Differences in pupil characteristics and motives in being a victim, perpetrator and witness of violence in secondary education

Abstract

Socially problematic and violent behaviour of pupils in and around schools is undesirable from pedagogical, social and societal perspectives. The motives underlying violence between different social actors in school may help explain and improve this behaviour. The aim is to investigate the relationship patterns between characteristics of secondary school pupils and their problem social behaviour including the motives they attribute to being a victim, perpetrator or witness of six types of violence, in relation to the complementary social roles of other pupils, teachers, other school staff and pupils’ relatives, respectively. Data was collected with the aid of a Dutch nationwide Internet‐based survey in secondary schools. A total of 80,770 pupils from 215 school locations completed the questionnaire. Data was checked for reliability, scale homogeneity and representativeness. Pearson correlations show that pupil characteristics indicating educational attainment level (low), feeling at home in the Netherlands (not feeling at home), gender (male) and age (being older) are most important in problem social behaviour and violence motive patterns. Being religious is less relevant; degree of urbanisation is least relevant. The motives attributed most often refer to physical appearance, behaviour, level of school achievement, handicap, being religious, gender, sexual preference and ways of dealing with non‐conforming behaviour or punishments. Social interactions between pupils and between pupils and teachers are generally most important for eliciting violence‐related motives, followed by interactions between pupils and pupils’ relatives. The results are comparable with those from printed questionnaires in nationwide research, which supports the validity of the Internet‐based survey. It is concluded that the substantive results provide a more complete, thorough and systematic picture of social discrimination and motive aspects than has hitherto been customary. The resulting indicators can be used in Internet‐based feedback procedure cycles to inform school policy about evidence‐based results comparing the school’s own pupils, teachers, educational support staff and management to national benchmarks so as to support efforts to improve social safety in and around school.

Keywords:

• school safety
• social discrimination
• motives for violent behaviour
• complementary social roles
• pupils
• teachers
• Internet‐based survey

1. Introduction

There is significant evidence that individual genetic and environmental conditions influence the development and concrete manifestations of a person’s social behaviour (Loeber et al. 2008; Moffitt 1993). From a very young age, a person’s personal, cultural, educational and demographic characteristics relate in different ways to either prosocial behaviour or socially problematic or violent behaviour. Relevant personal characteristics are, for example, age and gender (Pellegrini, Bartini, and Brooks 1999; van Lier and Koot 2008). Generally, adolescents demonstrate more antisocial behaviour than persons at other ages, while boys generally behave more violently than girls (see also Arbeitsgruppe Schulforschung 1980; Fekkes 2005; Mooij 1994; Olweus 1991, 1993). One cultural characteristic that plays a role in schools with pupils from various ethnic minority or immigrant backgrounds is whether or not pupils feel at home. School studies by Carbines, Wyatt, and Robb (2006) and Gillison, Standage, and Skevington (2008) suggest that this feeling reflects social integration, safety and cohesion in a specific institution, region or country. Their research also reveals that not feeling at home may be related to different types of antisocial or violent behaviour. They furthermore illustrate the relevance of being religious. On the one hand, religious pupils may behave more socially than non‐religious pupils and help or support other persons; on the other, being religious appears to be related to more dogmatic and antisocial behaviour. One educational variable that plays a role here is the pupil’s educational attainment level (Beirn, Kinsey, and McGinn 1972; National Education Association 1994). Pupils with low attainment may display more antisocial or aggressive behaviour (Loeber and Farrington 2001). Another relevant factor is the degree of urbanisation of the region in which a pupil lives. Studies by the American Psychological Association (1993), Beirn, Kinsey, and McGinn (1972) and Mooij (2001) demonstrate that, compared to living in rural areas, pupils who live in a city experience more violent behaviour.

Empirical research usually concentrates on one or a few of the potentially relevant pupil characteristics and on one or a few types of violence. This precludes estimating the relevance of pupil characteristics across different types of violent behaviour. Moreover, the pupil’s role in the violent incident is crucial. A victim’s experience of violence will be very different from a perpetrator’s or a witness’ experience. At the same time, various actors may be included in a violent relationship as complementary social actors. With respect to school, complementary social actors may be one or more other pupils, teachers, other school staff or pupils’ relatives. What may be particularly important whenever violent behaviour is initiated or elicited are the specific characteristics of other persons or situations used to motivate, justify or explain social rejection, including violent behaviour (cf. Allport 1948; Magnusson and Allen 1983; Polman 2008). In schools, for example, some pupils may interact regularly because they speak the same language or have the same skin colour, social behaviour or country of origin, while excluding pupils who do not comply with these specific characteristics, ignoring them or approaching them aggressively or violently. This process of ‘social discrimination’ between pupils may also occur between pupils and teachers, pupils and other staff or between pupils and the relatives of pupils (Carbines, Wyatt, and Robb 2006; Gillison, Standage, and Skevington 2008). In this respect, research has clarified that the social or behavioural dimension should be regarded separately from the pupil’s cognitive or achievement behaviour (Giannopulu et al. 2008; Sutherland and Oswald 2005).

