Bullying at School: Basic Facts and Effects of a School Based Intervention Program

Full text

/
Child PsychoL
Psychiat.
Vol. 35. No. 7, pp. 1171-1190, 1994
Elsevier Science Ltd
@ jgg^ Association for Child Psychology and Psychiatry-
Printed in Great Britain. All rights reserved
0021-9630/94 $7.00
+
0.00
0021-^9630(94)E0039-I
Annotation: Bullying at School:
Basic Facts and Effects of a School Based
Intervention Program
Dan Olweus
Bullying among schoolchildren is certainly a very old phenomenon. The fact that
some children are frequently and systematically harassed and attacked by other
children has been described in literary works, and many adults have personal
experience of it from their own school days. Though many are acquainted with
the bully/victim problem, it
was
not until fairly recently—in the early 1970s—that
the phenomenon
was
made the object of more systematic research (Olweus, 1973a,
1978).
For a number of
years,
these efforts were largely confined to Scandinavia.
In the 1980s and early 1990s, however, bullying among schoolchildren has attracted
attention also in other countries such as Japan, U.K., The Netherlands, Australia,
Canada, and the U.S.A. There are now clear indications of an increasing societal
as well
as research interest into bully/\ictim problems in several parts of the world.
Overview
The first part of this article gives an overview of research findings on bully/\ictim
problems among schoolchildren. The presentation does not aspire to be exhaustive,
in particular since several of the early Scandina\ian
studies were
of a
very
preliminary
nature, with small sample sizes and no clear definition of what
is
meant
by
bullying.
In addition, they were often conducted by undergraduate students with
little supervision from more experienced researchers. The main research
findings discussed, therefore, are from the following four studies: a
longitudinal project initiated in the early 1970s and comprising some 900 boys
from Greater Stockholm, Sweden (Olweus, 1978) and three large-scale studies
conducted in connection with a nationwide campaign against bully/victim problems
in Norwegian
eleinenUir)'
and secondar)/junior high schools (grades 1-9), launched
Keyxoords:
Bully, victim, intervention, abuse
Accepted manusmpt
received
21 February 1994
Requests
for reprints to: Dan Olweus, Department of Psychosocial Science, Division of Personality
Psychology, University of Bergen, Oisteinsgate 3, N-5007, Bergen, Norway.
1171

1172
D.
Olweus
in 1983 (e.g. Olweus, 1991, 1993b). See Table 1 for an overview of the projects.
Many of the findings to be reported have been replicated in several different
samples and with dififerent metihods.
Although most of the generalizations and conclusions drawn in the article are
derived from this research (e.g. Olweus, 1973a, 1978, 1979, 1980, 1984, 1991,
1992a, 1993a, 1993b; Olweus & Alsaker, 1991),
I will
give reference to other sources,
where appropriate. In this context, it should be mentioned that an excellent
overview of the research area has been recendy completed by Farrington (1993).
There are also other, more practically oriented presentations for example
by
Besag
(1989),
Elliott (1991), Smith
&
Thompson (1991) and an extensive annotated
bibliography of literature and resources (Skinner, 1992).
The second part of the article reports on the positive effects of an intervention
program against bullying that I developed and evaluated over a
2-year
period in
42 schools in Bergen, Norway, in connection with the nationwide campaign. The
Table I. Overview of studies
Nationwide study
in Norway (1983)
Units of 715 Schools,
study grades 2-9
(130,000 boys and
girls)
Number of One
measurement
occasions:
Measures Questionnaire on
include: bully/victim
problems
(aggregated to
grade and school
level)
data on
recruitment area
of the school:
population
density.
socioeconomic
conditions, per
cent immigrants
School size.
average class size
Composition of
staff
Large-scale study
in Sweden
(1983/84)
60 Schools, grades
S-9 (17,000 boys
and girls)
One
Questionnaire on
bully/victim
problems
School size.
average class size
Intensive study in
Bergen, Norway
(1983-1985)
Four cohorts of
2500 boys and
girls in grades
4-7 (1983)
3-400
teachers
1000 parents
Several
Self-reports on
bully/victim
problems.
aggression.
antisocial
behavior, anxiety,
self esteem.
attachment to
parents and peers.
etc.
grades, some
peer ratings
Teacher data on
characteristics of
class,
group
climate, staff
relations, etc.
Study in Greater
Stockholm,
Sweden (1970- )
Three cohorts of
boys (900 boys in
all),
originally in
grades 6-8 (1973)
Several
Self-reports and
reports by
mothers on a
number of
dimensions
peer ratings
teacher
nominations
Official records on
criminal offences.
drug abuse for
subgroups:
interviews on
early child
rearing, hormonal
data, psychophysi-
ological data

