Term Paper on "Bulling and Academic Performance"

Quoted Instructions for "Bulling and Academic Performance" Assignment:

For the article below, please provide a review that includes the following;

1. Incidence rates (measurement issues, definitional issues, perceptions)

2. Causation (who is blamed is another way to look at this)

Prevention efforts (evidence of effectiveness – for whom, under what conditions)

3. Intervention (what level - 1-1, group, policy; evidence of effectiveness – see above; ramping up?)

4. School/college climate issues (if they apply)

Concept of victimization

5. What do we know regarding race, sex, age, SES

Direct and relational bullying among primary school

children and academic achievement

Sarah Woodsa,*, Dieter Wolkeb

a Department of Psychology, University of Hertfordshire, UK

bDepartment of Community Based Medicine, University of Bristol, ALSPAC, UK

Received 28 March 2003; received in revised form 30 July 2003; accepted 8 December 2003

Abstract

The association between bullying behaviour and academic achievement was investigated in 1016

children from primary schools (6–7-year-olds/year 2: 480; 8–9-year-olds/year 4: 536). Children

were individually interviewed about their bullying experiences using a standard interview. Key Stage

I National Curriculum results (assessed at the end of year 2) were collected from class teachers, and

parents completed a behaviour and health questionnaire. Results revealed no relationship between

direct bullying behaviour and decrements in academic achievement. Conversely, higher academic

achievement at year 2 predicted bullying others relationally (e.g. social exclusion at year 4).

Relational victimisation, Special Educational Needs (SEN), being a pupil from a rural school or

small classes and low socioeconomic status (SES) predicted low academic achievement for year 2

children. Findings discount the theory that underachievement and frustration at school leads to direct,

physical bullying behaviour.

D 2004 Society for the Study of School Psychology. Published by Elsevier Ltd. All rights reserved.

Keywords: Bullying; Victimisation; Academic achievement; Special needs; Primary school students

According to Olweus (1993, 1999), victimisation refers to a student being exposed to

negative actions on the part of one or more other students’ with the intention to hurt.

Bullying must be a repeated action and occur regularly over time (Olweus, 1999) and it

usually involves an imbalance in strength, either real or perceived (*****, 1998;

Whitney & Smith, 1993). Bullying can be physical, verbal, or relational (Bjo¨rkqvist,

0022-4405/$ - see front matter D 2004 Society for the Study of School Psychology. Published by Elsevier Ltd.

All rights reserved.

doi:10.1016/j.jsp.2003.12.002

* Corresponding author. Computer Science Department, University of Hertfordshire STRC, Hatfield

Campus, College Lane, Hatfield, Herts AL10 9AB, UK. Tel.: +44-1707-281-133; fax: +44-1707-285-073.

E-mail address: s.n.woods@herts.ac.uk (S. Woods).

Journal of School Psychology

42 (2004) 135–155

1994; Bjo¨rkqvist, Lagerspetz, & Kaukianen, 1992), whereby relational victimisation is

defined as the purposeful damage and manipulation of peer relationships leading to

social exclusion (Crick & Grotpeter, 1995). The first two forms of victimisation have

sometimes been labelled as ‘‘direct bullying’’ as they include direct aggressive acts

such as hitting, kicking, pinching, taking belongings or money, pushing or shoving, or

direct verbal abuse (name calling, cruel teasing, taunting, threatening, etc.). In contrast,

relational or ‘‘indirect’’ bullying refers to social exclusion by spreading malicious

gossip or withdrawal of friendships (Wolke, Woods, Bloomfield, & Karstadt, 2000).

The prevalence of physical and verbal victimisation in primary schools has revealed

ranges from 8% to 46% and for bullying others between 3% and 23% across countries

worldwide (Smith et al., 1999; Wolke & Stanford, 1999; Wolke, Woods, Schulz, &

Stanford, 2001).

Initial research studies into the characteristics of bullying behaviour solely considered

those children who were ‘pure’ bullies or ‘pure’ victims. However, more recently, research

has revealed that a sizeable group of children cannot be simply classified as ‘pure’ bullies or

‘pure’ victims but both bully other children and are victimised at other times, and have been

termed bully/victims (Kumpulainen et al., 1998; Sutton & Smith, 1999; Wolke & Stanford,

1999; Wolke et al., 2000). Bully/victims have been found to be a distinct group of children

in terms of their behavioural characteristics (Schwartz, 2000) and are rated as being the least

popular by peers (Forero, McLellan, Rissel, & Bauman, 1999; Wolke & Stanford, 1999),

easily provoked, provoke others and are hot tempered (Schwartz, 2000). Furthermore,

bully/victims are characterised as having more behaviour problems with hyperactivity,

impulsivity, and conduct problems compared to ‘pure’ victims, ‘pure’ bullies, or neutral

children (Duncan, 1999; Kumpulainen et al., 1998; Wolke et al., 2000).

Less is known about the characteristics of those children involved in relational

bullying and an inconsistent profile emerges. Children who are relationally aggressive

have been found to be less liked by other children (Crick & Grotpeter, 1995, 1996)

and there is also evidence that relational aggression is related to maladjustment in

terms of depression, loneliness, anxiety, and social isolation (Bjo¨rkqvist, 1994; Crick,

Casas, & Hyon-Chin, 1999). In contrast, other findings have revealed that those

children who are relational ‘pure’ bullies are physically healthy, enjoy going to

school, have few absenteeism’s from school, have few behaviour problems in terms

of hyperactivity and conduct problems, but are characterised by low prosocial

behaviour (Wolke et al., 2000; Wolke, Woods, Bloomfield, & Karstadt, 2001; Wolke,

Woods, Schulz, et al., 2001).

