Introduction

During the last two decades, bullying at school has became the target of expanding research activity and action. A child 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 children, and there is an imbalance in strength. The child who is exposed to the negative actions has difficulty defending him/herself, and is rather helpless against those who do the harassing. Negative actions can be carried out not only physically, but also through words or making faces or gestures, or by exclusion from a group [27].

In surveys on children 8–12 years of age, the prevalence of bullying is found to vary between 3 and 23%, and the prevalence of victimization between 8 and 46% (see review in 47). The large variability of prevalence rates may partly be explained by differences in the instruments used, but most likely also results from real national and regional differences. Younger children [27] and boys [3, 9, 10, 16, 24, 37, 47, 48] are likely to bully more often than girls and older children, and boys are found to be victims of bullying more often than girls in many [3, 9, 10, 16, 24, 30, 47], but not all studies [37]. Being bullied has been shown to be associated with a range of psychiatric symptoms and social difficulties: anxiety [13, 33, 48], depressive symptoms [5–7, 13, 14, 37, 44, 48], suicidal thoughts [6, 13, 17, 44], physical health problems such as sleep problems, headache and stomach ache [7, 14, 45], low self-esteem [6, 48], and higher rates of absenteeism from school [9, 18]. In the prospective study of Fekkes and colleagues [8], it was found that victims of bullying were at significantly higher risk of developing new psychosomatic and psychosocial problems compared to other children, and also that children with depressive symptoms and anxiety had a significantly higher risk of being newly victimized.

Those who do the bullying have been found to have high rates of problems such as depressive symptoms [13, 14, 33, 37, 48], suicidal thoughts [13, 44], psychosomatic symptoms [9, 14], and conduct symptoms [12, 46]. Two types of bullies are identified, those who are socially unskilled and those who are socially skilled [15]. In the study of Juvonen and colleagues [12], it was found that despite increased conduct problems, bullies enjoyed high social standing among their classmates. In several studies, those who are both bullies and victims have been found to be the most troubled [9, 12–14, 17, 21].

Concerning the family backgrounds of bullies and victims, it has been found that bullying and victimization are associated with poorer family functioning [29], interparental violence [2], and parental maltreatment [38]. Olweus [27] has found no relation between aggression levels of boys and the socioeconomic conditions of their families. In the study on German and English children, a weak association between bullying and victimization and socioeconomic status of the family was found [47]. In the Nordic study children in families with low education and low socioeconomic status had increased risks of being bullied [26].

Victimization is a source of severe distress in many children and it can have long-lasting harmful effects even in later life. In the large prospective study (n = 1300) of Kumpulainen and colleagues [23], it was found that those children who were bully-victims at early elementary school age and those who were victims in their early teens, had more psychiatric symptoms at the age of 15 years than other children. Furthermore, bullying and victimization are found to be rather stable across developmental years [22, 41]. In a longitudinal study, it was found that, in a sample of over 800 children, almost all boys who were victims at age 16 had been victims already at the age of 8, while among the girls, half of those who were victims at age 16 had been victims already at the age of 8. There was also considerable consistency concerning bullying. Half of the boys and one fourth of the girls who bullied at age 16 had been bullying already at the age of 8 [41]. Several intervention studies have been done in Europe and North America to decrease bullying in school but the efficacy of such interventions has been somewhat contradictory, some showing a significant decrease in bullying and victimization, and some not [39].

Despite the growing interest in bullying, there are only few population-based time-trend studies on changes in prevalence of bullying and victimization. Schnohr and Niclasen [36], in their study based on a WHO collaborative study on health behaviour in school-aged children (HBSC), found that student- reported bullying and victimization increased among 11, 13 and 15-year-old children in Greenland from 1994 to 2002. According to the same HBSC study, victimization decreased among 11,13 and 15-year-old children in Lithuania, but the number of children reporting bullying others remained about the same from 1994 to 2002 [49]. In a Nordic study of children aged 2–17 years, parents reported a small increase in bullying from 1984 to 1996 [26].

The aims of this study were 1. to find out if there were changes in rates of bullying and victimization between the years 1989 and 1999 among eight-year-old children living in south-western Finland, and 2. to study the associations between victimization and bullying and psychiatric symptoms and the possible differences in the statistical strengths of these associations between the years 1989 and 1999.

