doi:10.1111/camh.

12261 Bullying, mental health and friendship 3

relationships, hyperactivity/inattention and prosocial. The

Methods SDQ has high reliability (Cronbach’s a = .73), test–retest stabil-
ity and external validity predicting clinical diagnosis (Hawes &
Setting and participants Dadds, 2004; Mellor, 2005). The present study focused on the
Children were recruited into the Childhood to Adolescence emotional subscale for child internalising symptoms (e.g. ‘Many
Transition Study (CATS). The CATS design and methodology worries or often seems worried’), and the conduct problems sub-
are briefly outlined here, as they are detailed in a protocol publi- scale for externalising symptoms (e.g. ‘My child often loses their
cation (Mundy et al., 2013). Children were recruited via primary temper’). Subscales are comprised of five items with response
schools in Melbourne, Australia. Schools were first randomly options of 1 (not true), 2 (somewhat true) or 3 (certainly true),
selected from a stratified sample (Catholic, Independent, with scores totalled for subscales. Australian norms (7–
Government) across metropolitan Melbourne. In total, 101 11 years) for the emotional symptoms subscale are M = 2.3
schools were approached to participate and 43 consented (43% (SD = 2.0) (Mellor, 2005). Clinical significance bands are 0–3
uptake). In the CATS, all grade 3 students at participating pri- (not significant), 4 (slightly raised risk) and 5–10 (high risk).
mary schools (2289) received a recruitment package with an Australian norms for the conduct subscale are M = 1.3
information statement and written consent form to take home (SD = 1.5), with clinical significance bands 0–2 (not significant),
to their parents/guardians. A total of 1239 students and their 3 (slightly raised risk) and 4–10 (high risk).
parents/guardians were recruited to participate in the CATS
To tap children’s self-reported emotional well-being, four key
study (54%).
internalising symptom items were selected from validated and
Two parent surveys were used to assess family sociodemo- widely used psychometric measures by child anxiety and
graphic details and children’s mental health. The first short depression experts. The wider CATS project spans a broad
questionnaire was included with the information and consent range of health and developmental measures, and the child
form package. The second longer questionnaire was emailed or questionnaire was only 20 min of class time. Therefore, a small
posted to parents after the child’s data collection session. The number of emotional well-being items were necessary (Mundy
child-report measures in the present study utilised an Apple et al., 2013). Two key items were selected from the Spence Chil-
iPad application administered at schools during pre-allocated dren’s Anxiety Scale (SCAS: Spence, Barrett, & Turner, 2003),
class time. The children’s online questionnaire was completed namely ‘I worry about things’ and ‘I feel afraid’. Two key items
in a group setting under the supervision of trained research were selected from the Short Mood and Feelings Questionnaire
assistants. The research assistants read aloud items for clarity (SMFQ: Angold et al., 1995), namely ‘I felt miserable and
with the help of a standardised explanatory script and then unhappy’ and ‘I didn’t enjoy anything at all’. The internalising
responded to any queries of individual children to ensure all symptom items were scored on a 5-point Likert scale, ranging
were able to comprehend the items. Amongst items in the child from 0 (never) to 4 (almost always). Internal consistency was
