BCS-A Thomas 2019
BCS-A Thomas 2019
BCS-A Thomas 2019
*Correspondence should be addressed to Hannah J. Thomas, University of Queensland Centre for Clinical Research, Herston,
71/918 Royal Brisbane and Women’s Hospital, Herston, QLD, 4029, Australia (email: hannah.thomas@uq.edu.au).
DOI:10.1111/bjep.12223
2 Hannah J. Thomas et al.
school connectedness, social support, and personality), as well as predictive validity over
6 months.
Conclusions. The BCS-A has sound psychometric properties. This tool establishes
measurement equivalence across types of involvement and behavioural forms common
among adolescents. An improved measurement method could add greater rigour to the
evaluation of intervention programmes and also enable interventions to be tailored to
subscale profiles.
Current study
Measurement methods have critical implications for study results. Complex and multi-
faceted constructs such as bullying require greater attention to measurement. Specifically,
there is a need to examine whether there is psychometric support for a multi-dimensional
measurement model based on behavioural form. The aim of Study 1 was to develop a multi-
item traditional bullying and cyberbullying scale that measured both experiences of
victimization and perpetration, and test internal, concurrent, and convergent validity. The
aim of Study 2 was to evaluate two different measurement scales (ordinal and ratio) for
the newly developed self-report tool, and to prospectively examine predictive validity.
part were (1) too little time in school timetable and (2) already participating in another
research project. Six of the seven private schools had a religious affiliation, and four of
these were Catholic.
Participating schools sent 3,794 invitations to students’ households seeking parental
consent for their child’s participation in an in-school survey and a follow-up survey
6 months later (Study 2). Overall, 36.6% of parents provided consent and of those
87.6% of adolescents aged 12–17 years also consented to take part in the survey
(N = 1,217; Mage = 14 years; 66.2% male). Non-response was predominantly due to
absences on the day of the survey, although a small number withdrew (N = 5). The age
group selected to participate was chosen because they were likely highly familiar with
traditional as well as cyber forms of bullying (Thomas et al., 2017). Over 90% spoke
English at home, 85% were born in Australia, 74.7% had married biological parents
living together, 66.5% lived in urban centres, and 2.1% identified as indigenous. The
national index of relative socio-economic standing indicated that in the current sample
15% were ‘most advantaged’, 31% were ‘advantaged’, 16% were ‘neither advantaged
nor disadvantaged’, 30% were ‘disadvantaged’, and 8% were ‘most disadvan-
taged’(Australian Bureau of Statistics, 2011).
Procedure
University human ethical clearance was obtained, as well as approval from the relevant
education authorities. Parental informed consent was required to participate. Students
who had parental consent were offered an information sheet and consent form. A
researcher explained these documents and reminded participants the survey was
voluntary, they could withdraw at any time, and their responses were anonymous.
Participants completed a pencil-paper survey on bullying and mental well-being. The
study was conducted during school time over a 40-min lesson. Participants were asked to
complete the survey booklet independently, and the researchers were available to answer
any questions from participants. Surveys were administered within the last 2 weeks of a
school term, which run for 11–12 weeks.
Measures
Bullying and cyberbullying Scale for Adolescents (BCS-A)
The initial scale comprised a 20-item victimization scale and corresponding 20-item
perpetration scale (Appendix S1). Items were developed based on the revised version of
the Olweus Bully-Victim Questionnaire (Olweus, 1996), the Peer Relations Questionnaire
(Rigby, 1998), and the Forms of Bullying Scale (Shaw et al., 2013). A definition of bullying
and cyberbullying was presented to establish a shared meaning of bullying across
participants before questions were presented (Appendix S1). Two parallel scales,
victimization and perpetration, were further divided into behaviour that was ‘offline/face-
to-face’ (11 items; traditional bullying subscales) as well as ‘online/on the Internet or
Mobile Phones’ (nine items; cyberbullying subscales). A conceptual diagram of the four
fixed subscales is presented in Figure 1. The reference period for the scale was the past
3 months. Existing bullying measurement scales have traditionally used ‘past 30 days’,
‘past couple of months’, or ‘past Term’ as a reference period. This is generally considered a
reliable timeframe in which to ask respondents about their most recent bullying
experiences (Bovaird, 2010; Olweus, 1996; Shaw et al., 2013). The reference period is
Bullying and Cyberbullying Scale for Adolescents 5
1. Bullying
2. Victimization Perpetration
Figure 1. Conceptual diagram of BCS-A measurement model. Note. BCS-A = Bullying and cyberbul-
lying Scale for Adolescents; C = cyber; Perp = perpetration; T = traditional; Vict = victimization.