Bullying behaviour is a social, group process that is prevalent in the school environment

and there are well documented findings regarding the behavioural and health

consequences of bullying behaviour at school for both direct and relational bullying

profiles (Kumpulainen et al., 1998; Owens, Slee, & Shute, 2000; Williams, Chambers,

Logan, & Robinson, 1996; Wolke et al., 2000). However, there is a dearth of research that

has considered the association between bullying behaviour per se and academic achievement

among primary school children.

Olweus (1978, 1983) first speculated that aggressive behaviour of bullies towards

peers could be considered as a reaction to frustrations and failures at school. However,

data from a large sample of boys from Greater Stockholm provided no evidence to

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 136

suggest that aggressive behaviour was a consequence of poor grades at school. Rather, it

was found that both bullies and victims had lower than average marks than neutral

children (Olweus, 1978).

In a recent study, Schwartz, Farver, Chang, and Lee-Shin (2002) reported that children

who exhibited poor academic performance in school tended to emerge as frequent targets

of bullying. However, it was only a subset of victimised children, the aggressive victims

(or bully/victims) who were likely to be characterised by poor school performance

(Schwartz, 2000). What remains to be established by research studies is whether poor

academic achievement leads to bullying involvement or whether being bullied leads to

poorer school achievement, possibly mediated by less participation in school.

Research on peer rejection has also considered the relationship to academic

achievement and school adjustment. Peer rejection is predominantly assessed by standardised

scores that are comparable across classes and school, but does not take into

account individual bullying roles within classes. Ladd (1990) considered the academic

behaviour and school adjustment of children over the first year of school life and

reported that rejected children had less favourable school perceptions, significantly

higher levels of school avoidance and significantly lower school performance compared

to popular, average, and neglected children. In a similar vein, DeRosier, Kupersmidt, and

Patterson (1994) assessed three different cohorts of children over 4 consecutive years

and revealed that children who had experienced peer rejection displayed higher levels of

absenteeism, and that the youngest cohort of children who were more chronically

rejected, performed more poorly on academic tests compared to those children who had

never been rejected. However, it was concluded that there was no direct relationship

between peer rejection and later academic achievement, and instead an indirect pathway

was proposed. Further evidence for this contention was derived from a study by Buhs

and Ladd (2001) who found that the effects of peer rejection on children’s adjustment

were partially mediated through the processes of maltreatment and reduced classroom

participation.

Research studies to date have solely considered direct bullying and peer rejection,

and no consideration for the possible association between relational bullying status and

academic achievement is evident. The profile of relational ‘pure’ bullies remains

controversial, and academic achievement results for these individuals would shed light

on the overall characteristic profile for these children. For example, if relational ‘pure

bullies are found to be characterised by outstanding academic achievement, this would

provide further support for the claim by Sutton, Smith, and Swettenham (1999) that

‘pure’ bullies or ring leaders are socially intelligent and have superior theory of mind

skills resulting in enhanced manipulation skills in peer group situations.

To further complicate the picture, academic achievement and school adjustment

have been employed interchangeably within research studies, which is potentially

problematic as both place different emphasis on aspects of children’s academic school

life. Academic achievement has been assessed by school records of composite

achievement test scores (DeRosier et al., 1994), Task Mastery subscale of the

California Preschool Social Competence Scale, (Ladd, 1990), verbal and quantitative

subtests of the Metropolitan Readiness Tests (Buhs & Ladd, 2001; Tremlow et al.,

2001), and categorical variables rated by teachers as ‘this child is excellent’, ‘this

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 137

child is a good student’, or ‘this child has difficulties with school work’ (Schwartz,

2000; Schwartz et al., 2002).

Furthermore, no studies have considered the sole importance of academic achievement

and school adjustment in relation to bullying behaviour, but have instead included these

factors as much smaller aspects of larger research studies. School adjustment variables

have included negative school attitudes, social adjustment at school, school avoidance,

child–teacher relationships, cooperative classroom participation, involvement in school

activities, and adaptation to the classroom environment (Blankemeyer, Flannery, &

Vazsonyi, 2002; Buhs & Ladd, 2001; Ladd & Burgess, 2001).

Measurements that assess academic ability by tests and teacher assessments in the

same way across different schools are needed in order to provide consistency. The

National Curriculum Standard Assessment Tasks (SATs) employed to assess children’s

academic achievement in the U.K., combine tests with teacher assessments. Despite

scepticism, reliability on the SATs tests have revealed high values (Hurry, 1999;

Reeves, Boyle, & Christie, 2001). One strength of SATs assessments is that they

sample a broad range of skills compared to other standardised tests. SATs tests also

enable researchers to identify subtle and relative changes such as a decline in standards

(Hurry, 1999).

In summary, the literature reveals a noticeable gap concerning the possible causal

pathways between bullying behaviour and academic performance and the direction of

influence, i.e. does being involved in direct or relational bullying behaviour as a bully,

victim, or bully/victim contribute to the prediction of academic achievement or alternatively,

does poor academic achievement predict involvement in bullying behaviour?

Furthermore, previous studies have often concentrated on small sample sizes, on preschool

or grade one children, failed to consider relational bullying, and solely focused on

peer rejection and sociometric status, rather than bullying behaviour per se.

The current study with primary school children from the U.K. had two major aims:

(1) to assess the relationship between direct and relational bullying behaviour, and SATs

tests results and teacher assessments; (2) consider variables that predict children’s SATs

results and teacher assessments and determine whether SATs results and teacher

assessment results contribute to the prediction of being involved in direct and/or

relational bullying.

Method

Population

Parents of children in 82 classes in 39 primary schools in Hertfordshire and North

London in the United Kingdom were approached. Five schools (eight classes) failed to

provide Key Stage Level I National Curriculum results (SATs) resulting in an 87.2% return

rate and a total of 74 classes from 34 primary schools. Table 1 illustrates the sample data

for age, gender, non-consent rates, Special Educational Needs (SEN), and children with

full data sets (bullying interview, SDQ behaviour questionnaire, health questionnaire, and

SATs results).