Bullying at School 1173
content of the program and the principles on which it is based, are also briefiy
described.
Definition of bullying
I define bullying or victimization in the following general way: A student is being
bullied or victimized when he or she
is
exposed,
repeatedly
and
over
time,
to negative actions
on
the
part
of
one
or
more
other
students.
It
is
a negative action when someone intentionally
infiicts, or attempts to infiict, injury or discomfort upon another—basically what
is implied in the definition of aggressive behavior (Olweus, 1973b). Negative actions
can be carried out by physical contact, by words, or in other
ways,
such as making
faces or obscene gestures, and intentional exclusion from a group.
In order to use the term bullying, there should also be an
imbalance
in
strength
(an asymmetric power
relationship):
the student who is exposed to the negative actions
has difficulty in defending him/herself and
is
somewhat helpless against the student
or students who harass.
Bullying can be carried out by a single individual—the bully—or by a group.
The target of bullying can also be a single individual—the victim—or a group. In
the context of school bullying, the target has usually been a single student. Data
from the Bergen study indicate that, in the majority of
cases,
the victim is harassed
by a group of two or three students. A considerable proportion of the victims, some
35-40%,
report however, that they are mainly bullied by one student (Olweus,
1988).
In my definition, the phenomenon of bullying is thus characterized by the
following three criteria: (a) It is aggressive behavior or intentional "harmdoing"
(b) which is carried out "repeatedly and over time" (c) in an interpersonal
relationship characterized by an imbalance of power. One might add that the
bullying behavior often occurs without apparent provocation. This definition
makes it clear that bullying can be considered a form of abuse, and I sometimes
use the term peer
abuse
as a label for the phenomenon. What sets it apart from
other forms of abuse such as child abuse and wife abuse is the context in which it
occurs and the relationship characteristics of the interacting parties.
It is useful to distinguish between
direct
bullying/victimization—with relatively
open attacks on the victim—and
indirect bullying/victimization
in the form of social
isolation and intentional exclusion from a group.
In the present article the expressions bullying, victimization
2iad
bully/victim problems
are used synonymously.
Basic Facts About Bully/Victim Problems
Prevalence
On the basis of a survey of more than 130,000 Norwegian students with the
Bully/Victim Questionnaire (Olweus
&c
Smith, in press), one can estimate that
some 15% of the students in elementary and secondary/junior high schools (grades
1-9, roughly corresponding to ages 7-16) in Norway were involved in bully/victim
problems with some regularity—either as bullies or \dctims (Olweus, 1985, 1987,
1991,
1993b). This figure corresponds to 84,000 students (autumn 1983).

1174
D.
Olweus
Approximately 9%, or 52,000 students, were victims, and 41,000, or 7%, bullied
other students regularly. Some
9,000
students were both victim and bully (1.6%
of the total of
568,000
students or 17% of the victims). A total of some 5% of the
students were involved in more serious bullying problems (as bullies or victims or
bully/victim), occurring "about once a week" or more frequently.
Analyses of parallel teacher nominations in approximately 90 classes (Olweus,
1987) suggest that these results do not
give
an exaggerated picture of the prevalence
of bully/victim
problems*.
Indeed,
as
both the student and the teacher questionnaires
refer only to part of the autumn term, there is little doubt that the figures
underestimate the number of students who are involved in such problems during
a whole year.
It
is
apparent, then, that bullying
is
a considerable problem in Norwegian schools
and affects a very large number of students. Data from other countries (in large
measure collected with the Bully/Victim Questionnaire) such as Sweden (Olweus,
1992b), Finland (Lagerspetz, Bjorkqvist, Berts & King, 1982), England (Smith,
1991;
Whitney
&
Smith, 1993), U.S.A. (Perry, Kusel
Sc
Perry, 1988), Canada (Ziegler
Sc
Rosenstein-Manner, 1991), The Netherlands (Haeselager
8c
van Lieshout, 1992;
Junger, 1990),Japan (Hirano, 1992), Ireland (O'Moore
8c
Brendan, 1989), Spain
(Ruiz, 1992), and Australia (Rigby
&
Slee, 1991), indicate that this problem certainly
exists outside Norway, with similar or even higher prevalence rates.
Bully/victim
problems
in
different grades
As
seen in
Fig.
1,
the percentage of students
who
reported being bullied decreased
with higher
grades.
It was the younger and weaker students who were most exposed.
With regard to the ways in which the bullying was carried out, there was a clear
trend towards less use of physical means (physical \iolence) in the higher grades.
In the Bergen study it
was
also found that a considerable part of the bullying was
carried out by older students. This was particularly marked in the lower grades:
More than
50%
of the bullied children in the lowest grades (2 and
3,
corresponding
to ages 8 and 9 in this survey) reported that they were bullied by older students.
It is natural to invoke the latter finding at least as a partial explanation of the
form of the curves in Fig. 1. The younger the students are, the more potential
bullies they have above them; accordingly, an inverse relationship between percentage
of victims and grade level seems reasonable. It may also be the case that the form
of the curves refiects the possibility that a certain proportion of the victims are
able gradually to develop strategies for escaping bullying, as they grow older. In a
similar
vein,
it can be argued that a certain proportion of the students may become
With regard to the validit)' of self-reports on variables related to bully/victim problems, in the
early Swedish studies (Olweus, 1978) composites of
3—5
self-report items on being bullied or btillying
and attacking
others,
respectively, correlated in the .40-60 range (Pearson correlations) with reliable
peer ratings on related dimensions (Olweus, 1977). Perry, Kusel and Perry (1988)reported a correlation
of
.42
between
a
self-report scale of three victimization items and a reliable measure of peer nominations
of victimization in elementary schoolchildren. In the intervention study (below), we found class-
aggregated student rating estimates to be highly correlated
(.60-.70)
with class-aggregated estimates
derived from the students' own reports of being bullied or bullying others (see Olweus, 1991, for
details).

