Cross-Cultural Invariance and Gender Bias when Measuring ‘Fear of Crime’
| Author | Stefaan Pleysier,Lohan Goethals,Geert Vervaeke |
| DOI | 10.1177/026975800401000303 |
| Date | 01 January 2004 |
| Published date | 01 January 2004 |
International
Review
of
Victimology,
2004,
Vol.l
0,
pp.245-260
0269-758&04
$10
© A B
Academic
Publishers
-Printed
in
Great
Britain
CROSS-CULTURAL
INVARIANCE
AND
GENDER
BIAS
WHEN
MEASURING
'FEAR
OF
CRIME'
S1EFAAN
PLEYSIER,
GEERT
VERVAEKE
and
JOHAN
GOETHALS*
Department
of
Penal
Law
&
Criminology,
Catholic
University
of
Leuven,
Belgium
ABSTRACT
The
persistent
questioning
within
the
'fear
of
crime'
literature
of
its
own
conservative
measurement,
has
resulted
in
a
growing
tendency
towards
using
scaling
techniques,
as
a
far
better
way
of
measuring
a
complex
and
multidimensional
concept
like
'fear
of
crime'.
Choosing
more
complex
measures,
however,
gives
rise
to
a
number
of
'new'
issues
related
to
measurement
error,
which
have
remained
mostly
undiscussed
in
the
'fear
of
crime'
literature.
Cross-cultural
measurement
invariance
is
such
an
issue;
in
order
for
cross-cultural
comparisons
of
fear
of
crime
to
be
meaningful,
the
instruments
used
to
measure
the
theoretical
constructs
have
to
exhibit
adequate
cross-cultural
equivalence.
Gender
bias
is
another
potential
danger
to
measurement
invariance;
a
substantial
part
of
the
'fear of
crime'
literature
has
empirically
found
and
discussed
different
'fear
of
crime'
levels
between
men
and
women.
Without
testing
the
assumption
of
measurement
invariance
it
is
unclear
whether
these
gender
differences
in
'fear
of
crime'
are
a
reflection
of
true
differences,
or
the
result
of
a
different
understanding
of
the
central
concept
between
men
and
women.
Using
exploratory
factor
analysis
and
structural
equation
modeling,
both
cross-cultural
in
variance
and
gender
equivalence
will
be
assessed
in
this
contribution,
based
on
the
1999
General
Election
Study
of
Belgium
(N
=
4239).
INTRODUCTION
Since the empirical discovery
of
'fear
of
crime', originally
as
a 'widespread
public anxiety about crime', in the
US
victim surveys
of
the 'Commission on
Law Enforcement and Administration of Justice' (established by President
Johnson in July 1965), publications and research on 'fear
of
crime' have risen
almost exponentially. Notwithstanding the 'thickness'
of
this body
of
research,
Ditton and Farrall conclude their influential review of the literature rather
•
This
project
is
supported
by
the
Fund
for
Scientific
Research
Flanders
(FWO).
Address
corre-
spondence
to:
Stefaan
Pleysier,
Department
of
Penal
Law
&
Criminology,
Catholic
University
of
Leuven,
Hooverplein
10,
3000
Leuven,
Belgium,
phone:
0032
16
325306,
e-mail:
stefaan.pley-
sier@law.kuleuven.ac.be.
Stefaan
Pleysier
is
currently
working
as
a
research
assistant
at
the
Department
of
Penal
Law
and
Criminology
at
the
Catholic
University
of
Leuven,
Belgium,
where
he
is
preparing
a
PhD
in
Criminology
on
the
'conceptualisation
and
measurement
of
fear
of
crime'.
This
project
is
sponsored
by
the
Fund
for
Scientific
Research
Flanders
(FWO).
Geert
Vervaeke
is
a
Full
Professor
at
the
Department
of
Penal
Law
and
Criminology
at
the
Catholic
University
of
Leuven,
Belgium.
Johan
Goethals
is
a
Full
Professor
at
the
Department
of
Penal
Law
and
Criminology
at
the
Catholic
University
ofLeuven,
Belgium.
