Intended for healthcare professionals


Bullying in schools: self reported anxiety, depression, and self esteem in secondary school children

BMJ 1998; 317 doi: (Published 03 October 1998) Cite this as: BMJ 1998;317:924
  1. G Salmon, senior registrar in child and adolescent psychiatrya,
  2. A James, consultant in child and adolescent psychiatry (Tony.James{at},
  3. D M Smith, senior medical statisticianb
  1. aHighfield Adolescent Unit, Warneford Hospital, Oxford OX3 7JX
  2. bCentre for Statistics in Medicine, Institute of Health Sciences, Oxford OX3 7LF
  1. Correspondence to: Dr James
  • Accepted 13 May 1998

Evidence exists of considerable problems with bullying and bullied children in secondary schools. In the largest survey in the United Kingdom to date 10% of pupils reported that they had been bullied “sometimes or more often” during that term, with 4% reporting being bullied “at least once a week.”1 The impact of the introduction of policies on bullying throughout a school seems to be limited.1 The commonest type of bullying is general name calling, followed by being hit, threatened, or having rumours spread about one.1 Bullying is thought to be more prevalent among boys and the youngest pupils in a school.2

Summary statistics and details of fitted models. Values are numbers of schoolchildren unless stated otherwise

View this table:

We are unaware of any study that has examined the mental health problems of children who are being bullied. We assessed self reported anxiety, depression, and self esteem in bullied children and those who were not bullied and in bullies and those who were not bullies.

Subjects, methods, and results

Four questionnaires (the Olweus bully/victim,2 the short mood and feelings,3 the revised children's manifest anxiety incorporating a lie scale,4 and the Rosenberg self esteem5 questionnaires) were anonymously completed by 904 pupils aged 12-17 in years 8-11 in two coeducational secondary schools. School A is a non-selective school in a socially disadvantaged urban area. School B is a rural grant maintained school in an area with a higher than average proportion of high social class households.

Logistic regression models were fitted to the proportions of bullied or bullying children using STATA. Categorical variables were school, school year, and sex. Anxiety, lying, esteem, and depression scores were treated as continuous variables. The table shows the odds ratios of the fitted logistic regression models. For anxiety, esteem, lying, and depression the odds ratios are for a change of one standard deviation of 6.22, 4.92, 2.13, and 5.44 respectively (pooled SD). The prevalence of being bullied “sometimes or more often” was 4.2%. Significant variables (P<0.05) for being bullied were school, sex, and anxiety and lying scores; school year approached significance (P=0.06). The prevalence of bullying others “sometimes or more often” was 3.4%. Significant variables (P<0.05) for being a bully were school year, sex, and anxiety, lying, and depression scores. The esteem score featured in neither model.

Boys in year 8 in school A with high anxiety and lying scores were most likely to be bullied. Girls in year 9 in school B with low anxiety and lying scores were least likely to be bullied. Boys in year 10 with low anxiety and lying scores and high depression scores were most likely to be bullies. Girls in year 8 with high anxiety and lying scores and low depression scores were least likely to be bullies.


Bullied children tend to be in the lower school years. The low prevalence of bullying (4.2%) may reflect the effectiveness of bullying interventions already in place in the two schools. Our data support the idea that bullied children are more anxious and bullies equally or less anxious than their peers.2 New findings from the study are the relation between having a high lying score and being bullied and having a high depression score and being a bully. The male to female ratio of bullies (3:1) is lower than that previously reported (4:1).2 This may indicate that bullying interventions are having more of an impact on the direct bullying characteristic of boys and less on the indirect bullying more common among girls.

Our results should be viewed with caution because our study is small, but they suggest factors that could be important.


Contributors: GS and AJ initiated the study and designed the protocol. GS collected and collated the data. DS performed the analyses. GS, AJ, and DS interpreted the results, discussed their meaning, and wrote the article. AJ is the guarantor for the study.


  • Funding Oxfordshire Health Services Research Fund

  • Conflict of interest None


View Abstract