Bullying
at
School1175
GIRLS
PERCENT
VICTIMS
BOYS
I 17.5
GIRLS!
16
BOYS
Fig.
1.
Percentage
of
students
in
different grades
who
reported being bullied (being exposed
to
direct
bullying).
(A'
for
boys
=
42,390:
N for
girls
=
40,940).
less vulnerable with increasing
age and,
accordingly, will report being bullied less.
All of these explanations
(and
perhaps additional ones) maybe partly correct,
and
more detailed analyses
of the
factors affecting
the
shape
of the
curves
are to be
undertaken.
As regards
the
tendency
to
bully other students, depicted
in
Fig.
2, the
changes
with grades
are not
so clear
and
systematic
as in
Fig. 1.
The
relatively marked drop
in
the
curves
for
grade
7
(around
age
13),
in
particular
for
boys,
may partly reflect
the fact that these students were
the
youngest ones
in
their schools
and
accordingly
did
not
have "access
to
suitable victims"
in
lower grades
to the
same extent.
(The
majority
of
Norwegian students make
a
transfer from primary schools
to
separate
secondary/junior high schools with
the
start
of
grade
7.)
Bullying among
boys
and
girLs
u n •
As is evident from Fig.
1,
there is
a
trend
for
boys
to be
more exposed
to
bullying
than girls. This tendency
is
particularly marked
in the
secondary/junior high
school srrades.
. , , . , ^^ i
o
These results concern what
was
called direct bullying, with relatively open attacks

1176
D.
Olweus
PERCENT BULLIES
Fig. 2. Percentage of studetits in different grades who reported having bullied other students (N for boys
=
42,423;
A^
for girls = 40,877).
on the
victim.
It
is
natural to ask whether girls were more often exposed to indirect
bullying in the form of social isolation and intentional exclusion from the peer
group. Analyses of the Questionnaire data confirm that girls were more exposed
to indirect and more subtle forms of bullying than to bullying with open attacks.
At the same time, however, the percentage of
boys
who were bullied in this indirect
way was approximately the same as that for girls. In addition, a somewhat larger
percentage of
boys was
exposed to direct bullying, as mentioned
above.
(Moreover,
there
was
a fairly strong association between being
a
victim of direct and of indirect
bullying.)
An additional result from the Bergen study is relevant in this context. Here it
was found that boys carried out a
large
part of
the
bullying to which girls
were
subjected.
More than 60% of bullied girls (in grades 5-7) reported being bullied mainly by
boys.
An additional 15-20% said they were bullied by both boys and girls. The
great majority of
boys,
on the other hand—more than 80%—^were bullied chiefly
by boys.
Figure 2 shows the percentage of students who had taken part in bullying other
students with some regularity. It
is
evident here that a considerably larger percentage
of
boys
than girls had participated in bullying. In secondary/junior high school,
more than four times as many
boys
as girls reported having bullied other students.

Bullying at School 1177
It should also be reported that bullying by physical means was more common
among
boys.
In contrast, girls often used more subtle and indirect
ways
of harassment
such as slandering, spreading of rumours and manipulation of friendship
relationships (e.g., depriving a girl of her "best friend"). Nonetheless, harassment
with non-physical means (words, gestures, etc) was also the most common form
of bullying among boys.
In summary,
boys were more often
victims and in particular
perpetrators
of
direct
bullying.
This conclusion is in good agreement with what can be expected from research
on sex differences in aggressive behavior (Maccoby & Jacklin, 1974,1980; Ekblad
8c
Olweus, 1986). It is well documented that relations among boys are by and large
harder, tougher, and more aggressive than among girls (Maccoby, 1986). These
differences certainly have both biological and social/emdronmental roots.
The results presented here should by no means be construed as implying that
we do not need to pay attention to bullying problems among girls. As a matter of
course, such problems must be acknowledged and counteracted, whether girls are
the victims of bullying or they themselves perpetrate such beha\dor.
The pattern of results found in the Norwegian data have been replicated in all
essentials in the corresponding analyses with Swedish students (Olweus, 1992b)
and with students in the Sheffield area (though the levels of problems were somewhat
higher for these students; see Whitney & Smith, 1993). For a number of other
research findings on bullying, concerning, for example, teacher and parent
awareness of the problems, bullying on the way to and from school, and levels of
problems in "big-cit>" versus small-town schools, see Olweus (1993b).
Three
common
''myths "
about bullying
A common view holds that bully/\ictim problems are a consequence of large
classes and/or schools: the larger the class or the school, the higher the level of
bully/victim problems. Closer analysis of this hypothesis, making use of the
Norwegian survey data from more than 700 schools and several thousand classes
(with great variations in size), gave clear-cut results: There were no positive
associations between level of bully/victim problems (the percentage of bullied
and/or bullying students) and school or class size. Thus, the size of the class or
school appears to be of negligible importance for the relative frequency or level
of bully/victim problems (Olweus, 1993b).
It
is,
nevertheless, a fact that the
absolute
number of: huWmd and bullying students
is greater on average in big schools and in big classes. One might therefore think
that it would be somewhat easier to do something with the problems in a small
school or a small class. Analyses of data from the Bergen study conducted so far,
however, do not support this assumption. It may be added, that international
research on the "effects" of class and school size agrees in suggesting that, in
general, these factors are of no great significance, at least within the ranges of size
variation typically found (e.g., Rutter, 1983).
In the general debate in Scandinavia (and probably elsewhere) it has been
commonly maintained that bullying is a consequence of competition and striving
for grades in school. More specifically, it has been argued that the aggressive