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 138

Procedure

The study received ethical permission from the University of Hertfordshire Ethical

Committee and all instruments and information was lodged with the Hertfordshire

Education Council. When the head teacher and class teachers consented to participate

in the study written information about the study and a non-consent form (parents were

asked to sign when they wanted their child not to take part) was passed to all parents via

the pupils in sealed envelopes. On prearranged dates all pupils were interviewed

individually in a private room in the school by one of four trained interviewers

(postgraduate psychologists). The school was asked to provide a copy of the Key Stage

1 National Curriculum and Teacher Assessment results for those children that had

participated in the bullying interview.

Instruments

Bullying interview

Children were interviewed individually in a quiet, private room within the school

using a standard structured interview that enquired about friendships and peer relationships

(previously described in detail, see Wolke et al., 2000; Wolke, Woods, Bloomfield,

et al., 2001; Wolke, Woods, Schulz, et al., 2001). The part of the interview that is

subject to this report was adapted from the Olweus (1993) Bullying Questionnaire. First,

children were asked whether they had experienced any of six behaviours in the last 6

months that had upset them: (1) having been called bad or nasty names, (2) having

belongings taken, (3) having lies told about them, (4) having nasty tricks played on

them, (5) having been threatened or blackmailed, (6) having been hit or beaten up. If the

child answered that s/he had experienced any of the six above behaviours, the child was

Table 1

Sample data

Variable/instrument Sample size (n)

Total available sample (bullying interview, SDQ N: 1639

behaviour questionnaire, health questionnaire) N: 722 (44.1%), year 2

N: 917 (55.9%), year 4

Total sample with complete data sets (SATs results,

bullying interview, SDQ behaviour questionnaire,

N: 1016 (62% of total

available sample)

health questionnaire) N: 480 (47.2%), year 2

N: 536 (52.8%), year 4

SATs return rate N: 34/39 schools (87.2%)

N: 74/82 classes

Age Overall: x: 7.52, S.D.: 1.02

year 2: x: 6.62 years, S.D.: 0.53

year 4: x 8.33 years, S.D.: 0.58

Gender Males: n: 498 (49%)

Females: n: 518 (51%)

SEN Yes: n: 44 (4.1%)

No: n: 972 (95.9%)

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 139

asked to give examples and describe how this happened. This was done to ascertain that

the behaviours experienced were carried out with intent by the perpetrator(s) to upset the

child rather than having occurred by accident or during play fighting, etc. Researchers

were also able to ensure that there was an imbalance of power between the perpetrator

and victim. Those children who had experienced one or more of these behaviours were

asked how frequently these incidents happened in the last 6 months (seldom: one to

three times during past 6 months, frequently: four times or more during past 6 months,

very frequently: at least once per week). To aid children’s reference to approximately 6-

month periods, anchors such as ‘‘since last Christmas’’, ‘‘since the summer holidays’’,

etc. were used. The six behaviours were then repeated and the child was asked whether

they have used these behaviours to upset other children and how often they had done

this over the last 6 months (never or seldom: one to three times during past 6 months,

frequently: four times or more during past 6 months, very frequently: at least once per

week).

Subsequently, children were asked four questions relating to relational bullying at

school: (1) other children saying that they did not want to play with them; (2) other

children saying that they would not be the child’s friend anymore; (3) other children telling

nasty stories that were not true about them; (4) Other children deliberately spoilt their

games. If the child responded that they had experienced any of the above behaviours, the

child was asked to supply a description with examples. This was carried out to ensure that

the behaviours had been deliberate, that there was a perceived imbalance of power, and

to ascertain that the perpetrator(s) were children that the child normally played with.

Children were then asked to express how frequently the incidents occurred in the last

6 months for each of the four questions.

Children’s frequency responses were subsequently coded according to three categories,

seldom, frequently, and very frequently (seldom: one to three times during the past 6

months, frequently: four times or more during the past 6 months, very frequently: at least

once per week).

The four types of relational bullying were then repeated to the child and they were

asked whether they had ever used any of the behaviours to upset other children over the

past 6 months (never or seldom: one to three times during the past 6 months, frequently:

four times or more during the past 6 months, very frequently: at least once per week).

At no time during the interview was the term ‘bullying’ used, i.e. only behavioural

(operational) descriptions were used.

According to the results of the interview and the frequency of bullying events reported,

children were classified using a standardised coding manuscript into the following groups

(Whitney & Smith, 1993; Wolke & Stanford, 1999; Wolke et al., 2000) for physical direct

bullying and relational bullying, separately: physical bullies [children who were involved

in physically bullying others frequently (at least four times during past 6 moths) or very

frequently (at least once every week) but are never or only rarely physically victimised],

physical victims (children who experienced any of the above described behaviours: being

called bad/nasty names, being threatened, having belongings stolen, having lies told about

them, being hit/beaten, having nasty tricks played on them, frequently or every week but

bully others rarely or never); physical bully/victims (children who both physically bully

others and become physical victims of the six described behaviours frequently or every

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 140

week), physical neutrals who neither physically bully others nor become physical victims

(never or rarely only).

For relational bullying the same classification system was employed to categorise

children as: relational bullies, relational victims, relational bully/victims, and relational

neutrals.

Each post-graduate psychologist received intensive training in a standard interviewing

and coding protocol which required them to carry out pilot supervised bullying

interviews and subsequently code them. Any inconsistencies or discrepancies were

individually discussed and rectified until there were no problems before the main study

commenced. The basis of determining whether behaviour constituted bullying and the

bullying classifications was made according to guidelines stipulated in the research

coding manual which as a result of a pilot study clearly outlined direct and relational

bullying behaviours. Researchers used standardised probing questions to determine the

frequency of bullying behaviours and researchers always asked the children to back up

statements by describing the behaviour they had made and to ensure consistency in

responses. Behaviours were recorded in standard datasheets and any difficulties in

decision making discussed in the coder group (consensus rating). Interviews were not

audio taped as this would have required a second consent form which was likely, from

previous experience to lead to higher and selective drop out (which ideally should be

avoided).