1. Bullying as an overarching construct; 2. Bullying by the experience of victimization and perpetration;
3. Bullying as an experience within a traditional versus cyber domain; 4. Four fixed subscales developed for
the BCS-A.
important to consider given the need for greater consistency of measurement across
studies. The BCS-A used the reference period of the past 3 months which fits with school
cycles that use either a four-term system (each term is 10–12 weeks duration or a three-
term system (each 10–14 weeks duration). Responses were recorded on a ratio scale
consistent with ‘how many times’ a given behaviour was experienced (0, 1, 2, 3, 4, 5, . . .
etc.). The initial 2 9 20 multi-item scale was piloted in a group of 12 students (aged
13 years) who completed the questionnaire and were invited to provide verbal feedback
to the researchers on the content and the wording of the definition and individual items.
Based on the group’s feedback, minor changes were made to improve the readability of
the scale.
week’, ‘every few weeks’, ‘about once a week’, ‘several times a week or more’. Items for
each scale were summed and a mean score calculated.
Social support
The classmates and close friends subscales from the Child and Adolescent Social Support
Scale were used to measure perceived social support (Malecki & Kilpatrick, 2002). Each
subscale consists of ten items measured on a 6-point scale, from ‘never’ to ‘always’. Social
support scores were obtained by summing the items such that higher scores indicate
greater social support.
School connectedness
The brief five-item School Connectedness Scale was used to measure the bond felt by
students towards their school (McNeely, Nonnemaker, & Blum, 2002). Responses are
measured on a 5-point scale from strongly disagree to strongly agree, and summed
together such that higher scores indicate greater connectedness to school.
Personality dimensions
The revised Junior Eysenck Personality Questionnaire-short form (Corulla, 1990; Eysenck
& Eysenck, 1975) was used to measure three personality dimensions: (1) psychoticism –
proneness to take risks, impulsiveness, anti-social or non-conformist behaviour; (2)
neuroticism – ease and frequency with which an individual becomes upset and
experiences negative emotions; and (3) extraversion – predisposition to outgoing social
behaviour. Each subscale comprises 12 dichotomous true/false items referring to
behavioural tendencies. Six items were reverse scored, and scores for each subscale were
summed such that higher scores indicate greater trait tendency.
Data analysis
Exploratory factor analyses and regression analyses were performed using SPSS Version
24. The distributions of all BCS-A items were significantly negatively skewed. Less than
0.5% of responses were Winsorized to limit the maximum value to 10, which fell in the
96th percentile or higher on all items. This was applied to lessen the impact of outliers on
the sample (Bieniek, 2016). Two random samples were drawn from the dataset to conduct
Exploratory Factor Analyses (EFAs) – labelled Data A (N = 613), and Confirmatory Factor
Analyses (CFAs) – labelled Data B (N = 604). A series of EFAs were conducted to explore
the potential factor structure of each of the four fixed subscales (Figure 1). The adequacy
Bullying and Cyberbullying Scale for Adolescents 7
of extraction and number of factors to retain were examined using four methods:
eigenvalues greater than one, Velicer’s minimum average partial correlation (MAP) test,
parallel analysis, and the scree test (Tabachnick & Fidell, 2013).
Confirmatory Factor Analyses were analysed in R Version 3.2.1, package extension
lavaan (Rosseel, 2012). Statistical estimates robust to non-normal distributions were
used in order to account for negative skew. Models were evaluated using a range of fit
measures: comparative fit index (CFI; values ≥ .95 indicating good fit) (Hu & Bentler,
1999), root-mean-square error of approximation (RMSEA; values < .07 are considered
acceptable, .08 to .10 are mediocre, and >.10 are poor) (Ho, 2006), and the standardized
root-mean-square residual (SRMR, values < .08) (Hu & Bentler, 1999). Non-significant
robust p-value for the Satorra-Bentler scaled chi-squared test was another index of model
fit (Satorra & Bentler, 1994). In the first step, models were tested to confirm factor
structure, and in the second step, models were tested to improve fit by eliminating poor
performing items. Modification indices and expected parameter change values were used
to decide which items to remove (Saris, Satorra, & van der Veld, 2009). The four subscales
were then tested by reducing the original scales item-by-item until the final models were
determined. The factor structure of the final models was then assessed for invariance by
sex.
Concurrent validity was evaluated in a series of hierarchical multiple regression
analyses, adjusting for age and sex. The BCS-A was compared to other bullying measures
using SDQ-Internalizing and SDQ-Externalizing as dependent variables. Convergent
validity was also examined using hierarchical multiple regression analyses to assess
hypothesized relationships between the BCS-A and measures of mental health, social
support, school connectedness, and personality dimensions, controlling for age and sex.
Finally, prevalence of bullying victimization and perpetration was estimated using the cut-
off of ‘three or more times’ in the past school term to meet the defining criterion of
repetition.