1178
D.
Olweus
behavior of the bullies toward their environment can be explained as a reaction
to failures and frustrations in school. Such ideas are in fact central elements in
many criminological theories. A detailed LISREL analysis of data on 444 boys in
the Swedish study (Olweus, 1983), who were followed from grade 6 to grade 9
(from ages 13 to 16), gave no support at all to these ideas. Though there was a
moderate association (r around .30) between poor grades in school and aggressive
behavior, both in grade
6
and in grade
9,
there
was
nothing in the results to suggest
that the behavior of the aggressive
boys was
a consequence of poor grades and failure
in school. A similar finding, but more specifically concerned with delinquent
behavior, has recently been reported by a Canadian research group (Tremblay, et
al., 1992).
Third, a widely held view explains victimization as caused
by
external deviations.
It is argued that students who are fat, red-haired, wear glasses, or speak with an
unusual dialect, etc. are particularly likely to become victims of bull)ing. Again,
this hypothesis received no support from empirical data. In two samples of boys,
victims of bullying
were by
and large found to be no more externally deviant (with
regard to 14 external characteristics assessed by means of teacher ratings) than a
control group of boys who were not exposed to bullying (Olweus, 1973a, 1978).
It was concluded that external deviations play a much smaller role in the origin
of bully/victim problems than generally assumed (see also Junger, 1990). In spite
of the lack of empirical support for this hypothesis, it
still
seems to enjoy considerable
popularity. Some probable reasons why this is so have been advanced, and the
interested reader is referred to this discussion (Olweus, 1978, 1993b).
All of these hypotheses have thus failed to receive support from empirical data.
Accordingly, one must look for other factors to find the origins of these problems.
The research evidence collected so far and presented in the next few pages clearly
suggests that personality characteristics/typical reaction patterns, in combination
with physical strength or weakness in the case of boys, are important for the
development of these problems. At the same time, other factors such as the teachers'
attitudes and behavior play a major role in determining the extent to which the
problems
will
manifest themselves in a classroom or a school (see Olweus, 1993b).
What
characterizes
the typical victims
?
A relatively clear picture of both the typical victims and the t^ical bullies has
emerged from research (e.g. Olweus, 1973a, 1978, 1981, 1984; Bjorkqvist, Ekman
8c
Lagerspetz, 1982; Boulton
8c
Smith, in press; Farrington,
1993;
Lagerspetz
et
al.,
1982;
Perry
et
al.,
1988).
By
and large, this picture seems to apply to both boys and
girls,
although it must be emphasized that much less research has so far been done
on bullying among girls.
The typical victims are more anxious and insecure than other students in general.
Further, they are often cautious, sensitive and quiet. When attacked by other
students, they commonly react
by
crying (at least in the lower
grades)
and withdrawal.
Also,
victims suffer from low self-esteem, they have a negative view of themselves
and their situation; they often look upon themselves as failures and feel stupid,
ashamed and unattractive.

Bullying at School 1179
The victims are lonely and abandoned at school. As a rule, they do not have a
single good friend in their
class.
They are not aggressive or teasing in their behavior,
however,and,accordingly,one cannot explain the bullying as a consequence of the
victims themselvesbeingprovocative totheir peers (see below) .These children often
have a negative attitude toward violence and use of violent means. If they are boys,
they are likely to be physically weaker than other boys (Olweus, 1978).
I have labelled this type of victim
the passive or submissive
victim,
as opposed to
the far less common type described below. In summary, it seems that the behavior
and attitude of the passive/submissive victims signal to
others
that
they are insecure
and worthless individuals who will not retaliate if they are attacked or
insulted.
A slightly
different way of describing the passive/submissive victims is to say that they are
characterized by an anxious or submissive reaction pattern combined (in the case of
boys) with physical
loeakness.
In-depth interviews with parents of victimized boys indicate that these boys were
characterized by a certain cautiousness and sensitivity from an early age (Olweus,
unpublished, 1993a). Boys with such characteristics (perhaps combined with
physical weakness) are likely to have had difficulty in asserting themselves in the
peer group and may have been somewhat disliked by their age mates. There are
thus good reasons to believe that these characteristics contributed to making them
victims of bullying (see also Schwartz, Dodge
8c
Coie, 1993). At the same time, it
is obvious that the repeated harassment by peers must have considerably increased
their anxiety, insecurity, and generally negative evaluation of themselves.
Some of our data also indicate that the victimized boys had closer contact and
more positive relationships with their parents, in particular with their mothers,
than boys in general. This close relationship is sometimes perceived by teachers
as overprotection on the part of the mothers (Olweus, 1978,1993a). It
is
reasonable
to assume that such tendencies toward overprotection are both a cause and a
consequence of the bullying.
As mentioned earlier, there is also another, clearly smaller group of victims, the
provocative
victims,
who are characterized by a combination of both anxious and
aggressive reaction patterns. These students often have problems with concentration,
and behave in ways that may cause irritation and tension around them. Some of
these students can be characterized as hyperactive. It is not uncommon that their
behavior provokes many students in the class, thus resulting in negative reactions
from a large part of, or even the entire
class.
The dynamics of bully/victim problems
in a class with provocative victims differ in part from problems in a
class
with passive
victims (Olweus, 1978).
A follow-up study of two groups of boys (Olweus, 1993a) who had or had not
been victimized by their peers in school (from grades 6 through 9) shows that the
former victims had "normalized" in many
ways
as young adults, at age
23.
This was
seen as an indication that the boys, after having left school, had considerably
greater freedom to choose their own social and physical environments. In two
respects, however, the former victims had fared much worse than their non-
victimized peers: they were more likely to be depressed and had poorer self-esteem.
The pattern of findings clearly suggested that this was a consequence of the earlier.
Persistent victimization which thus had left its scars on their minds.