In fairly stable institutions such as schools, long‐term social discrimination and corresponding behavioural processes between pupils and between pupils and school staff may act as key indicators of the degree of social cohesion within the institution. Social cohesion is defined as the degree of connectivity between individuals or groups of individuals in a specific environment (Beauvais and Jenson 2002; Farrington 1993; Peschar 2005). Specific feelings, emotions, ideas, activities, perceptions and practices reflect this connectivity and can also be used to increase or reduce the degree of social cohesion (Bayh 1975; Dijkstra et al. 2004; International Association for the Evaluation of Educational Achievement 2007; Lewin, Lippitt, and White 1939; ten Dam and Volman 2003, 2007). Low social cohesion reflects exclusion or segregation and antisocial or unsafe behaviour, such as bullying and violence in and around school (American Psychological Association 1993; Smith and Sharp 1994). For pupils, continuous negative social discrimination and the experience of antisocial behaviour may lead to their leaving school early; for staff it may mean their wanting to find work in other schools, or to work outside the school system altogether (Beirn, Kinsey, and McGinn 1972; Galand, Lecocq, and Philippot 2007; Tapola and Niemivirta 2008). These manifestations and consequences of antisocial behaviour are undesirable from pedagogical, social and societal perspectives (Collier 1994; Glover et al. 2000; Schuyt 1995; van der Ploeg and Mooij 1998). Increasing social cohesion may assist to overcome antisocial behaviour and segregation based on negative social discrimination between pupils and between pupils and school staff. Qualitative examples are presented in a report entitled ‘Encouraging tolerance and social cohesion through school education’, written for the Australian Department of Education, Science and Training (Carbines, Wyatt, and Robb 2006).

Systematic research into the different roles played by pupils in violent behaviour, the complementary social partners and the relevance of different motives in these relationships appears to be very scarce, however. Knowing more about such roles and their embedded motives could provide clues for reducing or preventing violent behaviour (cf. Loeber et al. 2008; Mooij 2005). One of the reasons for this lack of information appears to be that, in traditional survey or monitoring research that makes use of printed questionnaires, comprehensive clarification of the issues requires the involvement of large numbers of persons in rather complicated research. Nowadays, however, the Internet offers new opportunities for methodological data‐collection procedures (Blumenfeld et al. 2000; Mooij 2006). Using the Internet permits the flexible and large‐scale use of differentiated methods for measuring individual pupil characteristics, geared towards the various motives that may be involved in specific interaction types of violent behaviour and the pupil’s various social roles in relation to different complementary social actors.

Research providing this kind of information has been carried out for the Dutch Ministry of Education, Culture and Science. The Ministry wished to monitor variations in feelings of social safety, problem social behaviour and violent behaviour in and around secondary schools once every two years, starting in 2006. The researchers involved used the Internet to obtain representative data from pupils, teachers, other staff and school management in order to acquire a comprehensive overview of secondary school safety, including the patterning of social discrimination with respect to six types of violent behaviour. Types of violent behaviour included in the monitor study concerned verbal, material, social, mild physical, severe physical and sexual violence (Mooij, Sijbers, and Sperber 2006). Additional quantitative analyses were performed to reveal 12 motive patterns in relation to victim, offender and witness roles, and the complementary social roles of pupils, teachers, other staff and pupils’ relatives (Mooij 2007). Based on the latter study, the goal of the present article is to present some of the key results of the national monitor study on secondary school safety. The research question is: What are the relationship patterns between, on the one hand, the personal, cultural, educational and demographic characteristics of secondary school pupils and, on the other, their socially problematic behaviour and the motives involved in being a victim, perpetrator or witness of six types of violence, in relation to the complementary social roles of other pupils, teachers, other school staff and pupils’ relatives, respectively?

2 Method

2.1 Procedure

In 2005, the Dutch Ministry of Education, Culture and Science initiated the development of a monitor study to produce empirically controlled information about social safety in secondary education, including secondary education for pupils with special educational needs. A supervisory committee was set up during the preparation and execution of the monitor study. Represented on the committee were the Inspectorate of Education, an organisation representing school boards in secondary education (VO‐Raad), and a trade union for teaching staff (ABOP). In October 2005, the total number of Dutch school locations for secondary education and secondary special education was 1642. These school locations and their boards received a letter from the Ministry and a letter from the research institute explaining the goal of the monitor study and the overall procedure with respect to data collection and feedback of results. Each location was asked to nominate a ‘location monitor manager’ to facilitate data collection. It was explained that communication between the school location and the research institute would take place via the Internet. The location monitor manager was asked to organise data collection within the school location. He/she was able to create log‐in codes for classes of pupils, the teachers, other staff and the school management via the website www.veiligvo.nl and a confidential log‐in procedure in November and December 2005.

Research instrumentation was worked out in three separate questionnaires for the school management, staff and pupils. In December 2005, digital pilot versions were tested at some secondary school locations for all levels of attainment. A total of 10 classes of pupils, their teachers, two members of the administration, two hall‐porters, a member of the school management and the supervisory committee were involved in the pilot investigation. The recommendation to class teachers was to complete the questionnaire for themselves, either at school or at home, and then coach their class of pupils during the completion process. Completion by all respondents was anonymous. The pilot results led to minor adjustments regarding the number and nature of variables included, the wording used, the layout and the distribution of variables in the monitor study.

2.2 Participation

Within the population of 1642 locations for secondary education, 225 location monitor managers signed up to participate. Given that the monitor study was new in the field of education, and locations already had access to some other method of registering safety, this was a relatively large number. Each location monitor manager supervised the data‐collection process for his/her own location on the Internet. Pupils, school staff and the school management began to fill in the Internet‐based questionnaires at the end of the first week of January 2006, immediately after the Christmas holidays. If necessary, a location monitor manager could create new log‐in codes for pupils, staff or management, or remove codes that were no longer necessary. Such organisational activities remained possible throughout the entire completion period (January and February 2006). Additional information was provided by the research institute’s helpdesk.

A total of 215 (96%) of the 225 sites that had signed up actually participated in the data collection. Of these 215 sites, 191 sites (88.8%) collaborated with all three categories of social informants (pupils, staff and school management). The school safety monitor study was completed by 80,770 pupils, 5148 teachers, 1749 educational support staff and 629 members of the school management. Pupil participation was representative for the level of educational attainment, while school location participation was representative for the degree of urbanisation (Mooij 2006).

