The Modelling of Threats: Evidence from the British Crime Survey
| Published date | 01 September 1994 |
| DOI | 10.1177/026975809400300304 |
| Date | 01 September 1994 |
International
Review
ofVictimology,
1994,
Vol.
3,
pp. 235-261
0269-7580/94$10
©
1994
A B Academic Publishers-Printed
in
Great Britain
THE
MODELLING
OF
THREATS:
EVIDENCE
FROM
THE
BRITISH
CRIME
SURVEY
ANDROMACHI
TSELONI,
DENISE
R.
OSBORN
and
KEN
PEASE
Departments
of
Econometrics
and
Social
Statistics,
Social
Policy
and
Social
Work,
University
of
Manchester
ABSTRACT
Threats
to
kill,
to
injure,
or
to
damage property stimulate fear
and
anxiety
in
the victim
and
his/her
immediate human environment. Previous research
has
used
sociodemographic attributes
and
lifestyle or routine activities measurements to predict crime victimisation rates. The purpose of this
study
is
to
examine
how
personal characteristics
are
associated
with
the
probability of
being
threatened.
Data
from
three
sweeps
ofthe British
Crime
Survey (1982,
1984,
1988)
are
employed.
Logistic regression
is
the
main
analytic
tool
used,
with
some
interaction
terms
entering
the
set of
explanatory variables. Appropriate specification
tests
for
logit models, which
are
usually neglected,
are
conducted.
The
results of
the
tests
confirm
the
robustness of our models. The present
study
addresses
two
questions
in
particular:
(a)
which groups of people
are
most threatened?;
and
(b)
do
these groups change over
time
and,
if
so,
how?
INTRODUCTION
To
the authors' knowledge, the only published research
on
threats
is
that of Hough
(1990). His definition
is
as
follows: 'To threaten someone
is
to
declare one's
intention to punish or hurt,
to
inflict injury to reputation or property which
may
restrain a person's freedom of action.' (Hough,
1990;
p.
169).
Having possible
unpleasant consequences hanging over
one
has
long been identified
by
psycho-
logists
as
a debilitating condition (e.g. Seligman,
1975)
and
is
a major psycho-
logical component of torture (see Lifton,
1974).
It
is
thus perhaps surprising that
threats have been
so
neglected
in
analyses of victimisation surveys. They are
interestingly placed between
fear
of crime
and
direct crime victimisation. Max-
field (1984) defines fear
in
the
following terms ' .. .fear
is
an
emotional
and
physical response
to
a threat. When confronted with a situation which poses
danger, people experience physiological changes that help cope with
the
threat'
(p.
3).
A threat
may
be
explicit or implicit in a situation. Apart
from
its
role
as
an
offence
in
its
own right,
the
threat merits study
as
an
agent of fear evocation.
One
of
the
reasons for neglect
in
the
literature
may
be
the
marginal position of
threats
in
definitions of crime. 'Only a minority of threats can
be
prosecuted qua
"threats"
as
indictable offences (and classified
as
notifiable offences).' (Hough,
1990;
p.
169).
Hough combined data
from
the
1982
and
1984
sweeps of
the
British Crime
Survey (hereinafter BCS).
He
found
the
prevalence of threats
to
be
1.7
percent, 1
236
with
85
percent of threats being
to
'beat up'
the
respondent.
In
two
thirds of cases,
the
threatener
was
a stranger. Stranger threats occurred
in
public meeting places,
such
as
pubs, clubs
and
bars,
and
on
public transport. They were concentrated
in
urban areas. Threats
by
non-strangers took place mainly
in
and
around
the
home
and
workplace, showing less variation
by
area of residence.
Young
men seemed
more
likely
to
face
stranger threats,
as
is
the
case for most types of personal
victimisation. Other threats
were
experienced more uniformly across
age
and
gender.
People victimised
by
a non-stranger
were
more likely
to
report
the
event
as
very
upsetting, according
to
the
1982
(but not the
1984)
BCS
data.
Threats
' ... made
by
strangers will have
an
immediacy which will not carry through
into
the
future-
provided that
the
victim's identity
is
unknown
to
the threatener.'
(Hough,
1990;
p.
175).
Hough
found
that only a quarter of threats were reported
to
the
police. Most
victims, whether or not
they
reported
the
threat, considered a police warning
to
the
offender
to
be
adequate punishment. Nonetheless, Hough
found
the
average
rating of
the
seriousness of threats
to
its
victims
was
higher than that of vandalism
to
and
theft
from
cars,
and
only marginally lower than common assault.
Hough's data
on
the
upset which threats cause
and
the
seriousness with which
they
are
regarded means that
the
phenomenon merits further research attention.
In
this paper, a formal modelling approach
is
taken.
The
probability of being
threatened
is
analysed by sociodemographic
and
lifestyle attributes of individ-
uals,
and
their perception of their future crime victimisation. The analysis extends
to
the
1988
sweep of
the
BCS.
By
examining
the
role of different sets of
explanatory variables,
and
by
incorporating data from three sweeps of
the
BCS,
the
paper seeks
to
give a thorough estimation of threat risks.
The
second (following) section of
the
paper presents
some
introductory infor-
mation
on
the
data employed here, both for threats
and
for
the
explanatory
variables used
in
our modelling. Section 3 discusses
the
factors associated with
threats
in
each of
the
three
BCS
sweeps while section 4 explicitly examines
the
changes between
1982
and
1988.
A concluding section suggests
some
paths
for
future research. Technical details relating
to
the
BCS
information
and
the
logistic
regression technique used
are
contained
in
Appendices 1
and
2 respectively.
Examples of threat risks implied
by
our models for different types of individuals
are
included
in
Appendix
3,
which
also
discusses
how
other probabilities can
be
computed.
THE DATA
Definition and profile of threats in BCS
The
BCS
is
a national victimisation survey carried out
in
England
and
Wales
in
1982, 1984,
1988
and
1992
2. Here
we
use
the
first three sweeps, since data
relating
to
the
fourth
was
unavailable
at
the
time
this analysis
was
undertaken.
To continue reading
Request your trial