1180
D.
Olweus
The hypothesis that bullying
may
have
a
causal effect
on the
victims'
self-
evaluations was also supported
by
certain data from
the
Bergen study.
In
these
analyses (Olweus
&:
Alsaker, in
press),
self-reported increases
or
decreases in bullying
over time
(a
2-year period) were accompanied
by
parallel changes
in
global
self-
esteem,
as
measured by
a
Rosenberg-like scale.
What
characterizes
the
typical
bullies ?
A distinctive characteristic of the typical bullies
is
their aggression toward peers—
this is implied
in the
definition of a bully. But bullies
are
often aggressive towards
adults as well, both teachers
and
parents. Generally, bullies have
a
more positive
attitude towards violence than students
in
general. Further, they
are
often
characterized by impulsivity and
a
strong need
to
dominate others. They have little
empathy with victims
of
bullying.
If
they
are
boys, they
are
likely
to be
physically
stronger than boys
in
general,
and the
victims
in
particular (Olweus, 1978).
A commonly held view among psychologists
and
psychiatrists is that individuals
with
an
aggressive
and
tough behavior pattern
are
actually anxious
and
insecure
"under the surface". The assumption that the bullies have
an
underlying insecurity
has been tested
in
several
of
my own studies, also using "indirect" methods such
as stress hormones (adrenaline
and
noradrenaline)
and
projective techniques.
There
was
nothing in
the
results
to
support the common
view,
they rather pointed
in
the
opposite direction:
the
bullies
had
unusually little anxiety
and
insecurity,
or were roughly average
on
such dimensions (Olweus, 1981, 1984, 1986;
see
also
Pulkkinen
&
Tremblay, 1992). They
did not
suffer from poor self-esteem.
These conclusions apply
to the
bullies
as a
group
(as
compared with groups
of
control boys and
victims).
The results do not imply that there cannot be individual
bullies who
are
both aggressive
and
anxious.
It should also
be
emphasized that there are students who participate
in
bullying
but who
do not
usually take
the
initiative—these
may be
labelled
passive
bullies,
followers,
or
henchmen.
A
group of passive bullies is likely to
be
fairly mixed
and
may
also contain insecure
and
anxious students (Olweus, 1973a, 1978).
Several studies have found bullies
to be of
average
or
slightly below average
popularity (Bjorkqvist et al., 1982; Lagerspetz, et al., 1982; Olweus, 1973a,
1978;
Pulkkinen
&:
Tremblay, 1992). Bullies
are
often surrounded
by a
small group
of
two-three peers
who
support them
and who
seem
to
like them (Cairns, Cairns,
Neckerman, Gest
8c
Gariepy,
1988).
The popularity of the bullies decreases, however,
in
the
higher grades
and
is considerably less than average
in
grade
9
(around
age
16).
Nevertheless,
the
bullies
do not
seem
to
reach
the
low level of popularity that
characterizes
the
victims.
In summary,
the
typical bullies
can be
described
as
having
an
aggressive reaction
pattern
combined
(in the
case
of
boys) with
physical
strength.
It should
be
noted that
not all
highly aggressive children
or
youths
can be
classified as bullies.
In an
as yet unpublished study
of
one
of
the Swedish cohorts,
the boys belonging
to the
upper 15-20%
of the
distribution
of
peer ratings
of
aggressive behavior
("Starts
Fights"
and
"Verbal Hurt"
in
grades 6)-9,
see
Olweus,
1980) were examined
as to
whether they
had
been identified
as
bullies
by
their