Results
Construct validity
Exploratory factor analysis
Using Data A, Principal Axis Factoring analyses with oblique rotation were conducted to
estimate the number of factors for traditional victimization (BCS-A Vict–T) and cyber
victimization (BCS-A Vict–C), and the same for traditional perpetration (BCS-A Perp–T)
and cyber perpetration (BCS-A Perp–C) scales (Appendix S2). Two factors with
eigenvalues >1 were identified for the two traditional bullying subscales, and one factor
with an eigenvalue >1 was identified for the two cyberbullying subscales. Three alternate
tests for factor retention were also run: Velicer’s MAP test, parallel analysis, and the scree
test. Results are reported in Appendix S2. Although a range of recognized statistical
approaches were conducted, no definitive conclusion could be drawn from the EFA
analyses.
BCS-A Sample Model Items Factored items SRMR Robust p Robust CFI Robust RMSEA
Vict – T Data B 1F 11 All items load onto one factor .07 < .001 .83 .14
2F 11 A to G (seven items) F1- direct; .07 < .001 .92 .10
Hannah J. Thomas et al.
Notes. BCS-A = Bullying and cyberbullying Scale for Adolescents; CFI = Comparative Fit Index; F = factor; Perp = perpetration; P = physical; RMSEA = root-
mean-square error of approximation index; SRMR = standardized root-square residual; Vict = victimization. This table presents the models conducted to
determine the number of factors to retain. These are not the final measurement models. This was the first stage of the analytic process using CFA methodology.
N = 604 (Data B – Random Sample at Time 1).
a
Models retained.
b
Alternative models considered.
Bullying and Cyberbullying Scale for Adolescents 9
establishing the final number of factors could reduce the number of factors retained
(Worthington & Whittaker, 2006). For traditional victimization and perpetration
subscales, one-, two-, and three-factor models were assessed, and for cyber victimization
and perpetration subscales, one- and two-factor models were examined. Four- and five-
factor models of traditional bullying and three-factor models of cyberbullying did not have
a strong theoretical foundation, and so were eliminated on the basis of likely erroneous
model fit overestimation.
The results of the CFA models tested to decide the number of factors to retain are
presented in Table 1. The three-factor models of traditional bullying were retained
because they demonstrated the best statistical fit and could be applied across both
victimization and perpetration scales. The models for cyberbullying were less conclusive,
and the victimization and perpetration subscales preferred one-factor and two-factor
models, respectively. While the two-factor models were plausible (items loaded onto
direct vs. indirect behaviour), a single-factor model of cyberbullying was favoured for the
reason of parsimony. In this first stage of the CFA, some indices indicated mediocre model
fit. However, a second stage of CFA was conducted to determine the items to retain and
further test goodness of fit.
Item determination. The four models (Vict-T – three factors; Vict-C – one factor; Perp-T
– three factors; and Perp-C – one factor) were then tested to decide on the number of items
to retain. Models were systematically tested by sequentially eliminating poor performing
items. Results are reported in Table 2. The final models were established based on
superior goodness of fit: parallel three-factor models of traditional victimization and
perpetration (eight items each), and parallel single-factor models of cyber victimization
and cyber perpetration (five items each). The revised victimization and perpetration
scales comprised 13 items each (2 9 13).
Final measurement model. Parallel higher-order CFAs were then conducted, each
with four specified factors to test the hypothesized overarching measurement
models of victimization and perpetration: physical (four items), verbal (two items),
relational (two items), and cyber (five items). In both models, five pairs of error
terms were correlated to account for matching items across traditional and cyber
domains (Brown & Moore, 2012; Gerbing & Anderson, 1984). The results of the
combined traditional and cyber victimization and perpetration models are reported
in Table 2. A five-factor model was also considered (three factors for traditional
bullying and two factors for cyberbullying), but the addition of a second factor for
cyberbullying did not substantially improve model fit and so the more parsimonious
four-factor model was retained. The standardized maximum likelihood parameters for
the final four-factor models of victimization and perpetration are presented in
Figure 2. The influence of sex on model fit was also examined in separate tests of
model fit using CFA. Robust fit indices indicated equivalent model fit for both males
(SRMR < .05, robust CFI > .96; and robust RMSEA < .06, although robust p-values
were < .05), and females (SRMR < .07, robust CFI > .96; and robust RMSEA < .11,
although robust p-values were < .05).
Based on the results of the selected CFA models, four bullying factors were
created for both victimization and perpetration: physical, verbal, relational, and
cyber. A mean score for each factor was created based on the sum of the factor
10 Hannah J. Thomas et al.
Notes. BCS-A = Bullying and cyberbullying Scale for Adolescents; CFI = Comparative Fit Index;
C = cyberbullying; D = direct behaviour; F = factor; I = indirect behaviour; Perp = perpetration;
P = physical; R = relational; RMSEA = root-mean-square error of approximation index; SRMR =
standardized root-square residual; T = traditional bullying; V = verbal; Vict = victimization.
