- Roberto Forero, research and evaluation coordinatora,
- Lyndall McLellan, project officerb,
- Chris Rissel, epidemiologist ()b,
- Adrian Bauman, professor of public healthc
- a Health Promotion Unit, Liverpool Hospital, Liverpool, New South Wales 2070, Australia
- b Needs Assessment and Health Outcomes Unit, Central Sydney Area Health Service, Newtown, New South Wales 2042, Australia
- c School of Community Medicine, University of New South Wales, Kensington, New South Wales 2052, Australia
- Correspondence to: C Rissel
- Accepted 4 May 1999
Objectives: To examine the prevalence of bullying behaviours in schoolchildren and the association of bullying with psychological and psychosomatic health.
Design: Cross sectional survey.
Setting: Government and non-government schools in New South Wales, Australia.
Participants: 3918 schoolchildren attending year 6 (mean age 11.88 years), year 8 (13.96), and year 10 (15.97) classes from 115 schools.
Main outcome measures: Self reported bullying behaviours and psychological and psychosomatic symptoms.
Results: Almost a quarter of students (23.7%) bullied other students, 12.7% were bullied, 21.5% were both bullied and bullied others on one or more occasions in the last term of school, and 42.4% were neither bullied nor bullied others. More boys than girls reported bullying others and being victims of bullying. Bullying behaviour was associated with increased psychosomatic symptoms Bullies tended to be unhappy with school; students who were bullied tended to like school and to feel alone. Students who both bullied and were bullied had the greatest number of psychological and psychosomatic symptoms.
Conclusions: Being bullied seems to be widespread in schools in New South Wales and is associated with increased psychosomatic symptoms and poor mental health. Health practitioners evaluating students with common psychological and psychosomatic symptoms should consider bullying and the student's school environment as potential causes.
Bullying behaviour occurs in schools worldwide and is likely to be associated with poor health in schoolchildren
Research into bullying has been mainly focused on victims but there are other categories of bullying that deserve attention
At least three out of five students experienced or participated in bullying in schools in New South Wales, Australia
The psychosocial and psychosomatic health of the students varied according to their bullying status
Editorial by Chesson and p 348
Recent reports have highlighted the frequency of bullying in schools and the adverse consequences on bullying behaviour in adolescence.1–7 Despite the efforts of schools to prevent or stop bullying, it still occurs worldwide 2 8–10 Victims of frequent bullying have been reported to experience a range of psychological, psychosomatic, and behavioural symptoms including anxiety and insecurity,11 low self esteem and low self worth,2 3 12 considerable mental health problems, sleeping difficulties, bed wetting, feelings of sadness, and frequent headaches and abdominal pain 1 5 They are also more likely to be unhappy and depressed5 and absent from school.13
Although definitions of bullying behaviour vary, bullying has been defined as the “intentional, unprovoked abuse of power by one or more children to inflict pain or cause distress to another child on repeated occasions.”14 The most common form of bullying self reported by Australian students is verbal harassment—for example, teasing and name calling.13 This is consistent with students in Norway15 and England.1 Most students do nothing to help victims because they feel it is not their place to get involved.16 17
Reported estimates of bullying vary owing to differences in the type of measurements taken and the sex, age, and ethnic origin of students studied. For example, in Newham, east London, 22% of young people had been bullied at some time.1 In Yorkshire, 21% of children had been bullied and about 17% had taken part in bullying others “sometimes or more often.”18 Another study in the United Kingdom found these rates to be 4.2% and 3.4% respectively.11 The weekly incidence of bullying in Australian schools has been estimated at about 1 in 6 children during any one year.16 Bullying has been found to decrease with age, with boys more likely to have been bullied and to participate in bullying others compared with girls 5 11 13 18 Bullies are more likely to dislike school and to engage in behaviours that compromise their health such as smoking and drinking alcohol to excess.5 According to the problem behaviour theory, bullying others may be one of a cluster of problem behaviours.19
Our study sought to identify the prevalence of bullying behaviours among schoolchildren in New South Wales, Australia, and the association of bullying with psychological and psychosomatic health. Our study differs from earlier ones by differentiating students involved in bullying behaviour from those who bully and are themselves bullied, those who only bully, those who are only bullied, and those who are neither bullies nor bullied.
Participants and methods
Analyses are based on a statewide sample of year 6 (primary) and year 8 and year 10 (secondary) students attending one of 115 Catholic, government, and independent schools. We stratified the schools into primary and secondary schools, and we selected those to participate through cluster random sampling defined by school class.
Participation was voluntary and anonymous. We obtained ethical approval for our study from the relevant education sectors and the New South Wales Health Department.
