Classroom influences on bullying

Educational Research

ISSN: 0013-1881 (Print) 1469-5847 (Online) Journal homepage: http://www.tandfonline.com/loi/rere20

Classroom influences on bullying Erling Roland & David Galloway To cite this article: Erling Roland & David Galloway (2002) Classroom influences on bullying, Educational Research, 44:3, 299-312, DOI: 10.1080/0013188022000031597 To link to this article: http://dx.doi.org/10.1080/0013188022000031597

Published online: 02 Dec 2010.

Submit your article to this journal

Article views: 1445

View related articles

Citing articles: 61 View citing articles

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=rere20 Download by: [Durham University Library]

Date: 03 November 2015, At: 03:04

Educational Research Vol. 44 No. 3 Winter 2002 299–312

Classroom in uences on bullying Erling Roland and David Galloway, Centre for Behavioural Research, Stavanger University College, Box 2557, Ullandhaug, 4091 Stavanger, Norway

Downloaded by [Durham University Library] at 03:04 03 November 2015

Summary The relationship between estimates of teachers’ management of the class, the social structure of the class and bullying others was investigated by questionnaires answered by 2,002 pupils and 99 teachers in Norwegian primary schools. A path analysis demonstrated that the social structure of the class had a direct impact on bullying behaviour. Classroom management had a direct impact on the prevalence of bullying other children, and an indirect impact on this variable via the social structure. The joint impact of management and social structure on bullying others was substantial. Class-level estimates of family conditions of the pupils were included in the analyses. Implications for preventing bullying by general management of the class are discussed. Keywords: bullying, management, structure, family

Introduction This study investigates the relations between teachers’ management of the class, the social structure of the class and pupils bullying other pupils. Bullying is commonly de ned as long-standing, negative behaviour conducted by an individual or a group, and directed against a person who is not able to defend her/himself in the actual situation (Olweus, 1997, 1999; Rigby, 1997; Roland, 1993, 1998; Roland and Munthe, 1997). For nearly 30 years, bullying among pupils has been recognized as a serious problem in schools. Although prevalence rates vary, a conclusion from research in the Scandinavian countries and Finland, several other countries in Europe, North America, Australia and Japan is that at least 5 per cent of the pupils in primary and secondary schools are bullied weekly or more often. Also, approximately 5 per cent of the pupils bully others that frequently. The number of victims among girls is almost as high as among boys, but considerably more boys than girls bully others. Besides those who are bullied persistently and those who Address for correspondence: Erling Roland, Centre for Behavioural Research, Stavanger University College, Box 8002, N-4068 Stavanger, Norway. Tel: 47-51 83 29 00. Fax: 47-51 83 29 50. E-mail: [email protected] Educational Research ISSN 0013-1881/print/ISSN 1469-5847 online © 2002 NFER http://www.tandf.co.uk/journals DOI: 10.1080/001318802200003159 7


300

Educational Research Volume 44 Number 3 Winter 2002

bully others on a regular basis, many more pupils are involved in bullying on an occasional basis (O’Moore, Kirkham and Smith, 1997; Olweus, 1997, 1999; Pepler et al., 1993; Rigby, 1997; Roland, 1998; Roland and Munthe, 1997; Smith, 1997). Much research has concentrated on identifying the amount of bullying, the personalities of the victims and the bullies and their family conditions. A dominant concept in the dynamics of bullying, based on this research, is that adverse home conditions can create a stable, aggressive trait ‘within’ some pupils, the bullies (Smith and Myron-Wilson, 1998). Such pupils often bully certain kinds of peers who, for various reasons, are convenient victims (Olweus, 1978, 1999; Roland, 1998). Recently, some aspects of the interactions in bullying, in particular, the role of bystanders, have also been addressed (Cowie, 2000; Craig and Pepler, 1995). In general, however, home conditions and relatively stable aspects of personality in the pupils directly involved in bullying have been the dominant variables reported in the literature. The empirical evidence for this conclusion is substantial (Olweus, 1980, 1999; Roland and Idsøe, 2001). However, both schools and classes have been reported to differ in the amount of bullying. These differences have been reported independent of urbanization in the catchment area of the school, and also independent of the size of the school and the class (Olweus, 1993; Roland, 1998). Differences at school and classroom level could be attributed to parallel differences in home conditions of the pupils. One could, however, also suspect that social aspects of schools and classes in uence the amount of bullying. Little or no research has been conducted to answer this important question about possible in uences on bullying from school-level to classroom-level variables (Roland, 1998). The present study concerns the possible relationships between the teacher’s management of the class, the social structure of the class and bullying.