2.3 Variables

The first set of variables refers to six personal, cultural, educational and demographic pupil characteristics. Personal characteristics are age and gender. Each pupil was also asked whether he/she felt most at home in the Netherlands or in another country. This dichotomous cultural variable was used to indicate the pupil’s feeling of connection to major Dutch values and norms. A second cultural variable referred to whether the pupil was religious. Answer categories were: no; baptised but not attending church; and church attending (In 2008, these answer categories were changed into: no; religious but not attending church, mosque, synagogue or temple; and attending church, mosque, synagogue or temple.). Educational attainment level was coded in terms of pupil’s specific type of educational programme, ranging from low (special secondary education = 1) to high (university preparatory = 7). Degree of urbanisation was based on a geographical categorisation of schools using an approach developed by Vliegen (2005). His geographical system consists of four categories, ranging from ‘big city’ = 1 to ‘rural areas’ = 4.

A second set of pupil variables concerns feelings of safety with respect to various specific places in and around the pupil’s school. The answer categories ranged from never = 1 to always = 7. Pupils were asked to report on a period of approximately six months, in this case, from the 2005 summer holidays until the questionnaire date in early 2006. In addition, the degree of problem social behaviour, such as truancy and possession of or dealing in drugs and weapons, was measured using a procedure comparable to the one concerning feelings of safety.

Violent behaviour was divided into the six types, with descriptions of each one, referred to as ‘concepts’. The types and concepts are given in Table 1. The pupils indicated whether or not they had experienced each of the types of violence described in Table 1 between the 2005 summer holidays and the questionnaire date in early 2006.

Table 1. Types of violent behaviour and specifications.

Types of violence	Specifications
Verbal	Calling names, bothering someone on purpose, talking in an extra loud voice, making a lot of noise on purpose
Material	Scratching or damaging something, spray‐painting or dirtying something, hiding or mislaying something, destroying things, stealing
Social	Ignoring, excluding, threatening, intimidating, blackmailing, spreading false rumours
Mild physical	Striking or hurting someone on purpose, pushing or kicking someone on purpose, tripping someone on purpose, punching someone on purpose, hitting
Severe physical	Fighting with someone, beating or roughing someone up, threatening someone with a weapon, using a weapon
Sexual	Making sexual comments, sexual gestures, feeling someone up, sexually molesting someone, rape

The six types of violence were divided into three blocks with successive descriptions, and the pupil was asked whether he/she had experienced this violence at least once. If such was the case, the pupil was randomly asked to complete two of the six blocks. In each block, the pupil specified the frequency of the specific type of violent behaviour, the identity of the complementary social actor (other pupils, teacher, school staff or pupils’ relatives), the place where the incident or incidents had happened, the assumed underlying motives (maximum of 16), whether the incident(s) had been reported and, if so, to whom, and how effective reporting the incident had been. The 16 motives that pupils could select to explain why they were victims of violence were: because of my physical appearance, skin colour, country of origin, behaviour, excellent school achievement and marks, poor school achievement and marks, handicap, because I am religious, because I am not religious, because I am male, because I am female, because I am gay or lesbian, because I am bi‐sexual, because I wanted other persons to follow to the rules, because I corrected the behaviour of other persons, because other persons disagreed with a punishment given at school. Comparable motives explaining why a pupil was the perpetrator of violence were worded so as to describe the appearance, characteristics or behaviour of other persons.

If a pupil had experienced a specific type of violent behaviour in the six‐month period concerned, the score results per specific concept were dichotomised (no = 0, once or more = 1). The scores per type of violent behaviour were then included in principal factor analysis and Alpha scale analysis. For each type of violence, the factor results indicated the existence of a homogeneous group of items. The relevance of the factors was supported by the Alpha scale results, summarised in Table 2. Table 2 presents six reliable scales on types of violent behaviour.

Table 2. Types of violent behaviour and scale results.

Types of violence	N	N items	M	SD	Alpha
Verbal	26,727	4	0.65	0.39	.85
Material	26,471	5	0.26	0.32	.79
Social	26,188	6	0.30	0.31	.80
Mild physical	26,172	5	0.45	0.40	.86
Severe physical	26,249	4	0.19	0.25	.66
Sexual	25,763	5	0.15	0.23	.73

2.4 Analysis

Data analyses were carried out at the individual level by calculating means and standard deviations. Moreover, Pearson correlations were calculated at the individual level between the personal, cultural, educational and demographic pupil characteristics on the one hand, and the variables referring to feeling safe in and around school, playing truant, using drugs or weapons and motives behind a pupil’s being a victim, offender or witness of types of violent behaviour involving other pupils, teachers, other staff or pupils’ relatives, on the other hand. In particular, the analysis specifications concerning the 16 motives relevant to each type of violence, for each combination of being a victim, perpetrator or witness with the four complementary roles, result in tables with 6 × 16 = 96 correlations each. To save space, the relevant tables in the results section will present the numbers of significant motives per type of violence. Other significant outcomes regarding the specified motives will be discussed briefly. Detailed results can be found in Mooij (2007). With respect to the correlation coefficients in the present study, it should be noted that large numbers of units of analysis tend to reduce the magnitude of such coefficients (Pearson and Hartley 1972).

3 Results

3.1 Univariate results

The first set of pupil characteristics concerns personal, cultural, educational and demographic variables. Age varies between 9 and 24 years (M = 14.22; SD=1.52; N = 80,786), while boys make up 51.3% of the sample (N = 80,768). 89.3% of the pupils said they feel most at home in the Netherlands (N = 80,709). (The percentage of pupils born in the Netherlands was somewhat higher, 94.5%.). The percentage of pupils who are not religious is 40.1%; baptised but not attending church 19.7%, and church attending 40.3% (N = 80,425). A total of 80,317 pupils answered questions about their educational attainment level (special secondary education = 1 to university preparatory = 7). The distribution of percentages across attainment levels was representative for Dutch secondary education (Mooij 2006). Degree of urbanisation of the school was disaggregated to pupil level (N = 80,790).

Table 3 contains the univariate results of the variables related to feelings of safety in and around school and the pupils’ exposure to problem social behaviour such as truancy and the possession of or dealing in drugs and weapons.