The bullying interview has shown high predictive validity for direct and relational

victimisation over 2–4 years (Wolke, Woods, & Samara, article in submission). Concurrent

validity is evident between bullying classification established from the bullying

interview (bully, victim, bully/victim, and neutral) and sociometric status (popular,

average, neglected, rejected, and controversial status) (Wolke & Stanford, 1999; Wolke,

Woods, Kropp, & Schulz, unpublished manuscript).

Behaviour questionnaire

The Strengths and Difficulties Questionnaire (SDQ) (Goodman, 1997) was given to

parents to complete after children had been interviewed. The SDQ comprises of 25 SDQ

items falling into 5 scales of 5 items each: conduct problems, hyperactivity, emotional

symptoms, peer problems, and prosocial behaviour (Goodman, 1997). For each scale,

except for prosocial, higher scores indicate more problems. A total difficulties behaviour

score is computed by combining all scales except from the prosocial behaviour scale.

The current study reports on total difficulties behaviour scores only. The SDQ correlates

highly (>.80) with the Child Behaviour Checklist (CBCL) total score and discriminates

as well as the Rutter Behaviour Scales (Elander & Rutter, 1996) and the Achenbach

(1991) CBCL between children with clinical behaviour problems and no problem

behaviour children. Goodman (2001) reported high reliability and validity of the SDQ

regarding Cronbach a, cross-informant correlations, and retest stability over 4–6

months.

For categorical analysis, Goodman (1997) suggested the following bandings for the

total behaviour problems subscale: 0–80th percentile are normal, between 81 and 90th

percentile are borderline, and those >90th percentile are in the clinical range. All

children in the study were classified according to the Goodman bandings as either

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 141

having normal/borderline behaviour difficulties or behaviour difficulties within the

clinical range.

Health questionnaire

Parents of children also completed a health questionnaire (Wolke, Woods, Bloomfield,

et al., 2001) that comprised of two sections: (a) seven items about physical health

problems (PHP) over the previous 6 months rated on a seven-point scale (none to six or

more times) (headache, tummy ache, sore throat/ear ache, cold/cough, feeling sick,

breathing problems, skin problems); (b) seven items about emotional health problems

(EHP) (previously known as psychosomatic health problems) rated according to a fivepoint

scale (never to most days) (bed wetting, problems going to sleep, nightmares, woken

in the night, poor appetite, excessive appetite, worried about going to school). Two

variables were constructed to examine the total amount of physical and emotional health

problems that children experienced.

Key Stage 1 national curriculum assessments

The Key Stage 1 National Curriculum Assessment (SATs 1) (www.qca.org.uk;

www.dffes.gov.uk/ncat) for 7-year-olds (year 2) comprised of five tests: (1) The Writing

Task (levels 1–3), (2) Spelling Task, (3) The Reading Comprehension Task (level 2 and

level 3), (4) The Reading Task (levels 1 and 2), and (5) Mathematics Task. The Writing

Task examines children’s ability to plan and write a non-narrative piece of work in a

single session. The length, sentence structure, vocabulary, punctuation, spelling, and the

structure and organisation of the piece of work are assessed. The Spelling Task consists

of 30 words and assesses a range of letter combinations and regular and irregular

spelling patterns. The Reading Comprehension Task tests children’s literal comprehension

in terms of retrieving information from a preset text, ability to make simple

inferences, understanding of the text’s organisation, and the author’s use of language.

The Mathematics Test is split according to levels. Level 1 assesses children’s ability to

identify properties of 3D shapes, money, and working with larger numbers. Level 2–3

examines children’s knowledge of number, calculations in terms of addition and

subtraction, multiplication and division, problem solving, handling data, and shape,

space and measures. The tests are not based on a multiple choice format, but require the

child to write their answers on standard examination forms and demonstrate their

workings out for questions where appropriate.

Results from each of the tests are individually graded as follows: working towards

grade 1(W), grade 1, grade 2C, grade 2B, grade 2A, grade 3, and grade 4. Children

who received a ‘W’ may have special educational needs and were underachieving.

Grade 1 is below the national curriculum average, grade 2C is slightly below the level

that should be achieved in accordance with national curriculum average, grade 2B is in

line with the national curriculum average, and grade 2A is slightly above the national

curriculum average. Grades 3 and 4 are substantially above the national curriculum

average. The following categories were formed for the analysis of SATs test results

(TR): (1) working towards level 1 or below level 1, (2) low level 2 (2C), (3) average

level 2 (2B), (4) high level 2 (2A), (5) above level 2 (levels 3 and 4), (6 missing) did

not sit SATs tests.

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 142

Summative teacher assessment (TA) for Key Stage 1

Teacher assessments for children at Key Stage 1 are carried out at the end of a unit

or the school year in order to make judgements about pupils’ performance in relation

to national set standards. TA initially occurs in the form of formative assessment

which happens throughout the year in the classroom and involves the teacher and

pupil in a process of continual reflection and review about progress, leading up to the

Summative TA at the end of the year. TAs are based on level descriptions and are

frequently assigned a numerical value. The following teacher assessments were

recorded: (1) an overall English assessment, (2) speaking and listening, (3) reading,

(4) writing, (5) spelling, (6) mathematics (included: using and applying mathematics,

number, shape, space, and measures), and (7) science (included: experimental and

investigative science, life processes and living things, materials and their properties,

physical processes).