Management and social structure of the class Teacher management of the class is comprised of many interrelated aspects of teacher–pupil interactions, which are also related to both pupil behaviour and to learning outcomes (Doyle and Carter, 1987; Kounin, 1970; Mortimore et al., 1989; Teddlie and String eld, 1993). Competence in teaching, monitoring, intervention and personal caring for students are commonly reported as general and important dimensions of teacher behaviour (Jones, 1986; Smith, 1990), especially within the tradition of school and teacher effectiveness research (Teddlie and String eld, 1993). These four aspects of management were selected for the present study. As formal groups, classes in school will almost by de nition possess ‘nomothetic’ features (Getzels and Thelen, 1960), which broadly correspond to what Hargreaves (1995) calls the instrumental-social control domain. In primary school classes we could roughly understand nomothetic aspects of behaviour as centring on task activities, and we will therefore be interested in the quality of such activities and the smoothness of change between them. Another dimension of group life is the ideographic aspect (Getzels and Thelen, 1960), or the expressive-social cohesion domain (Hargreaves, 1995). This is the informal pattern of interactions between group members, also between members of formal groups. Finally, a third dimension is the social norms that operate between the pupils (Morland, Argote and Krishnan, 1996; Schmuck and Schmuck, 1977), which will

Bullying and classroom management

301

comprise both nomothetic and ideographic issues. The three aspects of the social structure of the class selected for this study concern informal relations between pupils, their concentration on different learning activities and their social norms. These three variables are regarded as interrelated (Cartwright and Zander, 1969; Getzels and Thelen, 1960; Roland, 1998). The hypotheses of the present study are therefore:


Teachers’ management of the class is related to the interactions in bullying. The social structure of the class is related to the interactions in bullying. Teachers’ management of the class is related to the social structure of the class. An important background variable is the home conditions of the pupils, as family variables have been reported as related to bullying (Olweus, 1999).

Methodology The present investigation was based on three pilot studies in which instruments and procedures were tested on different groups of teachers and pupils (Roland, 1998).

Instruments The key information required concerned classroom management by the teachers, formal and informal interactions within their classes, family conditions of the pupils and bullying. In addition, some information relevant for evaluating the validity of the key variables of the study was obtained. The information was collected by two parallel questionnaires, one for the pupils and one for the designated main teacher of each class. (In Norway each primary class has about four teachers, one of whom is the designated class teacher.)

Management Parallel data from teachers and from pupils on management of the class were available, but the correlation was low (r = 0.19). A reason for this low correlation is probably that the data given by the pupils re ect their perception of the overall management of their class, while the appointed class teacher reported on his or her style of managing the class. We have therefore used the pupil data. Pupils were asked about aspects of their teachers’ management of the class. Four subscales were constructed, each focusing on a broad aspect of management that had been identi ed as important in research on school effectiveness (e.g. Rutter et al., 1979; Mortimore et al., 1989) and/or in research on school in uences on pupil behaviour (e.g. Galloway, 1985). Caring for pupils (Caring) The items concerned whether the teachers of the class cared about the pupil, whether the teachers were interested in pupils’ out of school activities, if the teachers tried to help the pupil if he or she had problems, and about friendship between the teachers and pupil. Competence in teaching (Teaching) This variable comprised items about teachers’ competence in explaining subject-matter and pedagogy, plenary activities, group activities and individual work, as well as alternation between these activities.

302


Monitoring The pupils were asked about their homework, work during lessons, behaviour in classroom and behaviour during breaks, and whether the class teachers closely monitored this.


Intervention The items for this variable were problem orientated, as they concerned how the teachers reacted when something was not acceptable – e.g. different aspects of social interaction between pupils and between teacher and pupils. ‘Management’ was the sum of Caring, Teaching, Monitoring and Intervention. Management is treated as a rst-level variable, while Caring, Teaching, Monitoring and Intervention are treated as second-level variables. There are theoretical reasons for computing this rst-level variable based on the second-level ones, as the respective aspects of management are regarded as interrelated.