Table 3. Univariate results of feeling safe, playing truant, drugs and weapons.

	Analysis N ^a	Item mean^b	SD
Feeling safe
Classroom	79,929	6.55	1.17
Study or work areas	79,927	6.51	1.22
Hallways	79,929	6.37	1.29
Cafeteria, dining hall	79,927	6.44	1.27
Toilets	79,929	6.43	1.36
Vestibule, lockers	79,927	6.44	1.29
Schoolyard	79,928	6.32	1.34
School neighbourhood	79,924	6.27	1.36
At home	79,927	6.70	1.08
Playing truant
Have other pupils in your class skipped school?	55,773	2.73	1.63
Have you skipped school?	77,701	1.43	.93
Has the school taken any steps against truancy?	45,386	3.48	2.57
Drugs
Do pupils in your class have drugs on them?	51,761	1.69	1.46
Have you ever taken drugs to school?	77,455	1.11	.68
Are drugs sold in or around the school?	37,987	2.03	1.77
Do pupils at your school take drugs?	40,516	3.00	2.10
Has the school taken any steps against drugs?	33,594	2.57	2.35
Weapons
Do pupils in your class have weapons on them?	48,911	1.42	1.24
Have you ever taken a weapon to school?	75,046	1.16	.85
Are weapons sold in or around the school?	40,698	1.21	.97
Do pupils at your school use weapons?	39,720	1.32	1.08
Has the school taken any steps against weapons?	31,835	2.32	2.33
Notes: ^aExcludes the category: I don’t know.^bItem score: (1) = never to (7) = always.

Generally, Table 3 illustrates that Dutch secondary pupils feel safe in and around school, although there are differences between the various places in school. The pupils feel least safe in the school playground and in the neighbourhood around their school. They feel most safe at home. They perceive other pupils as playing truant relatively more often than pupils themselves report truancy. Comparable phenomena occur with respect to taking drugs or weapons into school. According to the pupils, schools take the most measures against truancy, followed by measures against drugs and then weapons.

Table 2 has already presented the univariate results with respect to six reliable scales on violent behaviour. It has been shown that verbal violence occurs most frequently, followed by mild physical, social, material, severe physical and sexual violence, respectively.

3.2 Pupil characteristics and feeling safe, truancy, drugs and weapons

The first set of Pearson correlations concerns the personal, cultural, educational and demographic characteristics of pupils on the one hand, and their feelings of safety in and around school and the items about problem social behaviour such as truancy, drugs and weapons, on the other hand. The bivariate results are shown in Table 4. With respect to feeling safe, educational attainment level produced the highest significant correlations: the higher the level of attainment, the more pupils feel safe at school, in the school playground, in the school neighbourhood and at home. The next highest correlations are related to feeling at home in the Netherlands: pupils who feel themselves at home also feel safer in and around school. Other trends are that girls feel safer than boys, and pupils in rural areas feel safer than pupils in big cities.

Table 4. Pearson correlations between pupil characteristics and feeling safe, playing truant, drugs and weapons.

	Age 9–24	Boy–girl	I feel at home in the Netherlands (yes–no)	I am religious (no–baptised–yes)	Level of educational attainment (low–high)	Urbanisation (big city–rural)
Feeling safe
Classroom	−.01**	.03**	−.08**	−.00	.14**	.03**
Study or work areas	.00	.03**	−.08**	−.00	.15**	.03**
Hallways	.03**	.02**	−.08**	.01*	.14**	.03**
Cafeteria, dining hall	.01*	.03**	−.08**	.00	.13**	.04**
Toilets	.01**	.04**	−.08**	.00	.13**	.04**
Vestibule, lockers	.01**	.02**	−.08**	.00	.13**	.03**
Schoolyard	.02**	.01**	−.08**	.01	.13**	.02**
School neighbourhood	.02**	.00	−.08**	−.00	.12**	.04**
At home	−.04**	.04**	−.07**	−.00	.11**	.02**
Play truant
Have other pupils in your class skipped school?	.30**	−.02**	.09**	−.04**	.07**	−.09**
Have you skipped school?	.25**	−.05**	.09**	−.02**	.04**	−.05**
Has the school taken any steps against truancy?	.06**	.06**	−.02**	.07**	.05**	.02**
Drugs
Do pupils in your class have drugs on them?	.24**	−.08**	.09**	−.07**	−.01	−.03**
Have you ever taken drugs to school?	.12**	−.07**	.09**	−.01**	−.05**	−.02**
Are drugs sold in or around the school?	.15**	−.05**	.08**	−.05**	.06**	.06**
Do pupils at your school take drugs?	.16**	.03**	.06**	−.08**	.09**	.03**
Has the school taken any steps against drugs?	−.08**	.03**	−.02**	.03**	−.05**	.05**
Weapons
Do pupils in your class have weapons on them?	.11**	−.14**	.12**	.01*	−.07**	−.02**
Have you ever taken a weapon to school?	.09**	−.12**	.09**	.02**	−.06**	−.01**
Are weapons sold in or around the school?	.10**	−.10**	.12**	.03**	−.06**	−.01*
Do pupils at your school use weapons?	.08**	−.07**	.14**	.03**	−.07**	−.02**
Has the school taken any steps against weapons?	−.08**	.01	.00	.03**	−.07**	−.01
Notes: *Correlation is significant at the 0.05 level (two‐tailed).**Correlation is significant at the 0.01 level (two‐tailed).