We randomly selected one class from each selected school year according to the World Health Organisation's survey protocol for health behaviour among schoolchildren.5 The self administered surveys were completed in classrooms under exam-like conditions in October and November 1996, at the end of the last term (term 3) of the Australian school calendar.
The survey was administered statewide by health promotion staff of the New South Wales Health Department. The questionnaire was adapted20 from the WHO's health behaviour survey in schoolchildren, which has been shown to be valid and reliable in Europe for over a decade.5 The core questionnaire was validated for Australian conditions in 1992 and 1996.20 Questions from the instrument that are the focus of our report relate to demographics, bullying, psychosomatic symptoms, mental health, and school and social contact.
Categorisation of bullying behaviours
Bullying behaviours were the dependent variables. We used the following definition: “Bullying is when another student, or group of students, says or does nasty and unpleasant things to him or her. It is also bullying when a student is teased repeatedly in a way he or she doesn't like. But it is not bullying when two students about the same strength quarrel or fight.”20 Students were asked to respond to two questions: have you ever been bullied in school this term? and, how often have you taken part in bullying other students in school this term? We dichotomised responses into students who had not experienced bullying behaviour or who had (once or more). We then further classified the students into four groups as bullies, bullied, both bullies and bullied, and neither bullies nor bullied.
Students were presented with a series of health symptoms (headache, stomach ache, backache, feeling low, irritable or bad temper, feeling nervous, difficulties getting to sleep, feeling dizzy) and asked to report the frequency with which they experienced each. These items were summed to form a unit weighted psychosomatic symptom scale, and frequency of symptoms was categorised as: low scores (0-4; up to three symptoms less than once a month or never); moderate (5-8; four or more symptoms about every month or up to three per week); frequent (9-14; four or more symptoms once per week or up to three per week); and high (15-32; four or more symptoms more than once a week or about every day) (mean 8.5 (SD 6.3)). The Cronbach's alpha of the scale was 0.81.
Smoking —Current smoking was identified by asking the students how often they smoked at present. We dichotomised responses as less than weekly and once or more per week.
Mental health—We examined four single items assessing dimensions of mental health. Happiness was assessed by asking the students how generally they felt about life at present (responses dichotomised as happy or unhappy). Loneliness was measured by asking the students if they ever felt lonely (responses dichotomised as lonely or not lonely). Students were then asked how often it happened (during the last school term) that other students did not want to spend time with them and they ended up being alone (responses dichotomised as alone or not alone). Confidence was measured by asking the students whether they felt confident in themselves (responses dichotomised as confident or not confident).
Social contact—We assessed social contact by asking the students how often they spent time with friends straight after school, and how many evenings per week they usually spent out with their friends. We dichotomised responses as frequent (four or more days and three or more evenings) or infrequent (three or less days and four or more evenings).
Reactions to school—To assess the students' reactions to school we asked three pertinent questions: how they felt about school at present (dichotomised into like or dislike); whether school was a nice place to be (dichotomised into agree or disagree); and absenteeism (whole days) during the past four weeks of school without parental permission (“wagging”) before the survey was administered (dichotomised as frequent or infrequent).
We assessed the prevalence of the studied items by sex and school year with descriptive univariate statistics produced by SPSS release 6 software. The selected independent variables were dichotomised as healthy and non-healthy, and we examined their associations with bullying behaviours.
We used STATA software (release 5) to test the associations between the independent and dependent variables, adjusting for the design effect of clustering within schools. For this analysis we have assumed that there were four different bullying outcomes. In this case bullying outcomes were unordered events.
Multinomial logistic regression (STATA; College Station, TX) is an extension to the usual model for binary data21 and is used to model an outcome that is measured on a nominal scale (no natural order). We used svymlog21 (STATA) to estimate the adjusted relative risk ratio of age, sex, psychosomatic symptoms, and sociodemographic factors on children from each of the bullying categories of bully, bullied, and “both” versus the comparison group “neither.”
Complete questionnaires were received from 3918 students—2129 girls (54.3%), 1764 boys (45.0%), and 25 (0.64%) not specified—representing a participation rate of 84% on the basis of enrolment figures for secondary school years 8 and 10 and 88% for primary school year 6. The mean age of our sample was 11.88 years (SD 0.52) for year 6 students, 13.96 (0.54) years for year 8 students, and 15.97 (0.51) years for year 10 students. Boys and girls in each year were of identical age. Student distribution across the school years was similar: 1222 students in year 6 (31.1%), 1403 in year 8 (35.8%), and 1268 in year 10 (32.4%). Overall, 443 students (11.3%) were born overseas, 3381 (86.3%) were born in Australia, and 94 (2.4%) were from an indigenous background.