Bullying at School 1181
teachers (Olweus, 1978). Depending upon the cut-off point chosen, between 40
and
60%
of the boys rated as "highly
aggressive"
were nominated
as
bullies
by
their
teachers. Exploratory comparisons between the teacher-nominated bullies and
the aggressive non-bullies suggested interesting differences. Because of their
preliminary nature, they will not be reported on here, but they clearly indicate
the need for more detailed analyses of possible differences between students who
are identified as bullies and other aggressive youngsters.
As regards the possible psychological sources underlying bullying behavior, the
pattern of empirical findings suggests at least three, partly interrelated motives.
First, the bullies have a strong need for power and dominance; they seem to enjoy
being "in control" and to subdue others. Second, considering the family conditions
under which many of them have been reared (see below), it is natural to assume
that they have developed a certain degree of hostility towards the environment;
such feelings and impulses may make them derive satisfaction from infiicting injury
and suffering upon other
individuals.
Finally, there
is
an "instrumental component"
to their behavior. The bullies often coerce their victims to provide them with
money, cigarettes, beer, and other things of value (see also Patterson, Littman
8c
Bricker, 1967). In addition, it
is
obvious that aggressive behavior
is
in many situations
rewarded in the form of prestige (e.g. Bandura, 1973).
Bullying can also be viewed as a
component
of a
more generally antisocial
and
rule-
breaking
("conduct-disordered")
behavior pattern. From this perspective, it is natural to
predict that youngsters who are aggressive and bully others run a clearly increased
risk of later engaging in other problem behaviors such as criminality and alcohol
abuse.
A
number of sttidies confirm this general prediction (e.g., Loeber
8c
Dishion,
1983;
Magnusson, Stattin
8c
Duner, 1983).
The Swedish follow-up studies have also found strong support for this view.
Approximately 60% of boys who were characterized as bullies in grades 6-9 (on
the basis of a combination of teacher nominations and peer ratings) had been
convicted of at least one officially registered crime by the age of
24.
Even more
dramatically, as much as 35-40% of the former
bullies
had three or more convictions
by this age, while this was true of only 10% of the control boys (those who were
neither bullies nor victims in grades 6-9). Thus,
as
young
adults,
the former school
bullies had a 4-fold increase in the level of relatively serious, recidivist criminality
as documented in official crime records (Olweus, 1993b).
It may be mentioned, that the former
victims
had an average or somewhat below
average level of criminality in young adulthood.
Development of an
aggressit>e reaction
pattern
In light of the characterization of the bullies as having an aggressive reaction
pattern—that is, they display aggressive behavior in many situations—it becomes
important to examine the question: What kind of rearing and other conditions
during childhood are conducive to the development of an aggressive reaction
pattern? Veiy briefly, the following four factors have been found to be particularly
important (based chiefly on research with boys; for details, see Olweus, 1980; see
also Loeber
8c
Stouthamer-Loeber, 1986):
(1) The basic emotional attitude of the primary caretaker (s) toward the child

1182
D.
Olweus
during early years (usually the mother). A negative emotional attitude,
characterized by lack of warmth and involvement, increases the risk that the
child will later become aggressive and hostile toward others.
(2) Permissiveness towards aggressive behavior by the child. If the primary
caretaker is generally permissive and tolerant, without setting clear limits
on aggressive behaviors towards
peers,
siblings and
adults,
the child's aggression
level is likely to increase.
(3) Use of power-assertive child-rearing methods such as physical punishment
and violent emotional outbursts. Children of parents who make frequent
use of these methods are likely to become more aggressive than the average
child. "Violence begets violence".
We can summarize these results by stating that
too little love
and
care
and
too much
"freedom"
in
childhood are conditions
that
contribute strongly
to the
development
of an
a^ressive
reaction
pattern.
(4) Finally, the temperament of the child.
A
child with an active and "hot-headed"
temperament is more likely to develop into an aggressive youngster than a
child with a quieter temperament. The effect of this factor is less powerful
than those of the two first-mentioned conditions.
The factors listed above can be assumed to be important for both younger and
somewhat older children. For adolescents, it is also of great significance whether
the parents supervise the children's activities outside the school (Patterson, 1986;
Patterson
8c
Stouthamer-Loeber, 1984) and check on what they are doing and with
whom.
It should also be pointed out that the aggression levels of the boys participating
in the analyses above (Olweus, 1980) were
not
related to the socioeconomic conditions
of their families, measured in several different
ways.
Similarly, there were no (or
only very weak) relations between the four childhood factors discussed and the
socioeconomic conditions of the familv (Olweus, 1981).
Som£
group mechanisms
When several students jointly engage in the bullying of another student, certain
group mechanisms are likely to be at work. Several such mechanisms have been
discussed in detail in Olweus (1973a, 1978). Because of space limitations, they are
only listed
here:
(1) Social "contagion"; (2) Weakening of the control or inhibitions
against aggressive tendencies; (3) "Diffusion of responsibility"; and (4) Gradual
cognitive changes in the perceptions of bullying and of the victim.
A
question
offundam£ntal
democratic rights
The victims of bullying form a large group of students who are to a great extent
neglected by the school. We know that many of these youngsters are the targets
ofharassmentforlongperiodsof
time,
often for many years (Olweus, 1977,1978).
It does not require much im agination to understand what it is to go through the
school years in a state of more or less permanent anxiety and insecurity and with
poor self-esteem. It is not surprising that the victims' devaluation of themselves