246
pessimistically,
to
say
the
least;
after
almost
40
years
of
research
into
this
topic
'surprisingly
little
can
be
said
conclusively
about
fear
of
crime'
(2000:xxi).
An
important
factor
underlying
and
explaining
at
least
partially
the
above
conclusion
is
the
conservatism
inherent
to
the
era
-'the
heyday
of
survey
research' -
when
'fear of
crime'
research
originated.
Large
scale
crime
and
victim
surveys
were
developed
and
administered,
mainly
by
telephone,
in
order
to
have
a
more
accurate
estimate
of
the
victimization
rate
and
the
'widespread
public
anxiety
about
crime'
(Block,
1993).
As
these
large
scale
surveys
have
a
longitudinal
design
and,
therefore,
are
administered
on
a
recurrent
base,
conser-
vatism
not
only
points
to
the
dominant
methodology
-
survey
-
but
within
that
methodology
to
the
item
level
as
well.
The
concept
'fear of crime'
in
these
surveys
is
predominantly
measured
by
a
single,
'standard'
item:
'How
safe
do
or
would
you
feel
being
out
alone
in
your
neighbourhood
at
night?',
with
answering
categories:
'very safe',
'reasonably
safe',
'somewhat
unsafe',
or
'very
unsafe'
(Ditton
and
Farrall,
2000;
Hale,
1996;
Ferraro
and
LaGrange,
1987).
Ever
since,
numerous
authors
have
questioned
this
conservative
tenacity
with
profound
persistence,
resulting
in
a
tendency
towards
using
scaling
techniques,
as
a
far
better
way
of
measuring
a
complex
and
multidimensional
concept
as
'fear of
crime'.
Choosing
more
complex
measures,
however,
gives
rise
to
a
number
of
'new'
issues
related
to
measurement
error,
which
remain
until
today
mostly
undiscovered
in
the
'fear of crime'
literature
1•
Cross-cultural
measurement
invariance
is
such
an
issue;
in
order
for
cross-
cultural
comparisons
of
fear
of
crime
to
be
meaningful,
the
instruments
used
to
measure
the
theoretical
constructs,
e.g.
'fear of crime',
have
to
exhibit
adequate
cross-cultural
equivalence
(Billiet,
2002).
Gender
bias
is
another
potential
danger
to
measurement
invariance
that
is
perhaps
particularly,
although
not
exclusively,
related
to
fear
of
crime
research;
a
substantial
part
of
the
'fear of
crime'
literature
has
empirically
found
and
discussed
different
'fear of crime'
levels
between
men
and
women.
To
our
knowledge,
the
assumption
of
measure-
ment
invariance
has
never
been
tested,
so
gender
differences
in
'fear of crime'
are
not
necessarily
a
reflection
of
true
differences,
but
possibly
a
reflection
of a
different
understanding
of
the
central
concept
(and
its
measurement)
between
men
and
women.
Using
exploratory
factor
analysis
and
structural
equation
modeling,
both
cross-cultural
invariance
and
gender
equivalence
will
be
as-
sessed
in
this
contribution,
based
on
the
1999
General
Election
Study
of
Belgium
(N
=
4239)
(ISPO/PIOP,
2002)
2•
The
1999
General
Election
Study
is
a
large
general
population
survey
includ-
ing
a
variety
of
topics.
It
was
not
specifically
designed
to
focus
on
fear
of
crime,
but
it
does
contain
four
items
on
this.
Analysing
these
four
items
allows
us
to
see
whether
assumptions
of
measurement
in
variance
hold,
or
whether
in
fact
differ-
ent
items
display
cultural
variance
or
gender
bias.
We
are
not
holding
up
the
survey
as
a
particularly
good
way
of
measuring
fear
of
crime,
but
will
merely
be
using
it
to
demonstrate
the
techniques
which
could
be
used
to
test
other
sets
or
scales
aiming
to
measure
fear
of
crime.
To continue reading
Request your trial