Participants and methods

1989 sample

The target population was all Finnish-speaking children born in 1981 and living in one of the five University hospital catchment areas of Finland (Turku University Hospital), of whom 13% were chosen by selecting a representative sample of school districts. The school districts were located in seven different municipalities. Any child registered in the district and born in 1981 was included in the sample; the children belonging to the district, but attending a school outside the district, were also included (for example, because of a need for special education for disabled children, or the desire to attend schools with special language classes, such as English or German classes). The names and school locations of the children attending school outside their own district were given by the school authorities. The sample is a subsample of the Finnish Epidemiological Multicenter Child Psychiatric Study [1].

The sample consisted of 1,038 children. Three classes in the 1989 sample were excluded from the analysis, because in the 1999 sample (see below) three teachers did not want to participate in the study, so the children in the corresponding classes in the 1989 sample were left out, leaving 985 eligible children in the 1989 sample. In the case of the 985 eligible children 95% (936) of parents, teachers and children themselves filled in the questionnaires and were included in the analysis (90% of the total sample) [35]. Almost all children in this age group are in their second school year.

1999 sample

In 1999, a corresponding sample to that of the 1989 sample was chosen. All children born in 1991 in the same school districts as in the 1989 sample were included. Because three teachers were not willing to participate in the study 73 children were excluded. As in the 1989 sample, the children who lived in the selected school districts but attended school outside their own district were traced and invited to participate in the study.

In the case of the 962 children to whom questionnaires were distributed 86% (831) of parents (80% of the total sample), 85% (820) of children and 85% (818) of teachers returned the questionnaires [35]. Identification codes and information about gender in 180 children’s questionnaires (Children’s Depression Inventory) were lost due to a technical mistake. Thus, the data from 180 children’s questionnaires could not be linked to parents’ and teachers’ questionnaire data and were excluded from those analyses, where associations between children’s self-reported bullying and victimization and parent- and teacher-reported psychiatric symptoms were studied.

The genders were equally represented in both samples (48% boys and 52% girls), more mothers had completed upper secondary school in the 1999 than in the1989 sample (50% and 32%, respectively), and fewer children lived in families with two biological parents in 1999 than in the 1989 sample (77 and 85%, respectively) [42].

Procedures

Data collection was organized through teachers, and the procedure was similar at both time-points. The study was approved by the school authorities of all municipalities and by the ethics committee of Turku University Hospital. The teachers were contacted by phone or visited, and those teachers who were willing to participate were informed about the study protocol. The teacher sent a Rutter’s parent questionnaire via the child to the parents and the parents returned it in a sealed envelope to the teacher. In 1999 (but not in the 1989 sample), the parents’ written consent was required for their child to participate in the study, and only after this was given did the teacher fill in a Rutter’s Teacher Questionnaire about the child, and the child filled in a Child Depression Inventory questionnaire in the classroom. In 1989, it was assumed that parents who returned the Rutter’s parent questionnaire and did not forbid their child’s participation had accepted the study. The teachers sent the filled-in questionnaires (parents’ questionnaires in sealed envelopes) to the researchers. The study design and procedure have been described earlier in the study of Sourander and colleagues [42].

Methods

Questions about bullying and victimisation

The child was asked about bullying by giving him or her three alternatives to choose from: 1. “I bully other children almost every day”, 2. “I bully sometimes”, 3 “Usually I do not bully”. Victimization of bullying was asked about with the alternatives: “Other children 1. bully me almost every day”, 2. “bully me sometimes”, 3. “usually do not bully me”.

Parents and teachers were presented with the statement: The child bullies other children, and the alternatives given were: 1. doesn’t apply, 2. applies somewhat, and 3. certainly applies. The statement about a child being a victim of bullying was: The child is bullied, and the alternatives were: doesn’t, somewhat, certainly applies.

Questions about psychiatric symptoms

The child was asked to fill in The children’s depression inventory (CDI). The CDI is a 27-item self-report questionnaire on depressive symptoms in children during the previous two weeks developed by Kovacs [19, 40]. In both the 1989 and the 1999 survey, the question about suicidal ideation was omitted from the CDI because we did not consider it appropriate in a questionnaire filled in a classroom setting by relatively young children.