online survey were questions about peer relationships as well as reasonable for the four items (Cronbach’s a = .70). Children’s
socio-emotional well-being. self-report internalising score was the average of these items.
Of the recruited CATS participants, 1221 (99%) families pro-
vided data at child age 8–9 years, forming the sample for this
Friendships. A variable was created to represent four possi-
study. The sample of children was broadly population represen-
ble combinations of children’s friendships: no friends, best
tative (Australian Bureau of Statistics, 2013). Children ranged
friend only, group of friends only or best friend + group of
in age from 8 years, 1.6 months to 9 years, 11.9 months
friends. Children’s friendships were measured via self-report.
(M = 8 years, 11.7 months, SD = 4.3 months). The recruited
The first item asked children ‘Do you have a best friend?’ with
sample contained a slightly smaller proportion of boys (45.9%)
response option yes or no. The second item asked, ‘Do you have
than girls (54.1%) compared with census data for 8- to 9-year-
a group of friends?’ with response option of not many, some or
old children enrolled in grade 3 across the state (51% males,
lots. Children who responded ‘not many’ were considered not to
49% females, p = .028). Children were in the main born in Aus-
have a group of friends. Children who responded ‘some’ or ‘lots’
tralia (85.3%, n = 1042), and also in the United Kingdom (2.0%,
were classified as having a group of friends.
n = 24), New Zealand (1.1%, n = 14) and ‘other’ countries
(8.1%, n = 99). This Melbourne sample scored slightly higher on
a measure of socioeconomic status compared with the entire Analyses. Frequencies were generated to estimate the
Australian population (M (SD) 1012.3 (66.3) vs. 1000 (100), prevalence of bullying (including subtypes) for the full sam-
p < .001); Australian Bureau of Statistics, 2013). Also, a higher ple and by sex. Independent samples t-tests then compared
percentage of the sample was indigenous compared with all the group of children who experienced frequent bullying ver-
grade 3 children in Victoria (4.7% vs. 1%, p < .001)). sus the remainder on mental health symptom scores. Chi-
square was applied to compare friendship categories for the
group of children experiencing frequent bullying versus the
Measures remainder. Analysis of variance (ANOVA) with the frequently
Frequent bullying. Bullying was measured via child self- bullied children compared their mental health symptoms
report with key items from the Gatehouse Bullying Scale (Bond, across different friendship categories, with post hoc Bonfer-
Wolfe, Tollit, Butler, & Patton, 2007). An item assessed physical roni tests exploring where the significant between-group dif-
bullying (Have you been hit, kicked or pushed by another stu- ferences lie. All analyses were conducted using Stata 13.1.
dent?). An item assessed verbal bullying (Has anyone teased or This study was approved by Human Research Ethics com-
called you names?). Both items specified a time frame for bully- mittees at the Royal Children’s Hospital Melbourne (31089)
ing experiences ‘in the past month’. Children responding yes and La Trobe University (FHEC13-NR25). Permission was
were then asked how often they had each experience (response granted from the Victorian Department of Education and
options less than once a week, about once a week, most days). Early Childhood Development office and the Catholic Educa-
In line with prior field research, children were classified as fre- tion Office Melbourne to recruit through their schools.
quently bullied if they reported facing physical or verbal bullying
about once a week or on most days.