The following grading system was used to assess each teacher assessment: grade 0

to grade 4. In addition, children were rated according to whether they did not sit the

tests due to behavioural problems and whether they were absent on the day of the

assessment. (labelled ‘‘disapplication due to absenteeism’’). Children who had missed

a substantial amount of the curriculum relevant to their year group were exempt from

sitting the SATs tests. Level 0 denotes a child who is working towards level one and

usually has special educational needs. Level 1 refers to a child who is working below

the national average. Level 2 represents the national curriculum average. Levels 3 and

4 are both above the national curriculum average. For quantification purposes derived

from guidelines from the QCA (www.qca.org.uk), the following categories were

formed for TA statistical analysis: (1) working towards level 1 (W) or below the

threshold (X), (2) level 1 or (L) (substantially below curriculum average), (3) below

threshold of 2 but reached level 1 (not quite reached curriculum average, e.g. 2C),

(4) level 2 (2B or 2A, curriculum average), (5) level 3 or 4 (above curriculum

average).

Although National Curriculum SATs TR and TA are universally employed as the

standardised measure of assessment in U.K. schools, there are few studies which have

considered their reliability and validity. Hurry (1999) reported that the actual tests that

make up part of the SATs levels (SATs TR) have face validity according to curriculum.

Reliability has been measured using internal consistency and Cronbach’s a’s of between

.77 and .87 have been reported (Cronbach, 1960).

Reeves et al. (2001) argue that the teacher assessment (SATs TA) is an essential part of

the National Curriculum assessment arrangements.

The results of end of year key stage teacher assessment are reported alongside the test

results. Both have equal status and provide complementary information about the

pupils’ attainment. The tests provide a standard ‘snapshot’ of attainment at the end of

the key stage, while teacher assessment, carried out as part of teaching and learning in

the classroom, covers the full range and scope of the programmes of study, and take

account of evidence of achievement in a range of contexts, including that gained

through discussion and observation. (QCA, 1999, p. 7)

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 143

Rose (1999) stated that TAs are an under used source of information and better use

should be made of them, particularly in the test development phase. An important question

concerns the relationship between TR and TA. Teacher expectations and test results are not

completely independent of one another and teachers are able to consult TR to assist them

with TA. Reeves et al. (2001) explored the relationship between SATs TR and SATs TA for

Key Stage 2 results for over 6000 children. Comparison of SATs TR and TA for individual

pupils revealed a remarkably high level of consistency across years in all three subjects

(Math, English, and Science) ranging from 73% to 77%. Where a difference was found

between TR and TA this was nearly always just by one level (e.g. level 2 on English TA

and level 3 on English TR). Tymms (1996) reported exact agreement between TR and TA

for 74% of pupils and agreement within one level for 99.6% from a study involving over

7000 pupils from schools in Avon, U.K.

The result from the study by Reeves et al. (2001) and Tymms (1996) suggest that a

certain level of non-agreement is not only acceptable but desirable, otherwise one measure

becomes redundant. The SATs TR and TA appear to be complimentary assessment

measures, which usually concur but have variability to justify the application of both.

However, what remains unknown is how many teachers consult the SATs TR before

making their own TA.

Cronbach a (Cronbach, 1960) for the current sample revealed high internal consistency

for SATs TR and SATs TA (Math TR and Math TA: a=.77; English TR and English TA:

a=.92).

SEN

Information was recorded for those children who were deemed as having SEN.

Guidelines set out by the British Education Act (1996) to identify and assess SEN were

adhered to. There are 5 recommended stages for addressing the different levels of

children’s SEN. Stages 1–3 are school based, with support from specialists from outside

the school at Stage 3. Stage 4 involves the Local Education Authority (LEA), which

considers whether a multidisciplinary assessment of the child’s needs is necessary. Stage 5

involves the LEA issuing and monitoring a statement of SEN. Children at stages 2–5 were

considered as having SEN in the current sample, where the decision was made for a child

to receive extra support either from within the school or an outside party.

School variables

Data was collected from schools regarding social class distribution (records of free

school lunches were noted), school location in terms of urban (inhabitants of community>

50,000) or rural (inhabitants of community<50,000), and mean school and class size.

School absenteeism for each pupil was also recorded.

Statistical an*****s

First, frequency analysis was carried out for the SATs TR and TA. As a result of this, it

was decided that spelling TA (N: 229), Reading Task (N: 960), and Spelling Test (N: 711)

should be dropped from the final analysis due to substantially reduced sample sizes. Factor

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 144

analysis (extraction method: principal component analysis with varimax rotation) was

computed to determine whether TR and TA constituted the same or different factors.

Differences in SATs results according to direct and relational bullying status for year 2 and

year 4 on standardised scores from the factor analysis were computed using one-way

ANOVA. Final data analysis was carried out only for children who had full data sets (N:

1016) (bullying interview, SATs TR and TA, SDQ behaviour questionnaire, and health

questionnaire).

Prospective analysis was carried out between bullying status assessed before sitting

SATs tests (year 2 children) using ANOVA with bullying status as the IV and the SATs

result as the DV. Furthermore, logistic regression an*****s were computed to determine the

best combination of factors predicting SATs TR and SATs TA. For year 4 children, logistic

regressions were carried out with SATs results at year 2 as predictors for bullying status at

year 4 as the DV (Fig. 1).

Results

SATs TR and TA scales

Factor analysis (principal components analysis) with varimax rotation including nine

variables from the SATs results revealed two distinct factors, SATs TA and SATs TR,

respectively. Eigenvalues for the solution were 4.7 and 1.7 accounting for 70.7% of

the total variance. Factor I accounted for 51.7% of variance and Factor II 19% of

variance. All variables loaded onto the primary factors at .59 or greater (Table 2).

Factor I had six primary loadings (Table 2), and constituted all SATs TA items and

three variables loaded on the second factor (Table 2) identified as SATs TR items.

Fig 1. Sample characteristics for statistical analysis.

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 145

Standardised factor scores with a mean of 0 and S.D. of 1 were computed for SATs

TR and SATs TA.

SATs results and the relationship with bullying and victimisation behaviour for year 2 and

year 4 children

Table 3 illustrates the frequencies for direct and relational bullying classifications.