Parents filled in the Rutter parent questionnaire A2 (31 items), and teachers the Rutter teacher questionnaire B2 (26 items), both of which assess global child psychiatric symptoms divided into emotional, conduct and hyperactive symptoms within the past 12 months [31, 32].

The validity and reliability of the CDI, Rutter parent and Rutter teacher questionnaires are well documented both internationally and in Finland, and have been found to be good or moderate [19, 20, 40]. The responses from any participant were excluded from the analysis if, respectively, 20/26, 24/31 or 20/26 questions were not answered.

Statistical analysis

Bullying and victimization were the outcome variables in the analyses and were measured on an ordinal scale with three categories. The analysis of associations between different variables (CDI, Rutter parent scale including emotional, conduct and hyperactivity subscales, and Rutter teacher scale with emotional, conduct and hyperactivity subscales) and outcome variables (bullying and victimization) was carried out using cumulative logistic regression analysis. Cumulative logistic regression analysis is a comprehensive methodology suitable for univariate and multivariate analysis when the outcome variable is measured on an ordinal scale [11]. The associations were quantified with their cumulative odds ratios (COR) and 95% confidence intervals (95%CI). In all tests, P-values less than 0.05 were considered statistically significant. Statistical computations were performed using SAS system for windows, release 8.2/2000.

Results

Differences in prevalence of victimisation and bullying between 1989 and 1999

In both child and parental reports, significantly fewer children in 1999 than in 1989 had been victims of bullying. The difference between the 1989 and 1999 samples was similar among boys and girls (non-significant interaction between gender and year). On the other hand, according to the teachers’ reports, fewer boys were victims in 1999 than in 1989, while among girls, no significant change was found. Teacher- reported victimisation of girls was very low at both time-points. (Table 1).

Table 1 Prevalence of bullying and victimization reported by children, parents and teachers in the 1989 and 1999 samples in boys and girls and total samples
Full size table

Parents reported statistically significantly fewer children to be bullies in 1999 than in 1989.

Moreover children themselves and teachers reported fewer children to be bullies but the differences were statistically non-significant (Table 1).

Association of child-reported victimization and bullying with psychiatric symptoms

Victimization

In univariate analysis, girls’ self-reported victimization was significantly associated with self-reported depressive symptoms and with all parent-reported psychiatric symptom entities, emotional, conduct and hyperactivity symptoms, and with teacher-reported hyperactivity symptoms (Table 2).

Table 2 Univariate associations between child-reported bullying and victimization by psychiatric symptom scales in girls and boys (Child Depression Inventory, Rutter’s parent and Rutter’s teacher questionnaire and subscales of emotional, conduct and hyperactive symptoms).
Full size table

In univariate analysis, boys’ self-reported victimization was most strongly associated with self- reported depressive symptoms. In both parent and teacher reports, self-reported victimization associated significantly with conduct and hyperactivity symptoms, and in teacher reports, also with emotional symptoms (Table 2).

There were correlations between reported psychiatric symptoms. In multivariate analysis, the overlapping of these correlations was studied for parent- and teacher- reported psychiatric symptoms separately. In girls, self-reported victimization was found to be significantly associated with parent-reported conduct symptoms and teacher-reported hyperactivity symptoms (Table 3). In boys, self-reported victimization was found to be significantly associated with both parent- and teacher-reported conduct symptoms, and also with teacher-reported emotional symptoms (Table 3).

Table 3 Associations, adjusted by sample, between child-reported bullying and victimization and psychiatric symptoms in girls and boys, results of multivariate cumulative logistic regressiona analysis of parent-reported emotional, conduct and hyperactivity symptoms and, correspondingly, Teachers’ Questionnaire subscales
Full size table

Bullying

In univariate analysis, girls’ self-reported bullying was most strongly associated with self-reported depressive symptoms, the cumulative odds ratio being 4.3. Girls’ self-reported bullying was also significantly associated with both parent- and teacher-reported conduct and hyperactivity symptoms (Table 2).