Bullying
at
School
1183
sometimes becomes so overwhelming that they see suicide as the only possible
solution.
Bully/victim problems in school really concern some of our basic values and
principles. In my view, it is a fundamental
democratic
right for a child to feel
safe
in
school
and to
be
spared the
oppression
and
repeated,
intentional humiliation implied in bullying.
No student should be afraid of going to school for fear of being harassed or
degraded, and no parent should need to worry about such things happening to
his or her child!
Even though an intervention program against bully/victim problems in school
is particularly important in order to reduce the suffering of the victims, it is also
highly desirable to counteract these problems for the sake of the aggressive students.
As reported above, school bullies are much more likely than other students to
follow an antisocial path. Accordingly, it
is
essential to try to redirect their activities
into more socially acceptable channels.
Effects of
a
School Based Intervention Program
Against this background, it is now appropriate briefly to describe the effects of
the intervention program that I developed and evaluated in connection with the
campaign against bully/victim problems in Norwegian schools. For more details
about the program and its evaluation, the reader is referred to the book
Bullying
at
school—what we know
and
what we can do
(Olweus, 1993a) and to other sources
(e.g. Olweus, 1991, 1992a; Olweus
8c
Alsaker, 1991).
So far as is known, this is the first program against bully/victim problems that
has been evaluated in systematic research. It should be noted, however, that there
is at present at least one project in the United Kingdom, led by Peter Smith at the
University of Sheffield (e.g.. Smith
8c
Sharp, in press; Sharp
&
Smith, 1991), m
which an antibullying program is being systematically evaluated. In several ways,
this program
is
related to and actually based on the program developed in Bergen,
while it is at the same time somewhat different in other respects.
The major goals of the Norwegian program were to reduce as much as possible
existing bully/victim problems and to prevent the development of
new
problems.
Subjects
and design
Evaluation of the effects of the intervention program was based on data from
approximately 2500 students originally belonging to 112 grade 4-7 dasses m 42
primary and secondaiy/junior high schools in Bergen (modal ages at Time
1
were
11,12,13 and
14
years,
respectively). Each of the four grade/age cohorts consisted
of 600-700 subjects with a roughly equal distribution of boys and girls. The tirst
time of data collection (Time 1) was in late May 1993, approximately 4 months
before introduction of the intervention program. New measurements were taken
in May 1984 (Time 2) and May 1985 (Time 3). The design employed is usually
called a selection-cohorts design. s . , j ^ <- ^
Data were analyzed with ANOVA (analysis of variance) with students nested
within classes nested within schools nested within times/occasions (Time
1
versus

1184
D.
Olweus
Time 2, Time
1
versus Time 3). Sex of the subjects
was
crossed with times, schools
(within times), and classes (within schools).
Main
results
The main
findings
of the analyses can be summarized as follows:
• There were marked reductions in the levels of bully/victim problems (for
both direct and indirect bullying, and for bullying others) for the periods
studied,
8
and
20
months of intervention, respectively.
By
and
large,
reductions
were obtained for both boys and girls and across all cohorts compared.
• Similar reductions were obtained for the aggregated peer rating variables
"Number of students being bullied in the class" and "Number of students
in the class bullying other students". There was thus consensual agreement
in the classes that bully/victim problems had decreased considerably during
the periods studied.
• In terms of percentages of students reporting being bullied or bullying others
"now
and then" or more fi-equently, the reductions amounted to approximately
50%
or more in most comparisons.
• There was no displacement of bullying from the school on the way to and
from school. There were reductions or no changes as regards bully/victim
problems on the way to and from school.
• There
was
also a clear reduction in general antisocial behavior (see Olweus,
1989,
for details about the construction of the 23-item scale) such
as
vandalism,
fighting, pilfering, drunkenness, and truancy.
• In addition,
we
could register marked improvement
as
regards various aspects
of the "social climate" of the class: improved order and discipline, more
positive social relationships, and a more positive attitude to schoolwork and
the school.
• At the same time, there was an increase in student satisfaction with school
life as refiected in "liking recess time".
• For several of the variables, the effects of the intervention program were
more marked after 2 years than after
1
year.
• The intervention program not only affected already existing victimization
problems; it also reduced the number (and percentage) of
new
victims
(Olweus, 1989, 1992a); the program had thus both primary and secondaiy
prevention effects (Cowen, 1984).
In the majority of comparisons for which reductions were reported above, the
differences between base line and intervention groups were significant/highly
significant and with medium, large, or even very large effect sizes (a rf-value of
more than 1.0 for several variables; see Cohen, 1977).
A detailed analysis of the quality of the data and the possibility of alternative
interpretations of the findings led to the following general statements (Olweus,
1991):
It is very difficult to explain the results obtained as a consequence of (a)
underreporting by the students; (b) gradual changes in the students' attitudes to