SATs TR

ANOVA an*****s carried out on standardised factor scores for SATs TR revealed no

significant differences between bullying status (bully, bully/victim, victim, neutral) for

both direct [F(3,770)=0.33, p<.80] and relational bullying [ F(3,769)=0.52, p<.67] for

year 2 children (6–7-year-olds). In contrast, children in year 4 (8–9-year-olds) who were

involved in relational bullying had significantly higher SATs TR 2 years previously (in

year 2) than those children not involved in relational bullying, F(3,890)=4.25, p=.005.

A posteriori contrasts (Tukey-HSD test) indicated that those children in year 4 who

relationally bullied other children had significantly higher SATs TR previously in year 2

compared to relational victims and neutral children (relational bullies M=0.60 vs.

relational victims M=_0.12 and neutrals M=_0.05).

SATs TA results

ANOVA an*****s carried out on standardised factor scores for SATs TA revealed no

significant differences between year 2 children and SATs TA for direct [ F(3,714)=0.15,

p=.93] and relational bullying status [F(3,713)=0.52, p=.67]. However, significant differ-

Table 2

Factor analysis of SATs Teacher Assessment (TA) items and SATs TR items (N: 1016)

Item Factor 1 Factor 2

English TA .93 .15

Speaking and listening TA .86 .10

Reading TA .86 .16

Writing TA .85 .17

Math TA .81 .11

Science TA .78 4.87(E_02)

Reading comprehension TR 8.44(E_02) .86

Writing task TR .12 .90

Math TR .12 .59

Table 3

Physical and relational bullying classifications

Bully Victim Bully/victim Neutral

Direct bullying (N: 1015) N: 35 (3.4%) N: 419 (41.3%) N: 111 (10.9%) N: 450 (44.3%)

Relational bullying (N: 1014) N: 11 (1.1%) N: 424 (41.8%) N: 60 (5.9%) N: 60 (5.9%)

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 146

ences were uncovered for SATs TA and relational bullying status for year 4 children

[F(3,782)=3.78, p=.01].

The a posteriori contrasts highlighted that year 4 children involved in relationally

bullying others had significantly higher SATs TA results 2 years earlier than those

children who were relationally victimised or not involved in any relational bullying

(neutral) (relational bullies M=0.63 vs. relational victims M=_0.08 and neutrals

M=_0.05).

Predictors of children’s SATs results

Logistic regression an*****s were carried out to determine the best combination of

factors predicting the following dependent variables: (1) SATs TR in year 2 and (2)

SATs TA in year 2. The independent variables were categorised as follows for year 2

an*****s predicting SATs TR and SATs TA: (a) whether the child was involved as a

relational victim (yes vs. no); (b) whether the child was involved as a relational bully

(yes vs. no); (c) whether the child was involved as a direct victim (yes vs. no); (d)

whether the child was involved as a direct bully (yes vs. no); (e) gender (boy vs. girl);

(f) behavioural problems (clinical vs. normal/borderline); (g) total number of emotional

health problems (no emotional health problems vs. at least one emotional health

problem); (h) total number of physical health problems (0–2 physical health problems

vs. 3–6 physical health problems); (i) school absenteeism (0–11 vs. 12–21 days); (j)

ethnic minority (native English vs. non English native); (k) school location (urban vs.

rural); (l) social class (SES) (lower vs. upper/middle); (m) class size (small vs. medium/

large); (n) school size (small vs. medium/large); (o) whether the child had a statement

of SEN (SEN vs. no SEN).

First a full model was built forcing, all independent variables into the prediction

function and then removing those variables (backward stepping), which did not make a

significant contribution to the model (no significant change in fit when removing

variables).

SATs TR

The final model for predicting SATs TR is shown in Table 4 [v2(507)=40.17,

df=6, p=.000]. Factors which had a significant impact on predicting SATs TR were

Table 4

Final logistic regression models for predicting SATs TR (Year 2) (Backward Stepwise Method) (N: 507)

Predictor B S.E. Wald df Significance Exp. (B) 95% C.I. for Exp. (B)

Lower Upper

SEN (yes) _1.74 0.51 11.86 1 .001 0.18 0.07 0.47

Relational victim (yes) 0.99 0.35 7.93 1 .005 2.71 1.35 5.40

School location (rural) 0.91 0.41 4.94 1 .026 2.49 1.11 5.56

Class size (small) _0.83 0.37 5.01 1 .025 0.44 0.21 0.90

SES (lower) _0.77 0.35 4.71 1 .030 0.46 0.23 0.93

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 147

in order of importance: SEN (odds ratio: 0.18, CI (95%): 0.65–0.47), relational

victimisation (odds ratio: 2.71, CI (95%): 1.35–5.42), school location (odds ratio:

2.49, CI (95%): 1.11–5.56), class size (odds ratio: 0.44, CI (95%): 0.21–0.90),

and social class (odds ratio: 0.46, CI (95%): 0.23–0.93). Children were more

likely to underachieve on SATs TR if they had SEN compared to no SEN, were

relationally victimised compared to not victimised, went to rural schools compared

to urban schools, were in small class sizes compared to medium/large class sizes,

and went to schools in lower social class areas compared to middle/upper class

areas.

SATs TA

The final model for predicting SATs TA was significant [v2(462)=26.43, df=3, p=.000].

Only one factor remained in the final model and had a significant impact in predicting

SATs TA, SEN (odds ratio: 0.19, CI (95%): 0.75–0.49). Children were more likely to be

low achievers on SATs TA if they had SEN compared to no SEN.

Predictors of children’s bullying status

Logistic regression an*****s were carried out to determine the best combination of

factors predicting the following dependent variables in year 4: (1) direct bullying

involvement, (2) direct victimisation, (3) relational bullying involvement, (4) relational

victimisation.

The independent variables were the same as for the SATs results, but with the omission

of bullying status, and the inclusion of (a) achievement on SATs TR (average/high

achievement vs. low achievement), (b) achievement on SATs TA (average/high achievement

vs. low achievement).