Social structure of the class Data on the social structure of the class too were available from both the designated class teachers and the pupils. Here, though, the correlation was moderately high (r = 0.49), implying that the pupils and their designated class teacher had similar perceptions of the class social structure. We have used the teacher data-set because the high correlation between pupils’ perception of management and the social structure of the class (r = 0.70) could have been caused, in part, by a response set. In the teachers’ questionnaire, there were three scales concerning the social structure of the class; the three aspects of social structure are outlined above. Informal relations between pupils (Relations) This aspect of the social structure of the class concerned the cohesion domain. An examination of the items shows that many aspects of informal relations are included: friendship, support, attraction, isolation, power and relations between subgroups. Concentration on schoolwork (Concentration) It was preferred to use the concept ‘concentration’ related to learning activities rather than time on-task. A practical reason for not using time on-task as the estimate of the instrumental-control domain is that we were not in a position to observe or otherwise collect good quantitative data about pupils’ on-task activities. The items concerned how well the pupils concentrated when they performed different kinds of on-task activities, and how smoothly they changed from one activity to another when directed by the teacher. Informal norms (Norms) The items of this scale concerned both ideographic and nomothetic aspects of the class, in other words items concerning norms about both informal relations between the pupils (e.g. ‘Most pupils agree that it is right to help other pupils who are being bullied’) and about schoolwork (e.g. ‘Most pupils agree that school work is important’). In addition, items related to norms on teacher authority were included. The main variable ‘Structure’ is the sum of scores on Relations, Concentration and Norms. Structure is treated as a rst-level variable, while Relations, Concentration and Norms are treated as second-level variables. The principle followed is the same as with Management, since Relations, Concentration and Norms are regarded as interrelated.


303

Family conditions and bullying The pupils’ questionnaire comprised one scale concerning family conditions and two scales about bullying – i.e. being bullied and bullying others.


Six out of eight items from an instrument constructed by Ystgaard (1993) were adopted to obtain information on family conditions. The scale concerns different aspects of the relationship between the pupil and the family. Two items from the original scale were excluded because of consistency problems (Borg and Gall, 1989). The key questions presented to the pupils about bullying at school were, in principle, the same as those commonly used in previous major investigations, and the pupils were provided with a de nition or explanation of bullying (Olweus, 1991, 1993; Smith and Sharp, 1994).

Items and scoring system The items concerning Management, Structure and Family were statements. Alternatives for answering were ‘YES’, ‘yes’, ‘no’, ‘NO’, implying ‘De nitely’, ‘Probably’, ‘Probably not’ and ‘De nitely not’ (Management and Structure) and ‘Agree completely’, ‘Agree’, ‘Disagree’ and ‘Totally disagree’ (Family); these were scored 4, 3, 2 and 1 respectively. Items concerning bullying were formulated as questions about frequency of being bullied and bullying others in the current school year. Alternatives for answering were ‘Never’, ‘Sometimes’, ‘About every week’ and ‘About every day’, scored 0, 1, 2 and 3 respectively. Missing data on the two one-item variables bullying others and being bullied were not replaced. Missing data at item level on multi-item scales were replaced by the gure closest to the mean. For pupils, the mean number of missing data on an item was 17, and the maximum number was 31. For teachers, the mean number of missing data was 4, and the maximum was 9.

The sample The investigation was conducted among teachers and pupils at primary schools in one county in Norway. The county chosen is located in the southern part of Norway and has about 150,000 inhabitants. The population is slightly younger than in the rest of Norway. Income per capita is slightly below the mean for Norway, and there is no large difference between rich and poor. The pupil sample consisted of pupils in grades 4–6 (age 10–13), representing the three nal years of primary school in Norway. (Following the implementation of a recent school reform, these pupils would now be considered as being in grades 5–7.) The teacher sample comprised the appointed main teachers of the classes that were involved in the study. Only so-called ordinary primary schools took part. Small primary schools where pupils of two or more age levels composed a class were excluded. Excluded also were so-called combined schools; these are schools containing pupils at both primary and secondary level. In the county capital, by far the largest municipality, about half of the participating ordinary primary schools were chosen randomly. All the other schools in the county, except one, were asked to take part in the investigation. The one school that was not asked to participate was excluded because our research institute had recently run a project at this school about prevention and management of bullying and other behavioural problems. This left a total of 31 ordinary primary schools, which were asked to participate. The schools


304


in the sample had a total of 180 classes and approximately 3,440 pupils at grades 4–6. Of the 31 ordinary primary schools that were asked to participate, 22 took part in the investigation. There were different reasons why nine schools did not take part. Two of them were occupied with another investigation, and three schools in one municipality did not get the material in time because of a misunderstanding at the local post of ce. The remaining four schools did not want to participate. The 22 schools had a total of 128 classes at grades 4–6. For different reasons, a total of ten classes did not take part. The remaining 118 classes comprised a total of 2,389 pupils. Almost 400 of these were lost because of various reasons (Roland, 1998), and we ended up with 2,002 valid cases. The total response rate, in those classes that participated in the investigation, is 83.8 per cent. A total of 99 out of 118 appointed main teachers (84 per cent) agreed to take part in the study.

Procedure The investigation was conducted in May, when the schools were about to nish the school year. The main teachers of the classes were asked to answer the questionnaire anonymously. The pupils received a letter describing the investigation for their parents, and the parents were requested to reply whether their child could take part. The appointed main teacher of the class collected the answers. This teacher also administered the pupil questionnaire in their classes according to written instructions (Roland, 1998). Each school was asked to conduct the investigation at the same time for all the participating classes, and all schools in the sample were asked, if possible, to conduct it on one particular day suggested by us. The teachers were requested to provide the information without discussing it with anyone.