This table presents the models conducted to determine the number of items to retain. The first set of
CFAs tested separate models for Traditional bullying and cyberbullying. The second set of CFAs
presented in this table tested combined models of Traditional bullying and cyberbullying. N = 604 (Data B
– Random Sample at Time 1). All models used maximum likelihood estimation. For the traditional
victimization and perpetration models, both ‘ignore’ and ‘secret’ were eliminated. For consistency, these
two items were also eliminated from the cyber victimization and perpetration models.
a
Model chosen for traditional and cyber subscales.
b
Model chosen for victimization and perpetration scales.
items divided by the number of factor items. The factor scores for males, females,
Data A, Data B, and the complete sample, along with the zero-order correlations are
presented in Appendix S3.
Bullying and Cyberbullying Scale for Adolescents 11
.32 .63
Threat e1 Threat e1
Figure 2. Final four-factor victimization and perpetration models. Note. Ellipses represent latent
constructs; rectangles indicate measured variables; circles signify residuals. Standardized maximum
likelihood parameters are presented. All parameters are statistically significant at p < .001.
Prevalence of bullying
The final victimization and perpetration scales were analysed to estimate the prevalence
of each of the four forms of bullying. Participants were classified as having experienced
bullying if they indicated a frequency of three or more times for at least one item. For
victimization, 15% experienced verbal, 10.7% for relational, 4.9% for physical, and 4.2%
for cyber bullying. For perpetration, 2.7% reported verbal, 2.5% for relational, 1.5% for
cyber, and 1.4% for relational bullying.
Concurrent validity
Hierarchical multiple regressions were conducted to examine the concurrent validity of
the BCS-A compared to the Olweus global items and the Forms of Bullying Scale (FBS) in
explaining the variance of two dependent variables previously shown to be associated
with experiences of bullying victimization and perpetration: internalizing and external-
izing problems, controlling for sex and age. Demographic variables were entered at Step 1,
the Olweus global item or the FBS were entered at Step 2, and the four BCS-A factors were
entered at Step 3. This was carried out in order to test the unique contribution of the multi-
dimensional measurement structure which enables individual factor associations to be
examined, as compared to unidimensional measurement (Olweus global and FBS). For
both victimization and perpetration, the addition of the BCS-A physical, verbal, relational,
and cyber factors together explained significantly more variance in the dependent
variables (SDQ internalizing and externalizing problems) than the Olweus global
victimization and perpetration items, as well as the FBS-Victimization and FBS-
Perpetration scales (Appendix S4).
Convergent validity
Hierarchical multiple regressions controlling for age and sex were conducted to
investigate the convergent validity of the BCS-A physical, verbal, and relational factors
on a range of dependent variables, and whether the Cyber factor added significantly more
12 Hannah J. Thomas et al.
Measures
Bullying and cyberbullying Scale for Adolescents (BCS-A)
The initial 2 9 20-item BCS-A described in Study 1 was included. In addition, an alternative
version of the BCS-A was presented in a supplementary booklet for participants to
complete. The only difference was that responses were measured on an ordinal scale: ‘this
did not happen to me’, ‘once or twice’, ‘every few weeks’, ‘about once a week’, and
‘several times a week or more’ – the same response categories as the Forms of Bullying
Scale (Shaw et al., 2013). Mean scores for the two versions of the scale were computed
based on the four factors derived in the Confirmatory Factor Analysis conducted in Study 1.
Bullying and Cyberbullying Scale for Adolescents 13
Results
Sample attrition
Participants who were not followed at Time 2 were more likely to be male, older in age,
and have significantly higher scores on the K10 and SDQ-total difficulties at Time 1 than
those who completed surveys at both time points (p < .01). However, there were no
significant differences between groups on the four BCS factors (physical, verbal,
relational, and cyber) for victimization or perpetration.
Predictive validity
The BCS-A assessed in Study 1 (Time 1) was used in a series of hierarchical regressions to
test predictive validity in explaining variance in internalizing and externalizing problems
measured in the follow-up survey (Time 2). For victimization, there was a significant
positive association between physical, verbal, and relational bullying at Time 1, and
internalizing problems at Time 2, even after controlling for sex, age, and internalizing
problems at Time 1. There was also a significant association between Physical and
Relational victimization at Time 1 and externalizing problems at Time 2, controlling for
BCS-A Sample Model Items Estimation SRMR Robust p Robust CFI Robust RMSEA
Notes. BCS-A = Bullying and cyberbullying Scale for Adolescents; CFI = Comparative Fit Index; DWLS =
dedicated ordinal estimator; F =factor; ML = maximum likelihood; Perp = perpetration; RMSEA = root-
mean-square error of approximation index; SRMR = standardized root-square residual; Vict =
victimization. This table presents the CFA results for testing the validity of ordinal versus ratio scaling;
N = 870 (Sample at Time 2 – Study 2).
14 Hannah J. Thomas et al.
age and sex. However, this association was no longer significant after controlling for
externalizing problems at Time 1. There was a significant positive association between
cyber bullying perpetration at Time 1 and internalizing problems at Time 2, and between
physical and verbal perpetration at Time 1 and externalizing problems at Time 2, after
controlling for sex and age. However, these associations were no longer statistically
significant after controlling for internalizing and externalizing problems at Time 1
(Appendix S7).