Direct bullying

The final model for predicting direct bullying involvement in year 4 was significant

(v2(514)=21.67, df=2, p=.000]. Two factors had a significant impact in predicting

involvement in direct bullying: gender (odds ratio: 0.35, CI (95%): 0.21–0.57) and class

size (odds ratio: 0.56, CI (95%): 0.33–0.96). Those children involved in direct bullying

were more likely to be boys rather than girls, and from small class sizes compared to

medium or large class sizes.

Direct victimisation

The final model for predicting direct victimisation in year 4 was significant

[v2(514)=36.34, df=3, p=.000]. Three factors had a significant impact on predicting

direct victimisation: behaviour problems in the clinical range (odds ratio: 0.23, CI

(95%): 0.12–0.45, school location (odds ratio: 1.74, CI (95%): 1.20–2.52), and gender

(odds ratio: 0.62, CI (95%): 0.43–0.89). Direct victims were more likely to be

children who had behaviour problems within the clinical range compared to normal/

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 148

borderline, came from rural schools compared to urban schools, and were boys rather

than girls.

Relational bullying

A significant model was revealed for predicting involvement in relational bullying

[v2(514)=17.50, df=4, p=.002] (Table 5). Those factors that had a significant impact in

predicting involvement in relationally bullying others were in order of importance: high

achievement on SATs TR (odds ratio: 5.74, CI (95%): 0.77–42.72), gender (odds ratio:

0.42, CI (95%): 0.21–0.86), class size (odds ratio: 0.48, CI (95%): 0.23–0.99), and

emotional health problems (odds ratio: 0.50, CI (95%): 0.25–0.99). Children who were

involved in relationally bullying others were more likely to have average/above average

achievement compared to low achievement on SATs TR, be boys rather than girls, came

from schools with small class sizes as opposed to medium/large class sizes and had at least

one emotional health problem compared to no emotional health problems.

Relational victimisation

A significant final model for predicting involvement in relational victimisation in year 4

was revealed [v2(514)=23.44, df=4, p=.000]. Factors that had a significant impact in

predicting relational victimisation were in order of importance: school absenteeism (odds

ratio: 2.18, CI (95%): 1.12–4.25), behaviour problems (odds ratio: 0.52, CI (95%): 0.30–

0.93), school size (odds ratio: 0.60, CI (95%): 0.41–0.88), and school location (odds ratio:

1.48, CI (95%): 1.02–2.14). Children involved in relational victimisation were more likely

to have had fewer days (0–11) off school (compared to more days (12–21)), had

behaviour problems within the clinical range compared to the normal/borderline range,

came from small schools compared to medium/large sized schools, and went to a rural

school compared to an urban school.

Discussion

The present study investigated the association between direct and relational bullying

behaviour, and academic achievement among primary school children in the U.K.

Table 5

Final logistic regression models for predicting relational bullying (Backward Stepwise Method) (N: 514)

Predictor B S.E. Wald df Significance Exp. (B) 95% C.I. for Exp. (B)

Lower Upper

SATs TR

(average/above average)

1.75 1.02 2.91 1 .088 5.74 0.77 42.72

Gender (boys) _0.87 0.36 5.69 1 .017 0.42 0.21 0.86

Class size (small) _0.74 0.37 3.93 1 .048 0.48 0.23 0.99

Emotional health problems

(at least 1 problem)

_0.70 0.35 3.94 1 .047 0.50 0.25 0.99

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 149

Predictors of academic achievement and being involved in bullying behaviour were also

examined. The major findings of the present research study were:

1. There was a higher incidence of direct bullying behaviour among primary school

children compared to relational bullying.

2. No association between direct bullying and academic achievement was uncovered at

year 2. However, relational ‘pure’ bullies in year 4 had significantly higher SATs TR

and SATs TA at year 2 compared to victims and neutral children.

3. Important predictors of academic achievement for year 2 children were relational

victimisation, SEN, rural schools, small classes, and low socioeconomic status (SES).

4. Important predictors of involvement in bullying behaviour in year 4 were small classes,

behaviour problems, rural schools, being male, having average/above average

achievement, small schools, and emotional health problems.

The finding that primary school children were involved in more direct bullying than

relational bullying fits in with a developmental explanation of bullying behaviour.

Bjo¨rkqvist (1994) proposed that young children lack essential verbal skills, which results

in aggressive behaviour being predominantly physical in nature. Once verbal skills and

more importantly complex social skills develop, children will demonstrate more

sophisticated styles of aggression such as relational aggression, usually in secondary

school.

This is the first study to report results concerning the relationship between direct and

relational bullying behaviour and academic achievement among primary school children

in the U.K. No significant associations were revealed for children in years 2 and 4 and

direct bullying status for ‘pure’ bullies, ‘pure’ victims, bully/victims, and neutral

children and academic achievement as assessed by SATs TR and SATs TA. In contrast,

children in year 4 who were involved as relational ‘pure’ bullies had had significantly

higher SATs TR and SATs TA results in year 2. Olweus (1993) argued that aggressive

direct bullies’ behaviour could be explained as a reaction to frustration and failure at

school although future research studies were unable to confirm this claim (Tremblay et

al., 1992; Olweus, 1994). The current findings provide further substantiation that having

lower academic grades does not seem to be related to involvement in direct bullying

behaviour. No support was generated for the findings by Schwartz et al. (2002) where

children who exhibited poor academic performance in school tended to emerge as

frequent targets of bullying.