Truancy and the possession, sale and use of drugs and weapons is particularly typical of older pupils compared to younger pupils and of pupils who do not feel at home in the Netherlands compared to those who do. Younger pupils are more likely than older pupils to note school measures against drugs and weapons. Being religious is related to lower scores on truancy and drugs, but also to higher scores on weapons. Being religious is, moreover, related to attending a school that takes action against truancy, drugs and weapons. Furthermore, boys possess, sell and use weapons more than girls do. Girls take more notice than boys of measures taken by the school against truancy and drugs. Truancy is more characteristic of high‐attainment pupils than of low‐attainment ones, whereas taking drugs into school and school measures against drugs are typical for low‐attainment pupils. Selling drugs in and around school and taking drugs by pupils at school, is more typical of pupils at higher educational levels than those at lower educational levels. Pupils in low‐attainment educational programmes score higher on the weapon items than pupils in high‐attainment types. Finally, and relatively the least important given the size of the correlation coefficients: as the degree of urbanisation increases, pupils score higher on truancy, possession of drugs at school and possession and sale of weapons at school. The sale and use of drugs, but also the perception of school measures against drugs and truancy, are more typical of pupils in rural areas than of those in big cities.

3.3 Pupil characteristics and motives for verbal violence

The next series of Pearson correlations refers to the six pupil characteristics on the one hand, and being a victim, offender or witness of verbally violent behaviour according to the possible complementary roles of other pupils, teachers, other school personnel and pupils’ relatives, on the other hand. For each of these 12 specifications, Pearson correlations were calculated at individual level with respect to each of the 16 motives. Table 5 presents the main results for the total number of pupils, number of pupils relevant to each of the four victim–offender, offender–victim and witness–victim combinations for verbally violent behaviour, and the numbers of significant motive correlation coefficients found per combination for the specified six pupil characteristics. Only statistically significant correlation coefficients (p < .05, two‐sided) are counted. The + or − sign before the numbers indicates (generally) positive or negative types of relationships between the variables constituting the correlation coefficients in Table 5.

Table 5. Numbers of significant Pearson correlations between pupil characteristics and specified motives with respect to particular combinations of verbal violence.

Person himself/herself (No. of pupils = 26,726)	Age 9–24	Boy–girl	I feel at home in the Netherlands(yes–no)	I am religious (no–baptised–yes)	Level of educational attainment (low–high)	Urbanisation (big city–rural)
Victim (N)	Offenders
3229	Pupils	+10	−12	+11	+4	−13	−2
367	Teachers	+4	−1	+11	+2	−8	0
255	Other personnel	0	−2	+4	0	−1	0
220	Pupils’ relatives	−1	0	+10	+1	−4	0
Offender (N)	Victims
2287	Pupils	+6	−13	+6	0	−14	0
731	Teachers	+11	−14	+7	−1	−13	+1
316	Other personnel	+6	0	0	−1	−2	−1
203	Pupils’ relatives	0	−1	0	−2	0	0
Witness (N)	Victims
7622	Pupils	+6	−12	+7	−4	−13	+3
1307	Teachers	+2	−12	−1	−1	−12	−1
509	Other personnel	0	−12	−2	+1	−5	−5
369	Pupils’ relatives	0	−1	+1	0	0	−2

Table 5 shows that, as expected, motives are especially relevant in violent verbal relationships between pupils and between pupils and teachers in all three victim–offender, offender–victim and witness–victim combinations. Important in this respect is the educational attainment level: attaining relatively lower is associated with expressing more violent motives between pupils and between pupils and teachers in all three combinations.

In addition, not feeling at home in the Netherlands, being a boy, and being relatively older are all related to experiencing more verbal violence. Of the six pupil characteristics, being religious and degree of urbanisation are least important, with contrasting plus and minus signs across the three victim–offender–witness combinations.

One notable outcome is that motives do not play a role in the victimisation of boys and girls by teachers (the respective cell indicates −1), whereas 14 out of 16 motives are significant when boys victimise teachers (see the cell indicating −14). In addition, older pupils are more likely than younger pupils to have motives for behaving with verbal violence towards other school personnel (+6).

Boys are more likely than girls to perceive motives behind the verbal victimisation of other school personnel (−12). Compared with the pupils who feel at home in the Netherlands, pupils who do not feel at home are more likely to perceive motives behind verbal victimisation by other school staff and pupils’ relatives.

The specific motives attributed for verbal violence generally refer to social discrimination based on physical appearance and country of origin, being religious, gender, being gay or lesbian, or being bi‐sexual (for details, see Mooij 2007). Pupils also perceive teachers as perpetrating verbal violence because other pupils did not follow the school rules, or because other pupils did not accept disciplinary measures to correct their behaviour.

Compared to younger pupils, older pupils believe teachers discriminate on the basis of country of origin, skin colour, handicap or gender (being male). Girls report social discrimination on the basis of physical appearance.

3.4 Pupil characteristics and motives for material violence

Table 6 reports the results on material violence. Here the motives related to not feeling at home in the Netherlands and attaining relatively lower in education are the most important in all three victim–offender–witness combinations. This is particularly clear with respect to all four specifications of victimisation by other pupils, teachers, other personnel and pupils’ relatives. Concerning not feeling at home in the Netherlands, the significant correlations range from +10 to +16 in number, which implies that pupils perceive almost all specific motives as relevant. Being older comes third, being a boy comes fourth and being religious fifth. Here, we should note that victimised religious pupils attribute many motives to offending pupils (+14) and offending pupils’ relatives (+5), whereas offending religious pupils also give motives for their victimising other pupils (+8), pupils’ relatives (+8), other personnel (+7) and teachers (+3). Degree of urbanisation is of least importance.

Table 6. Numbers of significant Pearson correlations between pupil characteristics and specified motives with respect to particular combinations of material violence.