Discussion
This study developed the Bullying and cyberbullying Scale for Adolescents (BCS-A), a
multi-dimensional measurement model of bullying victimization and perpetration. The
results demonstrated evidence of two parallel 13-item scales measuring victimization and
perpetration experiences, each consisting of four factors: physical, verbal, and relational,
and cyber. The models were supported in both cross-sectional and prospective samples,
indicating satisfactory initial psychometric dependability. The analytic plan was driven by
the theoretical relationship among item measures and specifically the factorization of
behavioural form. Conceptual interpretability was the definitive factor-retention criterion
(Worthington & Whittaker, 2006). Some items were removed from the initial model to (1)
improve fit, (2) preserve equivalency across victimization and perpetration scales, and (3)
maintain the theoretical underpinning of a particular bullying behavioural form and (4)
ensure optimal content coverage of the behavioural domain. In applying the definitional
criterion of repetition to prevalence estimates (reported three or more times), the current
study found that different forms of bullying victimization experiences ranged from 4.2 to
15%, and bullying perpetration from 1.4 to 2.7%.
Evidence increasingly suggests that traditional bullying and cyber bullying should be
measured concurrently (Thomas et al., 2015). In this study, two forms of bullying were
conceptualized as distinct but related constructs. Given there were a number of matching
items across traditional and cyber subscales, initial models were tested separately. Once
factor structure was established, the traditional and cyber subscales were combined and
retested, resulting in a four-factor structure for both victimization and perpetration scales.
This study extends the empirical base by separately examining the dimensional structure
of the two related constructs of traditional and cyber bullying. Existing scale development
and validity studies are limited to traditional domains only, or conflate the measurement of
traditional and cyber experiences within unidimensional multi-items models (Breivik &
Olweus, 2014; Cheng et al., 2011; Kyriakides et al., 2006; Shaw et al., 2013). One
previous study that incorporated both traditional and cyber domains and tested multi-
dimensional structure was for victimization experiences only (Hunt et al., 2012).
externalizing outcomes. This information would not be captured using global bullying
measures.
The BCS-A victimization and perpetration scales were also associated in the expected
directions with a range of theoretically related variables including greater internalizing
and externalizing problems, lower school connectedness, social support from close
friends and classmates (for victimization only), lower extraversion (for victimization
only), higher psychoticism, and neuroticism. These associations are in line with the
existing literature (Bond et al., 2007; Felix et al., 2011; Mitsopoulou & Giovazolias, 2015;
Shaw et al., 2013; Slee & Rigby, 1993; Wang, Iannotti, & Nansel, 2009). Together, these
results suggest suitable convergent validity.
Consistent with longitudinal investigations (Wolke & Lereya, 2015), the BCS-A also
showed acceptable predictive validity for internalizing and externalizing problems. For
victimization, there was a significant positive relationship between physical, verbal, and
relational forms of bullying at baseline and increased internalizing problems at follow-up,
after controlling for baseline internalizing problems, age, and sex. The positive
relationship between physical and verbal perpetration at baseline and externalizing
problems at follow-up was no longer significant after controlling for externalizing
problems. Although this prospective evidence is somewhat limited, the current results
align with other studies that demonstrated that victimization experiences are most
strongly associated with internalizing problems, and perpetration experiences with
externalizing problems (Wolke & Lereya, 2015). Future studies are needed to further
examine the temporal sequence of different behavioural forms and their association with
maladjustment.
Limitations
The sample was non-representative; therefore, it is recommended that the models be
tested in other adolescent samples. It is acknowledged that participants in the current
sample formed only 32% of the target sample. Given the active consent procedures
required in the current study, it is possible that students who experienced bullying
victimization or perpetration were under-represented (Shaw, Cross, Thomas, & Zubrick,
2014). It is also possible that at follow-up the sample included adolescents with higher
levels of functioning, compared to those who dropped out. This would marginally
underestimate reported effects (Wolke et al., 2009). Self-report is a widely accepted
standard of measurement; however, it is acknowledged that by operationalizing bullying
in this way the measurement method is part of the construct being studied. In this way, the
construct being measured was the subjective experience of being bullied and bullying
others (Salmivalli & Peets, 2011).
In addition, it is important to acknowledge that the current study provided
preliminary validation of the BCS-A. While the final measurement model
demonstrates acceptable fit by most conventions, in some instances it was only
marginal. The majority of indicators produced strong factor loadings in the final
model, although there were some indicators that resulted in lower loadings.
These lower loadings were likely because for some factors there is greater
conceptual distinction between indicators compared to other factors. This is
simply because there is greater variation in the way specific indicators
behaviourally manifest. However, the low prevalence of some behaviours in this
sample likely played a role (i.e., cyber perpetration, and physical victimization
and perpetration for females). In addition, there were also instances where the
suitability of the model was challenging to interpret, and so the effect of retaining
or removing specific items did not produce a clear result. Future studies are
needed to confirm these findings.