Overall, 1650 students (42.4%) reported neither being bullied nor bullying others in the last term of the school calendar. Of the remaining 2268 students (57.8%), 928 (23.7%) bullied others, 843 (21.5%) both bullied and were bullied, and 497 (12.7%) were bullied. More boys (526, 29.8%) than girls (402, 18.9%) reported bullying or both being bullied and bullying (460 boys (26.1%), 358 girls (16.8%)), but slightly more girls (286, 13.4%) than boys (211, 12.0%) reported being bullied.
Prevalence of bullying behaviours
Table 1 presents the unadjusted prevalence of bullying behaviours and the studied variables by school year and sex. Table 2 presents the results from the multinomial logistic regression of students who bullied, who were bullied, and who were both bullied and bullied others. We used the group of students who reported neither having been bullied nor bullying other students as the comparison group.
After adjusting for survey sampling and the other social and psychological factors in our model, significantly more boys than girls were bullies (relative risk ratio 2.0); bullies experienced frequent (1.6) and high scores for psychosomatic symptoms (1.8), were current smokers (1.5), spent four days or more with friends after school (1.2), and spent three or more evenings out with friends (1.4). Bullies were significantly more likely to be in year 6 than years 8 or 10 (0.65 and 0.48 respectively), did not like school (0.48), and did not think that school was a nice place to be (0.55) (table 2).
Those who were bullied were more likely to be boys than girls (1.3), to be lonely (1.7), and to find themselves alone because other students did not want to spend time with them at school (3.4). Bullied students were less likely to be in year 8 (0.71) or year 10 (0.32), spent three or more evenings out with friends (0.52), did not like school (0.57), were current smokers (0.38), and missed days from school without permission from their parents (0.48) (table 2).
Bullied students who also bullied
Students who were both bullied and bullied others were significantly more likely to be boys (2.1), experienced frequent (1.6) and high scores for psychosomatic symptoms (2.0), reported being alone (2.5), were current smokers (1.8), and spent three or more evenings out with friends (1.4). These students were significantly less likely to be in year 8 (0.65) or year 10 (0.28), to spend time with friends after school (0.68), to feel school was a nice place to be (0.50), and to like school (0.57) (table 2).
Our results indicate that, in one term of the school year, more than three out of five students experienced or participated in bullying in schools in New South Wales. The estimate of 13% of students as victims of bullying is less than that reported in English schools, although this number increased substantially to 34% when we included students who both bullied and were bullied. Exposure to bullying may be substantial for those who are bullied as there were more students who bullied (but not also bullied) than students who were bullied.
Our results show a statistically significant association between bullying behaviour and psychosomatic symptoms and smoking, with those students who both bullied and were bullied reporting the highest frequency of symptoms. Bullies tended to be unhappy with school, and students who were bullied tended to like school more and to report feeling alone. Unsurprisingly, students who both bullied and were bullied exhibited the characteristics of disliking school and feeling alone, and they seemed to have the most psychological and psychosomatic symptoms.
Although our results are cross sectional, they are consistent with other findings that have identified bullied children as having few friends, being more introverted than others, and generally lacking social skills.2 13 Bullies also deserve attention. Dawkins says that bullying may be one component of a more general pattern of antisocial and rule breaking behaviour that shows considerable stability over time,14 although our data do not address this point.
As suggested by others,1 14 health practitioners evaluating students with common psychological and psychosomatic symptoms should consider bullying and the student's school environment as a potential cause. A positive school environment may increase health promoting behaviours 1 22 and it is likely that bullying and its consequences can be reduced if the school does not tolerate bullying. Engaging the assistance of children not involved in bullying may help to reduce tolerance of bullying and change the normal attitudes around bullying. Studies of policies and strategies that effectively reduce bullying are needed.
We thank all the schools, staff, and students who participated, the health promotion staff throughout New South Wales who assisted with data collection, and Ms Philayrath Phonsavan, who coordinated the data management.
Contributors: AB and LM initiated the study, designed the protocol and materials, collected and collated the data, and carried out reference study No 20. RF developed the model and conducted the multinomial logistic regression analysis. LM conducted univariate data analysis. RF, LM, CR, and AB interpreted the results, discussed their meaning, and contributed to the writing and editing of this manuscript. CR and RF contributed to reference study No 20. AB will act as guarantor for the paper.
Funding Epidemiology Unit, South Western Sydney Area Health Service. The health promotion branch of the New South Wales Health Department provided a small grant for the printing of the surveys.
Competing interests None declared.