Person himself/herself (No. of pupils = 26,470)	Age 9–24	Boy–girl	I feel at home in the Netherlands (yes–no)	I am religious (no–baptised–yes)	Level of educational attainment (low–high)	Urbanisation (big city–rural)
Victim (N)	Offenders
1415	Pupils	+10	−8	+16	+14	−14	−1
233	Teachers	−1	+1	+12	+1	−10	0
166	Other personnel	+5	0	+10	+2	−4	−1
130	Pupils’ relatives	+15	+3	+13	+5	−7	0
Offender (N)	Victims
687	Pupils	+15	−7	+12	+8	−16	0
308	Teachers	+13	0	+14	+3	−11	+3
198	Other personnel	+4	0	+12	+7	−2	0
120	Pupils’ relatives	0	−1	+2	+8	−7	0
Witness (N)	Victims
2967	Pupils	+10	−11	+12	0	−10	−2
664	Teachers	+3	−10	+4	+2	−8	0
415	Other personnel	+2	−8	+3	−1	−3	+2
272	Pupils’ relatives	0	−2	+1	0	−4	0

Generally, the most relevant motives in Table 6 are those that play a role between the pupils themselves, and only then between the pupils and teachers. Also of note is that pupils’ relatives tend to perpetrate material violence when the victimised pupils are older (+15), female (+3) and religious (+5). On the other hand, offending pupils who are religious (+8) and who attain relatively lower in education (−7) give motives for victimising pupils’ relatives.

Specific information about the motives behind material violence reveals that, according to the pupils, being victimised materially by teachers occurs among other things when pupils try to take care for other pupils’ conforming behaviour to the school rules, and because other pupils did not agree to punishment exercises obliged by the school.

3.5 Pupil characteristics and motives for social violence

Table 7 presents the motive results on social violence. It illustrates that, according to the pupils, socially violent behaviour is most common between the pupils themselves. In addition, victimised pupils attribute motives to teachers who exhibit socially violent behaviour against them. Pupil characteristics such as being older, being a boy, not feeling at home in the Netherlands, being religious and attaining relatively lower in education are relatively important in attributing specific motives for being a victim, perpetrator or witness of socially violent behaviour. There is little evidence that the relationships between pupils and other school staff are based on socially relevant motives. Similar to Tables 5 and 6, degree of urbanisation is of least importance.

Table 7. Numbers of significant Pearson correlations between pupil characteristics and specified motives with respect to particular combinations of social violence.

Person himself/herself (No. of pupils = 26,185)	Age 9–24	Boy–girl	I feel at home in the Netherlands (yes–no)	I am religious (no–baptised–yes)	Level of educational attainment (low–high)	Urbanisation (big city–rural)
Victim (N)	Offenders
2177	Pupils	+11	−8	+7	+4	−3	0
237	Teachers	+3	−5	+4	+4	−3	0
255	Other personnel	+3	0	+2	0	−3	−2
133	Pupils’ relatives	+2	−2	+2	+2	0	+2
Offender (N)	Victims
1233	Pupils	+10	−15	+3	+3	−13	+3
205	Teachers	+4	−1	+1	+2	+1	0
103	Other personnel	0	−4	+5	0	+1	+2
66	Pupils’ relatives	0	−4	+5	−1	+1	−6
Witness (N)	Victims
4393	Pupils	+3	−10	+12	+1	−9	+2
491	Teachers	0	−7	+1	+1	+2	+1
221	Other personnel	0	0	+1	+2	+1	−1
204	Pupils’ relatives	+6	−2	0	+4	+3	−1

3.6 Pupil characteristics and motives for mild physical violence

Results on the motives for mild physical violence are given in Table 8. The outcomes show that the relationships between pupils (victim–offenders; offender–victims; witness–victims) are particularly relevant here. The relationships are the same as previously: pupils attaining relatively low in education, those who do not feel at home in the Netherlands, pupils who are older, and boys tend to attribute more motives than their respective counterparts when it comes to mild physical violence.

Table 8. Numbers of significant Pearson correlations between pupil characteristics and specified motives with respect to particular combinations of mild physical violence.

Person himself/herself (No. of pupils = 26,173)	Age 9–24	Boy–girl	I feel at home in the Netherlands (yes–no)	I am religious (no–baptised–yes)	Level of educational attainment (low–high)	Urbanisation (big city–rural)
Victim (N)	Offenders
2663	Pupils	+10	−10	+14	+4	−15	−5
303	Teachers	+2	−4	+7	+10	0	0
208	Other personnel	+2	+1	+9	+2	−1	0
188	Pupils’ relatives	+1	+2	+3	+1	−1	0
Offender (N)	Victims
1878	Pupils	+15	−6	+15	+1	−16	+1
275	Teachers	+4	−1	0	0	−5	+1
201	Other personnel	+2	−2	+1	0	−5	+3
179	Pupils’ relatives	+11	0	0	0	−6	+1
Witness (N)	Victims
6445	Pupils	+6	−12	+6	0	−14	+6
499	Teachers	+2	+1	+1	+1	0	0
355	Other personnel	+1	0	+8	+8	0	−2
320	Pupils’ relatives	+1	+1	+2	0	−3	−1

Not feeling at home in the Netherlands and being religious are also associated with motives for violence between victimised pupils and teachers, and victimised pupils and other personnel. Older offending pupils tend to give motives for perpetrating violence on other pupils and pupils’ relatives.

Victimised pupils attribute motives to other pupils, teachers, other personnel and pupils’ relatives when they do not feel at home in the Netherlands. However, they themselves give motives for perpetrating mild physical violence only on other pupils. Victimised religious pupils attribute motives to teachers, other pupils, other personnel and pupils’ relatives for mild physical violence, but they do not attribute motives to themselves (where relevant) when they are the offenders.

Moreover, Table 8 demonstrates that victimised pupils who attain relatively low in education only attribute motives for mild physical violence used against them to other pupils (−15). However, when they are the offenders, these pupils give many motives for using mild physical violence against other pupils (−16), teachers (−5), other personnel (−5) and pupils’ relatives (−6).