Mental health. Children’s emotional well-being (internalising

Results
symptoms) was measured by parent report, as well as child self- Children’s mental health in the population-based sam-
report. Children’s behavioural functioning (externalising symp-
ple of 8- to 9-year-olds was comparable to SDQ Aus-
toms) was only measured by parent report. Parent report on
children’s mental health utilised the Strengths and Difficulties
tralian norms (Hawes & Dadds, 2004; Mellor, 2005). The
Questionnaire (SDQ) for ages 4–11 years (Australian version) sample’s score for the emotional symptoms subscale
(Goodman, 1997; Mellor, 2005). The SDQ has 25 items covering was M = 2.2 (SD = 2.1) with 14.1% (n = 168) scoring at
five subscales: emotional symptoms, conduct problems, peer ‘high risk’ of clinical significance. The sample’s score for

© 2018 Association for Child and Adolescent Mental Health.

 4 Jordana K. Bayer et al. Child Adolesc Ment Health 2018; *(*): **–**

the conduct problems subscale was M = 1.4 (SD = 1.5) Table 2. Frequent bullying and 8- to 9-year-old children’s mental
with 8.8% (n = 104) scoring at ‘high risk’ of clinical sig- health
nificance. On the child self-report measure, the sample’s
Mental health Bullying n M SD p-value
internalising symptoms score was M = 1.2 (SD = 0.8).
The prevalence of frequent bullying reported by 8- to Internalising difficulties
9-year-old children was examined, for the whole sam- Child self-report Frequent 356 1.6 0.9 <.001
ple and by sex (Table 1). A third of children reported Not frequent 808 1.1 0.8
the experiencing bullying on a weekly basis. The most com- Parent report Frequent 347 2.4 2.3 .026
prevalence mon form of frequent bullying was verbal, reported by Not frequent 790 2.1 2.0
of bullying one in four children. One in seven children reported
in school Externalising difficulties
children the experience of physical frequent bullying. Up to 10%
Parent report Frequent 348 1.6 1.6 <.001
of children reported a combination of verbal and physi- Not frequent 790 1.3 1.4
cal frequent bullying. The rate of verbal frequent bully-
ing was similar for boys and girls; however, boys Frequent bullying (at least once a week), not frequent (not bul-
reported slightly higher rates of physical and combined lied, or less than once a week).
verbal/physical bullying.
Independent samples t-tests compared children
reporting frequently bullying to the remainder of the Table 3. The relationship between experiencing bullying and
sample on the mental health symptom scales (Table 2). friendship
Children reporting frequent bullying had higher inter- Not
nalising and externalising symptoms on average. This Frequently frequently
pattern of difference was consistent across data sources. bullied bullied
Children reporting frequent bullying had significantly n = 811 n = 356
higher internalising difficulties as measured by self- and
parent report, and externalising difficulties as measured Friendship % n % n p-value
by parent report. Best friend 93.5 331 92.4 749 .488
Association between friendship and reporting bully- No best friend 6.5 23 7.6 62
ing was explored via chi-square (Table 3). Having a Group of friends 84.8 301 95.1 769 <.001
best friend was not related to reporting frequent bul- No group of friends 15.2 54 4.9 40
lying. Children who reported frequent bullying had a
best friend as often as those who did not report fre- All percentages are valid per cents.
quent bullying (over 90% of 8- to 9-year-olds have a
best friend). However, significantly fewer children who
reported frequent bullying said that they had a group Discussion
of friends. Amongst those reporting frequent bullying,
ANOVA models explored the mental health of children In early primary school almost one in three Australian
with different types of friendships (Table 4). For chil- children in the population sample reported experiencing
dren’s externalising difficulties, no protective effect of bullying on a weekly or daily basis. This frequent bully- statistics
friendships was apparent. However, children reporting ing commonly consisted of verbal (one in four children),
frequent bullying who had friends had less internalis- physical (one in seven children) or the combination of
ing difficulties (by self-report and parent report) com- both types (up to one in 10 children). Boys and girls both
pared with those without friends. Post hoc Bonferroni reported experiencing frequent bullying, which was
tests indicated that having a group of friends rather associated with higher mental health symptoms of an
than a best friend was potentially protective. On both emotional and behavioural nature. Children who
self- and parent report internalising symptoms, there reported frequent bullying were less likely to have a
were significant post hoc differences between the group of friends around them, and a group of friends
group with a best friend only and the group with both seemed potentially protective of these children’s mental
a best friend and a group of friends (p = .033 and health. Having a best friend, in contrast, did not appear
p = .007, respectively). potentially to protect the mental health of children
reporting bullying.
These findings with young Australian primary chil-
Table 1. Rates and types of frequent bullying experienced by 8- dren align with the large body of international findings in
to 9-year-old children (% (n)) adolescence. In both primary school and senior school,
many young people face frequent bullying, males and
Full sample Boys Girls Chi-square
females experience verbal bullying, and males are more
n = 1221 n = 661 n = 560 p-value
likely to experience physical bullying (Craig et al., 2009;
Bullied at least 29.2 (356) 32.1 (180) 26.6 (176) .017 Olweus, 1978; Rigby, 2000). As in adolescence, the expe-
once a week rience of frequent bullying in childhood relates to poorer
Verbal 22.7 (277) 24.1 (135) 21.5 (142) .214 mental health. Prior studies with youth (11 + years)
Physical 13.8 (169) 17.5 (98) 10.7 (71) <.001 have shown that frequent bullying is associated with
Physical + 7.4 (90) 9.5 (53) 5.6 (37) .007 mental health problems (Bond et al., 2001; Due et al.,
verbal
2005; Ford et al., 2017; Hodges et al., 1999; Klomek
Not frequently 66.4 (811) 62.3 (349) 69.9 (462) .017
et al., 2007; Lien et al., 2009; Thomas et al., 2017; Ttofi
bullied
et al., 2011; Undheim & Sund, 2010). The current
All percentages are valid per cents. results with younger children corroborate this concern.

© 2018 Association for Child and Adolescent Mental Health.