Boys’ self-reported bullying was also most strongly associated with self-reported depressive symptoms, the cumulative odds ratio being 2.7. Among boys, in the univariate analysis, self-reported bullying was significantly associated with both parent- and teacher-reported conduct and hyperactivity symptoms (Table 2).

In multivariate analysis performed separately for parents’ and teachers’ reports, girls’ self-reported bullying was significantly associated with both parent- and teacher-reported conduct symptoms. In boys, self-reported bullying was significantly associated with both parent- and teacher-reported conduct symptoms and with parent-reported hyperactivity symptoms (Table 3).

Differences in associations of self-reported victimization and bullying and psychiatric symptoms between girls and boys

There was only one significant difference (interaction) in the associations of victimization and bullying and psychiatric symptoms between girls and boys. Girls’ self-reported bullying was more strongly associated with teacher-reported conduct symptoms than boys’ (P-value for the interaction 0.0315).

Differences in associations between victimization and bullying and psychiatric symptoms between the 1989 and 1999 samples

There were only few significant differences in the associations of victimization and bullying and psychiatric symptoms between the 1989 and 1999 samples. In the total sample (boys and girls together), the association between self- reported bullying and teacher-reported conduct symptoms became slightly stronger (P-value for interaction 0.0317) from 1989 (P < 0.0001, COR 1.5, 95%CI 1.3–1.7) to 1999 (P < 0.0001, COR 1.8, 95%CI 1.5–2.3). The self-reported victimization and teacher-reported emotional problems also became stronger (P-value for interaction 0.0357) from 1989 (P = 0.070, COR 1.1, 95%CI 1.0–1.3) to 1999 (P = 0.0002, COR 1.4, 95%CI 1.2–1.7). On the contrary, the association between self-reported victimization and self-reported depressive symptoms became inversely associated from 1989 (P < 0.0001, COR 2.8, 95%CI 2.2–3.6) to 1999 (P = 0.028, COR 0.5, 95% CI 0.3–0.9) (P-value for interaction <0.0001).

Discussion

The present study showed that fewer 8-year-old children were victims of bullying in 1999 than in 1989. There was a decrease in the number of bullies but the change reached statistical significance only in the parental reports. The prevalence of and gender difference in victimization and bullying were within the same range as in earlier studies. Boys were more often bullies and victims than girls.

The samples were representative but the difference in response rates, even if high in both samples (95% in 1989 and 85% in 1999) may have slightly biased the findings. Unfortunately, we have no possibilities to analyse the characteristics of the non-respondents. A limitation of the study is that in the questionnaires no detailed definition of bullying was given and different types of bullying were not specified (physical aggression, physical bullying, verbal aggression, and social exclusion). Research on bullying has greatly developed during the last two decades and currently more sophisticated questions would have been appropriate. In the recent study of Naylor and colleagues [25] it was found that there are important differences between teachers’ and pupils’ definitions of bullying, with teachers expressing more comprehensive ideas in their definitions, including not only direct bullying but, for example, also social exclusion and a power imbalance in the bully’s favour. The main aim of the study was to compare the situation between 1989 and 1999, and therefore the questions formulated in the 1980’s were used in an identical manner.

In this study only one age group was included and the results can not be generalized to other age groups. The strength of the study was that three different informants, the child her/himself, a parent and a teacher, were used. The results according to all informants were similar. Another limitation was that peer reports were not used, but this was impossible because children going to schools outside their own school district were also traced and included.

The analyses were performed only with bully and victim categories; the bully-victim group was not analysed separately in this study because the sample size was too small, especially for girls. As the main aim was to compare the different time trend aspects (by three different informants and frequency categories) different frequency categories or answers of different informants were not combined so as not to miss detailed information.

During the years 1989–1999, no specific large-scale bullying intervention was undertaken in the schools of the study area. The law which made the principal in each school responsible for having an action plan against bullying came only later (2003). One possibility for the decreasing trend in bullying may be adults’ increasing awareness of bullying and its harmful effects. Schools have used their own models in the fight against bullying, some schools being more active than others (oral communication from principal school counsellor, P Danström, 10.10.03). The local university has a strong tradition of scientific research on children’s aggression and bullying, and this has included a collection of samples of bullying in schools in the study area in the 1990s [34], which may have aroused awareness of bullying and associated problems amongst professionals working in schools in this area. In the recent Nordic study of Nordhagen and colleagues [26], children in families with low education had increased risks of being bullied. This may partly explain the lower rate of victimization in the 1999 sample in the present study, where mothers’ education was found to have improved compared to the 1989 sample.