Finally, assessing concordance rates with existing measures of bullying were not
viable because they assessed bullying as a unidimensional construct. There are a
number of other additional measurement properties that could be evaluated in future
studies such as test-re-test reliability and responsiveness to change over time (Vessey
et al., 2014). Shorter follow-up periods could be used to establish test–retest
reliability, and intervention studies to test responsiveness to change. Related to these
issues is the validity of using global versus multi-scale measurement methods
specifically for estimating prevalence. While recent evidence suggests neither method
has demonstrated superior psychometric adequacy than the other (Renshaw,
Hammons, & Roberson, 2016), earlier studies preferenced global measurement
(Solberg & Olweus, 2003). Further investigations are needed to test which method is
most suitable for estimating prevalence.
Bullying and Cyberbullying Scale for Adolescents 17
Conclusion
Measurement is critically important to the replicability and robustness of research
findings, particularly regarding the rigorous evaluation of intervention programmes
(Thomas et al., 2015). In recent years, the construct of bullying has evolved to encompass
behaviour across traditional and cyber domains (Modecki et al., 2014), and there is a need
to ensure measurement tools capture the widened scope of bullying behaviour common
among young people. This study developed a brief 2 9 13-item self-report measure of
traditional and cyber bullying victimization and perpetration for use among adolescents.
The current findings provide preliminary evidence for the BCS-A as a measurement
strategy for making meaningful and comparable assessments of different forms of
traditional and cyber bullying behaviour among adolescents. This is both important for
extending the scope of hypotheses that can be tested, and for the evaluation and tailoring
of intervention programmes to target subscale profiles.
Acknowledgements
HJT is supported by the Bryan Foundation in partnership with ClearThinking Queensland, JGS
by a National Health and Medical Research Council (NHMRC) of Australia Practitioner
Fellowship Grant (1105807), JMC by an Australian Postgraduate Award scholarship, and JPC by
a NHMRC Career Development Fellowship (1031909).
References
Australian Bureau of Statistics (2011). Census of population and housing: Socio-Economic Indexes
for Areas (SEIFA), Australia. (ABS cat. no. 2033.0.55.001). Canberra, ACT: Commonwealth of
Australia.
Australian Bureau of Statistics (2016). Schools, Australia, 2016. Canberra, ACT: Commonwealth of
Australia.
Bandalos, D. L. (2014). Relative performance of categorical diagonally weighted least squares and
robust maximum likelihood estimation. Structural equation modeling. A Multidisciplinary
Journal, 21(1), 102–116. https://doi.org/10.1080/10705511.2014.859510
Bieniek, M. (2016). Comparison of the bias of trimmed and Winsorized means. Communications in
Statistics – Theory and Methods, 45, 6641–6650. https://doi.org/10.1080/03610926.2014.
963620
Bond, L., Butler, H., Thomas, L., Carlin, J., Glover, S., Bowes, G., & Patton, G. (2007). Social and
school connectedness in early secondary school as predictors of late teenage substance use,
mental health, and academic outcomes. Journal of Adolescent Health, 40(4), 357.e359-
357.e318. https://doi.org/10.1016/j.jadohealth.2006.10.013
Bovaird, J. A. (2010). Scales and surveys: Some problems with measuring bullying behavior. In S. R.
Jimerson, S. Swearer & D. L. Espelage (Eds.), Handbook of bullying in schools: An international
perspective (pp. 277–292). New York, NY: Routledge.
Breivik, K., & Olweus, D. (2014). An item response theory analysis of the Olweus bullying scale.
Aggressive Behavior, 41(1), 1–13. https://doi.org/10.1002/AB.21571
Brown, T. A., & Moore, M. T. (2012). Confirmatory factor analysis. In R. H. Hoyle (Ed.), Handbook of
structural equation modeling (pp. 361–379). New York, NY: Guilford.
Card, N. A., Stucky, B. D., Sawalani, G. M., & Little, T. D. (2008). Direct and indirect aggression during
childhood and adolescence: A meta-analytic review of gender differences, intercorrelations, and
relations to maladjustment. Child Development, 79, 1185–1229. https://doi.org/10.1111/j.
1467-8624.2008.01184.x
18 Hannah J. Thomas et al.
Cheng, Y.-Y., Chen, L.-M., Liu, K.-S., & Chen, Y.-L. (2011). Development and psychometric
evaluation of the school bullying scales: A Rasch measurement approach. Educational and
Psychological Measurement, 71(1), 200–216. https://doi.org/10.1177/0013164410387387
Corulla, W. J. (1990). A revised version of the Psychoticism scale for children. Personality and
Individual Differences, 11(1), 65–76. https://doi.org/10.1016/0191-8869(90)90169-R
Eysenck, H. J., & Eysenck, S. G. B. (1975). Manual of the eysenck personality questionnaire.
London, UK: Hodder and Stoughton.