Bullying at School 1185
bully/victim problems; (c) repeated measurement; and (d) concomitant changes
in other factors, including general time trends (Olweus, 1994). All in all, it was
concluded that the changes in bully/victim
problems
and
related
behavior patterns
were likely
to be mainly a consequence of
the
intervention program and not of
som£
other "irrelevant"
factor.
It was also noted that self-reports, which were implicated in most of the
analyses conducted so far, are probably the best data source for the purposes of
this study. At the same time largely parallel results were obtained for the peer
rating variables mentioned above and for teacher ratings of bully/victim problems
at the class level; in the latter case, however, the effects were somewhat weaker.
In addition, a clear "dosage-response" relationship has been established in
preliminary analyses at the class level (which is the natural unit of analysis in this
case).
Those classes that showed larger reductions in bully/victim problems had
implemented three presumably essential components of the intervention program
(including establishment of class rules against bullying and use of regular class
meetings) to a greater extent than those with smaller changes (additional information
on these analyses can be found in Olweus
8c
Alsaker, 1991). This finding provides
corroborating evidence for the hypothesis that the changes observed were due to
the intervention program.
The reported effects of the intervention program must be considered quite
positive in particular since many previous attempts to systematically reduce aggressive
and antisocial behavior in preadolescents/adolescents have been relatively
unsuccessful (e.g., Dumas,
1989;
Gottfredson,
1987;
Kazdiri,
1^87).
The importance
of the results is also accentuated by the fact that there has been a highly disturbing
increase in the prevalence of violence and other antisocial behavior in most
industrialized societies in the past decades. In the Scandinavian countries, for
instance, various fonns of registered criminality
have
typically increased
by
400-600%
since the 1950s or 1960s (see e.g.,
Stortingsmelding
nr.
23
1991-1992,
1992, for the
development in Norway).
As
mentioned in the first part of the article,
we
can estimate that approximately
80,000 students in Norwegian schools were involved in bully/victim problems (in
1983).
On the basis of the reported results the following conclusion can be drawn:
if all primary and secondary/junior high schools in Norway used the intervention
program the way it was used in Bergen, the number of students involved in
bully/victim problems would be reduced by 40,000 or more in a relatively short
period of
time.
Effective use of the intervention program would also have a number
of additional positive effects, as indicated above.
Basic
principles
The intervention program is built around a
limited set
of
key principles
derived
chiefly from research on the development and modification of the problem
behaviors concerned, in particular aggressive behavior. It
is
considered important
to try to create a school (and ideally, also home) environment charactenzed by
warmth, positive interest, and involvement from adults, on one hand, and firm
limits to unacceptable behavior, on the other. Third, in cases of violations of limits

1186 D. Olweus
and
rules,
nonhostile, nonphysical sanctions should be consistendy
applied.
Implied
in the latter two principles is also a certain degree of monitoring and surveillance
of the students' activities in and out of school (Patterson, 1982, 1986). Finally,
adults are supposed to act as authorities at least in some respects. In a sense, the
intervention program
is
based on an authoritative adult-child interaction, or child
rearing, model (cf e.g., Baumrind, 1967) apphed to the school setting.
The principles listed above were translated into a number of specific measures
to be used at the
school,
class,
and individual
levels.
It is considered important to
work on all of these levels, if
possible.
Space limitations prevent a description of
the various measures but such an account can be found in the book
Bullying
at
school—what we know
and
what we can do
(Olweus, 1993b) t. Table 2 lists a set of core
components which are considered, on the basis of statistical analyses and our
experience with the program, to be particularly important in any implementation
of the program.
Table 2. Overview of Core Program
General Prerequisites
+
+
Awareness and involvement on the part of adults
Measures at the School Level
+ + Questionnaire survey
+
+
School conference day
+
+
Better supervision during recess
+ Formation of
a
coordinating group
+ Meeting staff-parents (PTA meeting)
Measures at the Class Level
+ +
Class rules against bullying
+
+
Class meetings
Measures at the Individual Level
+ + Serious talks with hullies and victims
+ + Serious talks with parents of involved students
+ Teacher and parent use of imagination
+ +Core component; + highly desirable component.
With regard to implementation and execution, the program is mainly based on
a utilization of the existing social environment: teachers and other school personnel,
students, and parents. Nonmental health professionals thus play a major role in
the desired
""restructuring
of
the
social
environment"".
"Experts" such as school
psychologists, school counsellors, and social workers also serve important functions
fThe updated "package" related to the intervention program consists of the book Bully/Victim
Questionnaire (Olweus
&
Smith, in press), and a 20-minute video cassette showing scenes from the
everyday lives of two bullied children (with English subtitles; this video can be ordered from the
author).

Bullying at School 1187
such as planning and coordinating, counselling teachers and parents (groups),
and handling more serious cases.
It should be emphasized that this core program in many ways represents what
is usually called "a whole school policy approach to bullying" in the recent English
literature. It consists of a set of routines, rules, and strategies of communication
and action for dealing with existing and future bullying problems in the school.
Possible reasons for the effectiveness of the program have been discussed in some
detail in Olweus (1992a). They include a change of the "opportunity" and the
"reward structures" for bullying behavior. It should also be emphasised that
bully/victim problems are an excellent entry point for dealing with a variety of
problems that plague today's schools.
It may be added that the intervention program has been evaluated by more than
1000 Norwegian and Swedish teachers (Manger & Olweus, 1985). In short, their
reactions have generally been quite favorable, indicating, among other things,
that the teachers see the proposed principles and measures as useful and realistic.
Concluding words
As is hopefully shown in this article, there is now available a good deal of reliable,
research based knowledge about bully/victim problems in school. At the same
time,
it must be recognized that the empirical base on which robust conclusions
can be built is still somewhat limited in terms of methodology, contextual and
cultural variation, and other important parameters. This is particularly true with
regard to efforts to counteract and intervene systematically against the problems.
In the years to come, a number of issues in the area of bully/victim problems will
have to be dealt with in greater detail, with more methodological diversity, and
under more varied contextual and cultural conditions. Because the phenomenon
of bullying is of considerable interest to many developmental, educational and
child psychiatric researchers, and has many important practical and societal
implications, I feel fairly confident that the next decade will see a good deal of
solid research along these lines.
Acknowledgements—The
research reported in this chapter was supported by grants from the William
T. Gran