Our finding of a decreased proportion of bullied children corresponds to the finding in Lithuania, where in 1994, 41.7% of boys and 39.5% of girls were bullied, and in 2002 36.4 and 32.3%, respectively [49]. On the contrary, in Greenland, among 11, 13 and 15-year-old children, increasing rates of bullying and victimization were found from 1994 to 2002 [36]. According to the HBSC survey, both Lithuania and Greenland have higher rates of bullying and victimization than Finland [49]. The study conducted among Nordic children aged 2–17 years showed a small increase in victims of bullying from 1984 to 1996 (from 13.7 to 16.4), which was only significant in Denmark and Norway (not in Finland and Sweden). In that study, only parental reports were used, the questions were slightly changed, and the results are also difficult to compare because the study included children of various ages [26].

It may be that especially young primary school pupils are susceptible to influences to reduce bullying behaviour. In the anti-bullying intervention study of Stevens and colleagues [43], victimization was reduced among primary school but not among secondary school pupils. Like in our study, decreased victimization but no corresponding decrease in bullying was found in their study. It may be that when the awareness of bullying arises those who bully become more aware of their problematic behaviour and report being bullies. Another explanation is that the same number of bullies are harassing fewer victims.

Self-reported bullying was associated with self-reported depressive symptoms but not with teacher- or parent-reported emotional problems. It seems that parents and teachers fail to recognize emotional problems among bullying children. In earlier studies, it has been found that children and parents often disagree about issues that are abstract and ambiguous, such as internalizing symptoms and psychosomatic symptoms [4, 28, 35]. Victimization and emotional problems were associated with parent reports in girls and teacher reports in boys, and the results also remained significantly associated in the multivariate analysis. It seems that adults recognize emotional problems better in children who are victims than in children who are bullies.

The statistical strengths of associations of victimization and bullying with psychiatric symptoms were on the whole the same in 1989 as in 1999. Self-reported bullying was more strongly associated with teacher-reported conduct symptoms in 1999 than in 1989. Teachers also saw victims as more emotionally troubled in 1999 than in 1989. Teachers saw bullies and victims as more troubled in 1999 than in 1989, which may indicate that the children involved in bullying had more problems in 1999, but the results can also mean that more attention is paid to children who bully at school. As there is now more awareness of the negative psychological effects of being bullied and the problems of bullies are better understood, it might be that teachers paid more global attention to these children and their well-being in 1999.

The findings of this study are interesting from the children’s mental health promotion perspective. Several studies have shown that bullying is connected to concurrent psychiatric symptoms in children and adolescents, but recent prospective studies have shown that bullying experiences are also connected to future psychiatric symptoms [5, 8, 23]. Universal prevention of bullying is one way to improve children’s mental health. In schools, not only teachers, school counsellors and psychologists, but also school health services have the opportunity to carry out anti-bullying work.

In intervention studies examining the effectiveness of anti-bullying programmes in schools, control samples are necessary, because as shown in this study, changes in prevalences can occur even without specific large-scale interventions. As our understanding grows about different aspects of bullying, it is necessary that, in future epidemiological studies, large enough samples are included to be able to take into consideration different phenomena among children of different ages and genders, and to be able to study bullies, victims and bully-victims separately.

In the future adults should continue their efforts to diminish bullying in general. In individual cases adults should discuss the problem with a bullied or bullying child, especially with a bully-victim, and with the child’s family to consider the child’s well-being and need for extra support.

Conclusions

The associations between psychiatric symptoms and bullying and victimization seem to be rather stable. Slightly decreased levels of victimization among 8-year-old Finnish children is a promising result, but further time-trend studies are needed, as well as qualitative studies to obtain a deeper understanding of the bullying phenomenon and the factors affecting it. It might be that especially young children are susceptible to influences reducing victimisation and bullying.