Felix, E. D., Sharkey, J. D., Green, J. G., Furlong, M. J., & Tanigawa, D. (2011). Getting precise
and pragmatic about the assessment of bullying: The development of the California
Bullying Victimization Scale. Aggressive Behavior, 37(3), 234–247. https://doi.org/10.1002/
ab.20389
Gerbing, D. W., & Anderson, J. C. (1984). On the meaning of within-factor correlated measurement
errors. Journal of Consumer Research, 11, 572–580. https://doi.org/10.1086/208993
Gladden, R. M., Vivolo-Kantor, A. M., Hamburger, M. E., & Lumpkin, C. D. (2014). Bullying
surveillance among youths: Uniform definitions for public health and recommended data
elements, Version 1.0. Atlanta, GA: Centers for Disease Control and Prevention, Retrieved from
http://www.cdc.gov/violenceprevention/pdf/bullying-definitions-final-a.pdf
Goodman, R. (1997). The Strengths and Difficulties Questionnaire: A research note. Journal of
Child Psychology and Psychiatry, 38, 581–586. https://doi.org/10.1111/j.1469=7610.1997.
tb01545.x
Ho, R. (2006). Handbook of univariate and multivariate data analysis and interpretation with
SPSS. Boca Ratan, FL: Chapman & Hall. https://doi.org/10.1201/9781420011111
Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis:
Conventional criteria versus new alternatives. Structural Equation Modeling: A
Multidisciplinary Journal, 6(1), 1–55. https://doi.org/10.1080/10705519909540118
Hunt, C., Peters, L., & Rapee, R. M. (2012). Development of a measure of the experience of
being bullied in youth. Psychological Assessment, 24(1), 156–165. https://doi.org/10.1037/
a0025178
Kyriakides, L., Kaloyirou, C., & Lindsay, G. (2006). An analysis of the Revised Olweus Bully/Victim
Questionnaire using the Rasch measurement model. British Journal of Educational
Psychology, 76, 781–801. https://doi.org/10.1348/000709905X53499
Malecki, C. K., & Kilpatrick, M. D. (2002). Measuring perceived social support: Development of the
child and adolescent social support scale (CASSS). Psychology in the Schools, 39(1), 1–18.
https://doi.org/10.1002/pits.10004
McNeely, C. A., Nonnemaker, J. M., & Blum, R. W. (2002). Promoting school connectedness:
Evidence from the National Longitudinal Study of Adolescent Health. Journal of School Health,
72, 138–146. https://doi.org/10.1111/j.1746-1561.2002.tb06533.x
Mitsopoulou, E., & Giovazolias, T. (2015). Personality traits, empathy and bullying behavior: A meta-
analytic approach. Aggression and Violent Behavior, 21, 61–72. https://doi.org/10.1016/j.avb.
2015.01.007
Modecki, K. L., Minchin, J., Harbaugh, A. G., Guerra, N. G., & Runions, K. C. (2014). Bullying
prevalence across contexts: A meta-analysis measuring cyber and traditional bullying. Journal of
Adolescent Health, 55(5), 602–611. https://doi.org/10.1016/j.jadohealth.2014.06.007
Olweus, D. (1993). Bullying at school: What we know and what we can do. Oxford, UK: Blackwell.
Olweus, D. (1996). The revised Olweus bully/victim questionnaire. Bergen, Norway: University of
Bergen, Research Center for Health Promotion.
Renshaw, T. L., Hammons, K. N., & Roberson, A. J. (2016). General versus specific methods for
classifying US students’ bullying involvement: Investigating classification agreement, prevalence
rates, and concurrent validity. School Psychology Review, 45(4), 400–416. https://doi.org/10.
17105/SPR45-4.400-416
Rigby, K. (1998). Manual for the peer relations questionnaire. Point Lonsdale, Vic.: The
Professional Reading Guide for Educational Administrators.
Bullying and Cyberbullying Scale for Adolescents 19
Rosseel, Y. (2012). lavaan: An R package for structural equation modeling. Journal of Statistical
Software, 48(8), 1–36.
Salmivalli, C., & Peets, K. (2011). Bullies, victim, and bully-victim relationships in middle childhood
and early adolescence. In K. H. Rubin, W. M. Bukowski & B. P. Laursen (Eds.), Handbook of peer
interactions, relationships, and groups (pp. 322–340). New York, NY: Guilford Press.
Saris, W. E., Satorra, A., & van der Veld, W. M. (2009). Testing Structural Equation Models or
detection of misspecifications? Structural Equation Modeling: A Multidisciplinary Journal,
16, 561–582. https://doi.org/10.1080/10705510903203433
Satorra, A., & Bentler, P. M. (1994). Corrections to test statistics and standard errors in covariance
structure analysis. In A. von Eye & C. C. Clogg (Eds.), Latent variables analysis: Applications to
developmental research (pp. 339–419). Thousand Oaks, CA: Sage.