Finally, degree of urbanisation is of relative importance with respect to incidents of mild physical violence. More so than victimised pupils in rural areas, victimised urban pupils attribute the following motives to offending pupils for their aggressive behaviour: their poor school achievement, handicap, not being religious, trying to get other pupils to conform to the school rules and correcting other pupils’ behaviour. However, offending pupils in rural areas are more likely than offending urban pupils to motivate mild physical violence against other school staff by referring to physical appearance, skin colour and country of origin. Pupils in rural areas are more likely than urban pupils to regard physical appearance, skin colour, country of origin, being religious and being gay or lesbian as the motive behind mild physical violence they have witnessed against other pupils; they mention poor school achievement as a motive less than urban pupils do.

3.7 Pupil characteristics and motives for severe physical violence

Table 9 shows the numbers of motives perceived to be significantly relevant with respect to severe physical violence. There are clearly fewer incidents reported and fewer motives given here than in Tables 5–8. The most important numbers in relative terms refer to the motives involved between pupils in their various roles. With respect to the various pupil characteristics, age is of little relevance. The same is true for educational attainment. Here, only being a witness of severe physical violence against other pupils is of importance (−10).

Table 9. Numbers of significant Pearson correlations between pupil characteristics and specified motives with respect to particular combinations of severe physical violence.

Person himself/herself (No. of pupils = 26,242)	Age 9–24	Boy–girl	I feel at home in the Netherlands (yes–no)	I am religious (no–baptised–yes)	Level of educational attainment (low–high)	Urbanisation (big city–rural)
Victim (N)	Offenders
544	Pupils	+1	−4	+3	+4	0	−1
76	Teachers	0	+1	+3	0	−1	−1
63	Other personnel	0	0	+1	0	0	0
71	Pupils’ relatives	+1	−2	0	−1	+1	0
Offender (N)	Victims
519	Pupils	+2	0	+4	0	0	−4
116	Teachers	0	+1	0	0	−2	−3
72	Other personnel	+1	0	0	0	−1	+1
63	Pupils’ relatives	0	0	0	0	−1	0
Witness (N)	Victims
2557	Pupils	+3	−11	+6	−2	−10	+2
145	Teachers	+1	−5	+1	0	+1	0
112	Other personnel	+1	−3	0	+4	+1	0
105	Pupils’ relatives	−1	0	0	0	0	+4

Degree of urbanisation is relevant with respect to being an offender or a witness of severe physical violence (see Table 9). Offending pupils in rural areas are more likely than offending urban pupils to give the following four motives for severe physical violence: the other pupils’ poor school achievement, not being religious, being male and being female; the pupils give comparable motives when teachers are the victims (the one exception is being male).

Finally, compared to urban pupils who witness severe physical violence against other pupils, pupils in rural areas who witness such violence attribute skin colour and country of origin as motives; severe physical violence against pupils’ relatives is motivated by physical appearance, type of behaviour exhibited, the other person wanting them to conform to the rules and the other person wanting them to correct their behaviour.

3.8 Pupil characteristics and motives for sexual violence

Table 10 shows the numbers of motives given for sexual violence between the various social actors in and around schools. The table illustrates that age is a relevant pupil characteristic in all 12 types of relationships. Victimised boys are more likely than victimised girls to attribute motives to other pupils’ sexually violent behaviour (−15) and to sexual violence against teachers (−6). On the other hand, girls are more likely than boys to give motives for offending teachers sexually (+4), and they are also more likely than boys to attribute specific motives to teachers for sexually violent behaviour (+5).

Table 10. Numbers of significant Pearson correlations between pupil characteristics and specified motives with respect to particular combinations of sexual violence.

Person himself/herself (No. of pupils = 25,764)	Age 9–24	Boy–girl	I feel at home in the Netherlands (yes–no)	I am religious (no–baptised–yes)	Level of educational attainment (low–high)	Urbanisation (big city–rural)
Victim (N)	Offenders
363	Pupils	+13	−15	+11	0	−10	0
120	Teachers	+11	−6	+2	0	−3	0
49	Other personnel	+1	0	0	0	0	+1
48	Pupils’ relatives	+16	0	+9	+1	−2	−1
Offender (N)	Victims
184	Pupils	+11	+2	+3	+4	−16	0
60	Teachers	+15	+4	0	0	−15	−1
34	Other personnel	+10	+1	0	+3	−12	0
25	Pupils’ relatives	+1	0	0	0	−10	0
Witness (N)	Victims
644	Pupils	+1	−1	+3	−2	−11	0
42	Teachers	+3	+5	+11	+2	0	0
37	Other personnel	+8	0	0	0	−2	0
31	Pupils’ relatives	+11	+1	+5	+2	−6	−1

Pupils who do not feel at home in the Netherlands consistently perceive motives as underlying sexually violent behaviour between pupils, between pupils and teachers, and when pupils’ relatives are involved (see Table 10). Being religious does not seem to be that important here, while attaining relatively low in education is very important, in particular, for sexually violent behaviour against other pupils (−16), teachers (−15), other personnel (−12) and pupils’ relatives (−10). Degree of urbanisation is hardly or not relevant.

4 Discussion

The research focuses on relationship patterns between the personal, cultural, educational and demographic characteristics of secondary pupils and their problem social behaviour and the motives behind their being a victim, perpetrator or witness of six types of violence, in relation to the complementary social roles of other pupils, teachers, other school staff and pupils’ relatives. Types of violent behaviour concern verbal, material, social, mild physical, severe physical and sexual violence. Secondary analyses were carried out using reliable data from a Dutch nationwide monitor study of school safety in secondary education. The Internet‐based monitor study (2006) was completed by 80,770 pupils. Pupil participation was representative for the educational attainment level, and the participation of school locations was representative for the degree of urbanisation. Significant correlation results at the individual or pupil level illustrate the following main results:

•	pupil characteristics indicating educational attainment level (low), feeling at home in the Netherlands (not feeling at home), gender (male) and age (older) are most important in motive patterns; being religious is less relevant in evoking violence‐related motives, and degree of urbanisation is least relevant;
•	relatively most significant correlations concerning motives for violence are found with respect to verbal, material, social and mild physical violence; then comes sexual violence, and severe physicial violence comes in the last place;
•	with respect to the 16 motives specified, in particular, those referring to physical appearance, behaviour, school achievement, handicap, being religious, gender, sexual preference and ways of dealing with non‐conforming behaviour or punishment are mentioned most by the pupils;
•	furthermore, a rather broad array of motives are given for every type of violence, which implies that, in general, it is not the actual motive but a complex set of psychologically related arguments that are used to discriminate socially between specific persons or groups in socially, institutionally or culturally relevant ways. These interpersonal or intergroup relationships reflect various types of social cohesion, power or violence interaction characteristics between pupils, pupils and teachers, pupils and other staff, and pupils and pupils’ relatives, respectively; and
•	the empirical results also show that social interactions between pupils themselves, and between pupils and teachers, are generally most important for eliciting violence‐related motives, followed by the relationships between pupils and pupils’ relatives. Relationships between pupils and other school staff are least relevant.

Although our specifications of the 16 motives in the 12 social interaction combinations appear to be unique, the present results on the individual level relevance of (low) attainment in education, feeling at home in the Netherlands, gender and age agree with those of comparable research (e.g. Beirn, Kinsey, and McGinn 1972; Fekkes 2005; Mooij 1994, 2001; National Education Association 1994; Schuyt 1995). These findings support aspects of the external validity of this Internet‐based research. Validation is also supported by comparing the results of principal factor analysis carried out on school‐level data from three nationwide cohorts carried out in 1993, 2000 and 2006 of pupils, teachers, other staff and school management (cf. Mooij 1994, 2001, 2006). With safety data of all respondents aggregated to school level, factor results at the school level revealed social mirroring processes between pupils and between pupils and teachers and other staff in particular. This means that in schools where pupils show more problem social behaviour, teachers and other staff do the same, and the other way around. This mirroring phenomenon can be interpreted as indicating the degree of social cohesion or connectivity between the different school partners. Moreover, schools characterised by higher levels of problem social behaviour are also characterised by lower levels of educational attainment and by being smaller in size: they are attended by lower numbers of pupils. This last empirical fact has been found consistently and is contrary to the general stereotype that larger schools are responsible for more unsafe or problem social behaviour of pupils.

The development and initial results of the Internet‐based monitor study on school social safety also lead to several conclusions. First, from a methodological perspective, the empirical evidence demonstrates that main conceptual specifications and concrete operationalisations of variables were reliably implemented in a coherent system for assessment with pupils (see Tables 1 and 2). Comparable evidence was demonstrated for data of teachers and educational support staff, and school management (see Mooij 2006). The method resulted in response numbers and variable specifications that would have been difficult or impossible to achieve with printed questionnaires. The validity of the Internet‐based method was supported by comparing factor analysis outcomes from printed questionnaires (1993, 2000) with those from Internet‐based questionnaires (2006).

A second conclusion concerns the substantive results. The nationwide monitor study provides us with a more complete, thorough and systematic picture of the social discrimination and motive aspects related to social safety in and around school than has hitherto been customary. The secondary analysis clarified the relevance of the 16 motives or characteristics that pupils use in explaining their role as a victim, offender or witness of six different types of socially safe/unsafe or violent behaviour. Moreover, comparable information was also supplied by the teachers and educational support personnel, underlining the relevance of this procedure in assessing mutual interaction, social cohesion and social discrimination processes in and around schools.

Third, the research identifies contexts where improvements can be realised concerning social safety in and around school. In particular, in special needs education and other low‐attainment school programmes, and with respect to pupils not feeling at home in The Netherlands, psychological, pedagogical and educational processes require closer examination to determine how to improve support for pupils, teachers and parents (cf. also Cowie et al. 2008). There have already been some proposals for such investigations and systematic improvements (Hermanns, Öry, and Schrijvers 2005; Mooij and Smeets 2009; Sherman et al. 1998; U.S. Department of Health, Education and Welfare 1973). The most important among these are proposals to develop and implement early, preventive practices to reduce the incidence of truancy and the possession and use of weapons and drugs. Chen (2006) and Lim and Deutsch (1996) provide research overviews to demonstrate that specific social or safety policy measures lead to the effective improvement of social safety between pupils and between pupils and teachers. Teachers and educational support staff, but also parents, must work together from the very start of a pupil’s school career, and be given earlier and more effective support than is usually the case (see also Bogenschneider 2002; Chapman and Harris 2004; Georgiou 2008; Verdurmen et al. 2003; Vollebergh et al. 2006).

Fourth, the methodological features of the present monitor may also assist to improve social safety and social cohesion in schools where this issue seems to be at stake. The present monitor features make it possible to provide feedback to individual school locations on the Internet about their own results among pupils, teachers, educational support staff and school management. The feedback covers school location parameters and the respective national benchmarks, including indicators specified per educational attainment level and school year. These indicators can assist participating school locations in interpreting and evaluating their own school safety in various respects, and in defining the next policy steps to improve safety where needed. The expectation is that combining pedagogical, social and curriculum differentiation measures and integrating positive pupil social responsibilities where possible will go furthest in helping pupils function more effectively (Alschuler 1980; Howard and Jenkins 1970; Kirschner 1997; Lodewijks 2008; Mooij 1999a, 1999b; Salmivalli, Kaukiainen, and Voeten 2005; Stevens, de Bourdeaudhuij, and van Oost 2000). At the same time, specifically advancing and strengthening socially competent behaviour can preclude antisocial behaviour or an increase in such behaviour (Förrer, Kenter, & Veenman 2000; Sørlie, Hagen, and Ogden 2008). Next, Internet‐based measures of school safety can then be used to help evaluate whether or not the policy measures were successful, or should be followed up by more intensive or other school measures.

Notes on contributor

Ton Mooij is a professor of educational technology at the Open University of The Netherlands, CELSTEC (Heerlen) and a manager of educational research at the Radboud University, ITS (Nijmegen), The Netherlands.