Reliability and validity Management and Structure are sum scores, each based on several second-level variables. Therefore the reliability and validity of these second-level variables are also discussed below, although relationships between the second-level variables and bullying are not reported. The consistency of the scales was evaluated by Cronbach’s alpha (Cronbach, 1951). Together with mean score and standard deviation, this is given in Table 1. The terms ‘ rst-level’ and ‘second-level’ variables appear in the table: the consistency of a rst-level variable is the estimate of interrelation between the sum scores on the scales estimating the respective second-level variables. There are two such rst-level variables: Management and Structure. Alpha values for the rst-level variables Management and Structure are 0.88 and 0.67 respectively. All alphas of the second-level variables, except one, are above 0.70, and none is below 0.60. In general, the consistency of the variables seems to be good or acceptable. Validity was evaluated by inspecting how variables were related to each other and to other theoretically relevant variables. This procedure, often called construct validation (Borg and Gall, 1989; Brown, 1970), revealed no particular problems of validity (Roland, 1998, pp. 130–49). The validity of the variables seems to be good, and the consistency good or fairly good, when the evaluation is based on individual responses from teachers


305


TA B LE 1 Cronbach’s alpha for rst-level and second-level variables: mean scores and standard deviations

Management Caring Competence Monitoring Intervention Structure Relations Effectiveness Norms Family

Alpha

Mean

SD

0.88 0.70 0.80 0.73 0.63 0.67 0.77 0.81 0.71 0.74

3.23 3.00 3.41 3.22 3.28 3.14 2.83 3.02 3.58 3.44

0.48 0.64 0.57 0.57 0.63 0.36 0.50 0.53 0.37 0.42

TA B LE 2 Class-level estimates of Bullying Others (BO) and Being Bullied (BB): means, standard deviation, and minimum and maximum scores (N = 118)

BO BB

Mean

SD

Min.

Max.

0.38 0.54

0.18 0.25

0.00 0.00

0.88 1.12

and pupils. Since most analyses are based on class-level aggregated data, we were con dent about the reliability and validity of the variables.

Results Class-level scores on Bullying Others and Being Bullied The distribution of class-level scores on Bullying Others and Being Bullied is shown in Table 2. There are substantial differences between classes on both Bullying Others and Being Bullied, as the standard deviations and minimum and maximum scores demonstrate. The difference between classes is greatest on Being Bullied, according to the SDs. Also, the difference between maximum and minimum score is greater for Being Bullied than Bullying Others. In fact the minimum score for both Being Bullied and Bullying Others is zero. At the class level, the correlation between Being Bullied and Bullying Others is 0.40 (p < 0.001). This rather weak correlation may be explained, in part, by the pattern of bullying within and across classes.

Bullying within and across classes Most of the victims in the sample reported that they were bullied by pupils from their own class, but there were obviously a lot of victims who were bullied by pupils from other classes. Of those pupils who reported being bullied ‘sometimes’ and more often, 40.3 per cent reported that they were bullied by pupils from their own class only. Also, 19.7 per cent of the victims said that they were bullied by pupils from other classes only, and 29.1 per cent reported being bullied by pupils from both their own class and from other classes. The rest of the victims, 10.9 per cent, did not answer this question.

306 TA B L E 3

Educational Research Volume 44 Number 3 Winter 2002 Intercorrelations between Family, Management and Structure

Management Structure

Family

Bullying Others

0.40** 0.21**

–0.28** –0.41**


Key: ** p < .01

It can be seen that more than half of those victims who answered the question were bullied exclusively or also by pupils from other classes. It does not seem reasonable to argue that social relationships and the educational climate at class level should in uence the amount of bullying that class members experience from pupils in other classes. On the other hand, class-level experiences could reasonably be expected to in uence the amount of bullying for which class members were responsible. We therefore concluded that class-level estimates of being bullied are not a logical variable to relate to Management and Structure. Consequently, only Bullying Others will be considered in the analyses. A signi cant negative relationship between Management and Bullying Others is evident from Table 3. The relationship between Structure and Bullying Others is also negative, and is signi cant and stronger than that between Management and Bullying. These bivariate correlations demonstrate that both the management and social structure of the class are related to class-level estimates of bullying others. Two important issues deserve to be addressed, however: the possible in uence of family issues and the causal relationship between Family, Management, Structure and Bullying Others.

The family and bullying High scores on Family report positive family relationships. Data are aggregated to the classroom level. The correlation between Family and Bullying Others at the classroom level was –0.22 (