A number of explanations could account for the lack of association between direct

bullying and academic achievement. Direct victims and neutral children did not differ

from each other in terms of academic ability. One account previously proposed is that

children who are victims of bullying actually turn to school work and related activities

as an escape route from the problems they are experiencing with bullying behaviour

(Sharp, 1995) therefore increasing their academic capabilities and grades. Research

findings from work carried out on peer rejection and school adjustment could provide an

alternative explanation. Buhs and Ladd (2001) reported no direct pathway between peer

rejection and school adjustment including academic achievement and instead proposed a

model of mediation by other factors such as peer maltreatment and reduced classroom

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 150

participation. The current study did not consider levels of classroom participation or

cooperation and therefore no direct conclusions can be made. The school milieu and

ethos in the current study may have contributed to the lack of association between direct

bullying and academic achievement. Twemlow et al. (2001) carried out a highly

controlled intervention study based on zero tolerance for bullying. A significant increase

in academic performance and a reduction in disciplinary referrals was found in the

experimental school, but not in the control school. Schools in the current sample may

have had a similar ethos towards bullying behaviour.

The relationship uncovered between being a relational bully and obtaining significantly

higher academic achievement is new to the literature and proffers several

explanations. The current controversy regarding the individual characteristics of bullies

and whether they are ‘cool’ planners of their torment towards others, or are anxious,

depressed, insecure individuals with behaviour problems is one account for the current

findings. Support is generated for the theory proposed by Sutton and Smith (1999) and

Sutton et al. (1999) that bullies, in particular those children who use social manipulation,

are socially skilled and intelligent individuals who avoid being caught in the act of

bullying other children. Previous findings have also revealed that ‘pure’ relational bullies

enjoy going to school, have few days absent from school and are physically and

psychologically healthy (Wolke, Woods, Bloomfield, et al., 2001; Wolke, Woods,

Schulz, et al., 2001). The current findings add substance to the profile of ‘pure’

relational bullies and their strong and successful demeanour in school environments.

Conversely, the findings do not support the assertion that bullies lack self-esteem, are

anxious and depressed individuals (Farrington, 1993, 1995; Salmon, James, & Smith,

1998, Salmon, James, Cassidy, & Javaloyes). The developmental concept that relational

bullying follows on from direct bullying is also supported by the present results in that it

was only year 4 children who were classified as relational ‘pure’ bullies that had

significantly higher academic results (Bjo¨rkqvist, 1994). Children in year 4 (aged 8–9)

are more likely to have the developed social skills to use relational bullying as opposed

to younger children in year 2. It appears that children with high academic abilities have

the social skills available to employ relational manipulation.

Bullying role for both direct and relational bullying, with the exception of relational

victimisation, was not an important predictor for children’s academic achievement.

Children were nearly three times more likely to have under achieved on SATs TR if they

were relational victims. Relational victimisation could have more deleterious effects on the

child’s academic achievement in terms of firstly being more hurtful and long-term than

direct physical bullying and secondly, could be more classroom focused compared to

direct bullying, therefore distracting the child from concentrating on school tasks.

Interestingly, school absenteeism and health problems did not contribute to the prediction

of SATs TR or SATs TA.

Previous research has often discounted the association between bullying behaviour and

school factors such as school size, class size, school location, and SES. Results reported by

Wolke et al. (2001) found more bullying problems in small schools and classes in rural

locations compared to large urban schools. In a similar vein, the current study found that

small schools, small class sizes, rural locations and schools with lower SES significantly

contributed to the prediction of lower SATs TR. Indirect pathways between bullying

S. Woods, D. Wolke / Journal of School Psychology 42 (2004) 135–155 151

behaviour and school characteristics may be having an impact on SATs results. The nature

of these pathways is unknown resulting in the need for research studies to determine the

underlying factors that characterise these schools in terms of lower school attainment and

bullying problems.

A discussion of the limitations of the current study includes the reliance on individual

interviewers to classify children as being direct or relational bullies, victims, bully/victims,

or neutral. Ideally, reliability an*****s based on tape recorded interviews should have been

conducted across researchers to ensure that direct and relational bullying classifications

were assigned consistently and reliably. However, this was not within the capability of the

study for several reasons. Firstly, due to the study taking place within a school setting,

teaching staff would not permit the same child to be interviewed about bullying

experiences on repeated occasions by different researchers. This was deemed unethical

and impractical due to time constraints. Secondly, the research team did consider and

approach schools about the possibility of videotaping and audio taping each interview to

permit reliability testing. Again, teachers had serious reservations about breaching children’s

confidence and viewed this practise as unethical. Furthermore, this required a second

consent form and thus likely increased selective dropout of pupils. It would be imprudent

to assume that the bullying interview methodology is entirely objective as bullying

behaviour due to its complex social nature and the fact that 6–9-year-olds took part is

bound to reveal some subjective issues. Nonetheless, the bullying interview is highly

structured and all researchers received a highly structured comprehensive training

schedule. A pilot study revealed 90% agreement across bullying classifications across

four interviewers. The procedural manual was adhered to throughout the study, which

contained concrete examples of the distinctions between what constitutes bullying

behaviour and what does not. Any problems were discussed and consensus coding

applied. Interviewer ratings are at least as valid and reliable as would be obtained by

questionnaire and the interview method has the advantage to not be influenced by reading

age. The age range of children chosen for the current study only allows conclusions about

the association between bullying behaviour and academic achievement in primary school.

Other studies may want to consider adolescent samples in addition to primary school aged

children. Furthermore, the predictive power of bullying behaviour for academic achievement

could only be carried out with children in year 2 and the time lag between bullying

and SATS assessment was less than 1 year. Future studies may consider prospective

investigation of the causal pathways over a longer period of time. Finally, despite the

application of national tests with hundred of thousands of children every year, there are

few studies on the validity and reliability of SATs TR and TA. However, the ecological

validity is high as national markings of schools’ and supply of financial resources are

dependent on SATS results.

In sum, the current study provides empirical evidence that direct bullying behaviour is

not largely associated with decrements in academic achievement in primary school. No

support was generated for the theory that under achievement and frustration at school leads

to direct physical bullying behaviour in 6–9-year-olds. Conversely, it was ascertained that

relational bullies are often average or high achievers. Important factors contributing to the

prediction of SATs results were revealed including SEN, behaviour problems and school

characteristics such as school size, class size, and school location.