Article written by 2009 DPDF State Violence Fellow Andrew Linke and John O’Loughlin, featured in A Companion to Political Geography, 2nd edition, edited by John A. Agnew, Virginie Mamadouh, Anna Secor and Joanne Sharp:

Two
fundamental
concepts
in
spatial
analysis
resonate
both
in
political
geography
and
other
disciplines
that
examine
geographic
distributions
of
human
phenomena.
The
meanings
and
significance
of
context
(place)
and
nonstationarity
(spatial
autocorrelation)
have
to
date
been
examined
haphazardly.
Fundamental
questions
relating
to
the
definition
of
neighborhood
(and
the
related
“Modifiable
Areal
Unit
Problem”)
remain
only
partly
answered.
Similarly,
key
and
inconsistent
choices
in
the
handling
of
spatial
autocorrelation
(including
choices
of
spatial
weights
for
analysis)
are
as
important
as
four
decades
ago
when
systematic
examination
of
these
issues
began.
Three
recent
developments
–
increased
data
availability,
data
integration
across
scales,
and
new
analytical
tools
–
have
shifted
the
focus
of
spatial
analysis
in
political
geography.
After
reviewing
these
welcome
improvements,
an
illustration
through
a
stylized
example
of
merging
aggregate
violent
events
and
public
opinion
survey
data
from
a
representative
population
sample
of
the
North
Caucasus
of
Russia
demonstrates
spatial
analytical
opportunities
for
contemporary
political
geography.