Shaw, T., Cross, D., Thomas, L. T., & Zubrick, S. R. (2014). Bias in student survey findings from active
parental consent procedures. British Educational Research Journal, 41, 229–243. https://doi.
org/10.1002/berj.3137
Shaw, T., Cross, D., & Zubrick, S. R. (2015). Testing for response shift bias in evaluations of school
antibullying programs. Evaluation Review, 39(6), 527–554. https://doi.org/10.1177/
0193841x16629863
Shaw, T., Dooley, J. J., Cross, D., Zubrick, S. R., & Waters, S. (2013). The Forms of Bullying Scale
(FBS): Validity and reliability estimates for a measure of bullying victimization and perpetration
in adolescence. Psychological Assessment, 25, 1045–1057. https://doi.org/10.1037/a0032955
Slee, P. T., & Rigby, K. (1993). The relationship of Eysenck’s personality factors and self-esteem to
bully-victim behaviour in Australian schoolboys. Personality and Individual Differences, 14(2),
371–373. https://doi.org/10.1016/0191-8869(93)90136-Q
Solberg, M. E., & Olweus, D. (2003). Prevalence estimation of school bullying with the Olweus Bully/
Victim Questionnaire. Aggressive Behavior, 29(3), 239–268. https://doi.org/10.1002/ab.10047
Tabachnick, B. G., & Fidell, L. S. (2013). Using multivariate statistics (6th ed.). Boston, MA: Pearson
Education.
Thomas, H. J., Chan, G. C., Scott, J. G., Connor, J. P., Kelly, A. B., & Williams, J. (2016). Association of
different forms of bullying victimisation with adolescents’ psychological distress and reduced
emotional wellbeing. Australian and New Zealand Journal of Psychiatry, 50(4), 371–379.
https://doi.org/10.1177/0004867415600076
Thomas, H. J., Connor, J. P., Lawrence, D. M., Hafekost, J. M., Zubrick, S. R., & Scott, J. G. (2017).
Prevalence and correlates of bullying victimisation and perpetration in a nationally
representative sample of Australian youth. Australian and New Zealand Journal of
Psychiatry, 51, 909–920. https://doi.org/10.1177/0004867417707819
Thomas, H. J., Connor, J. P., & Scott, J. G. (2015). Integrating traditional bullying and cyberbullying:
Challenges of definition and measurement in adolescents – a review. Educational Psychology
Review, 27(1), 135–152. https://doi.org/10.1007/s10648-014-9261-7
Vessey, J., Strout, T. D., DiFazio, R. L., & Walker, A. (2014). Measuring the youth bullying experience:
A systematic review of the psychometric properties of available instruments. Journal of School
Health, 84, 819–843. https://doi.org/10.1111/josh.12210
Vivolo-Kantor, A. M., Martell, B. N., Holland, K. M., & Westby, R. (2014). A systematic review and
content analysis of bullying and cyber-bullying measurement strategies. Aggression and Violent
Behavior, 19(4), 423–434. https://doi.org/10.1016/j.avb.2014.06.008
Wang, J., Iannotti, R. J., & Nansel, T. R. (2009). School bullying among adolescents in the United
States: Physical, verbal, relational, and cyber. Journal of Adolescent Health, 45(4), 368–375.
https://doi.org/10.1016/j.jadohealth.2009.03.021
Wolke, D., & Lereya, S. T. (2015). Long-term effects of bullying. Archives of Disease in Childhood,
100(9), 879–885. https://doi.org/10.1136/archdischild-2014-306667
Wolke, D., Waylen, A., Samara, M., Steer, C., Goodman, R., Ford, T., & Lamberts, K. (2009). Selective
drop-out in longitudinal studies and non-biased prediction of behaviour disorders. The British
Journal of Psychiatry, 195(3), 249. https://doi.org/10.1192/bjp.bp.108.053751
20 Hannah J. Thomas et al.
Worthington, R. L., & Whittaker, T. A. (2006). Scale development research: A content analysis and
recommendations for best practices. The Counseling Psychologist, 34, 806–838. https://doi.
org/10.1177/0011000006288127
Supporting Information
The following supporting information may be found in the online edition of the article:
Appendix S1. Original 40-item (2x20) BCS-A.
Appendix S2. Results of Exploratory Factor Analysis.
Appendix S3. Final BCS-A items, factor scores, correlations, reliabilities, and
prevalence estimate.
Appendix S4. Regression models testing concurrent validity of BCS-A.
Appendix S5. Regression models testing convergent validity for BCS-A-Victimization.
Appendix S6. Regression models testing convergent validity for BCS-A-Perpetration.
Appendix S7. Regression models testing predictive validity of BCS-A Victimization
and Perpetration at baseline and internalizing and externalizing problems at follow-up.
Appendix S8. Scoring Instructions for the BCS-A.
Appendix S9. Final 26-item (2x13) BCS-A – Ratio Scale.
Appendix S10. Final 26-item (2x13) BCS-A – Ordinal Scale.