = μj + βdj _ + γxj _ + εj _, (1)
where yj _ is the outcomemeasure in grid cell _ in country j , μj is a
country fixed effect, dj _ is either our old conflict dummy which is
equal to 1 if there was historical conflict in grid cell _ in country j , and
xj _ are other grid cell controls.
o Luminosity at night for proxy for devel from NOAD
o Confl in cell 15% more likely confl in 1997-2010
o Pop dens pos corr with confl
o More confl on rough terrain
o Pos corr hist confl and modern confl
o Neg corr econ devel
o Robustness/Controls
Hist confl around urban centers, increases likelihood of confl by 14%
No corr between ethn diversity and confl
Slave pos corr but not sig
Cap city in cell, recent confl 15% more likely
With all controls, having confl in cell precol, 8% more likely postcol
Conclusions
o Is corr between hist confli and recent confl
o Country-level attitudes infl by conflict
o Regional pattern of devel corr with pattern of hist conf
Come rain, or come wells: How access to groundwater affects communal violence