validation of the perceived school bullying severity scale

15
This article was downloaded by: [University of Chicago Library] On: 19 November 2014, At: 18:53 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Educational Psychology: An International Journal of Experimental Educational Psychology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/cedp20 Validation of the perceived school bullying severity scale Li Ming Chen a , Kun Shia Liu a & Ying Yao Cheng a a Institute of Education, National Sun Yat-sen University , Kaohsiung , Taiwan Published online: 18 Nov 2011. To cite this article: Li Ming Chen , Kun Shia Liu & Ying Yao Cheng (2012) Validation of the perceived school bullying severity scale, Educational Psychology: An International Journal of Experimental Educational Psychology, 32:2, 169-182, DOI: 10.1080/01443410.2011.633495 To link to this article: http://dx.doi.org/10.1080/01443410.2011.633495 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms- and-conditions

Upload: ying-yao

Post on 24-Mar-2017

214 views

Category:

Documents


0 download

TRANSCRIPT

This article was downloaded by: [University of Chicago Library]On: 19 November 2014, At: 18:53Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Educational Psychology: AnInternational Journal of ExperimentalEducational PsychologyPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/cedp20

Validation of the perceived schoolbullying severity scaleLi Ming Chen a , Kun Shia Liu a & Ying Yao Cheng aa Institute of Education, National Sun Yat-sen University ,Kaohsiung , TaiwanPublished online: 18 Nov 2011.

To cite this article: Li Ming Chen , Kun Shia Liu & Ying Yao Cheng (2012) Validation of the perceivedschool bullying severity scale, Educational Psychology: An International Journal of ExperimentalEducational Psychology, 32:2, 169-182, DOI: 10.1080/01443410.2011.633495

To link to this article: http://dx.doi.org/10.1080/01443410.2011.633495

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Validation of the perceived school bullying severity scale

Li Ming Chen, Kun Shia Liu and Ying Yao Cheng*

Institute of Education, National Sun Yat-sen University, Kaohsiung, Taiwan

(Received 24 May 2011; final version received 17 October 2011)

Research on school bullying has tended to focus on its prevalence or frequencywhile ignoring its perceived severity. This study attempted to construct a per-ceived School Bullying Severity Scale (SBSS). The original 24-item instrument,revised from the Victim Scale of the School Bullying Scales, covered the fourcategories of physical, verbal, relational and cyber bullying. The partial creditmodel was used to conduct Rasch analysis with ConQuest software on dataderived from two samples of Taiwanese secondary school students. Sample 1and sample 2 consisted of 605 and 869 students, respectively. Three items weredeleted after examining the quality of the data from sample 1. The reliabilityand validity of the 21 items on the final scale were verified using data fromsample 2. Results demonstrated the reliability and validity of information col-lected by the SBSS. This study also found that secondary school students ratedrelational and cyber bullying as more severe than physical and verbal bullying.Differences between teachers’ and students’ perspectives on the perceived sever-ity of various bullying behaviours as well as implications for preventing andintervening in bullying are discussed.

Keywords: perceived severity; Rasch measurement; scale development; schoolbullying; victimisation

Introduction

Recently, school bullying has received international academic attention because ofits negative impact on the physical and mental health of students (Fleming & Jacob-sen, 2009; Hampel, Manhal, & Hayer, 2009; Hawker & Boulton, 2000; Totura,Green, Karver, & Gesten, 2009) and its effects on students’ perceptions of schoolsecurity (Gini, Pozzoli, Borghi, & Franzoni, 2008; Goldstein, Young, & Boyd,2008). Researchers around the world have been devoted to understanding the natureof bullying and trying to minimise its negative impact on students’ health (Bauman,Rigby, & Hoppa, 2008; Nabuzoka, Rønning, & Handegård, 2009; Novick & Isaacs,2010; Samara & Smith, 2008; Stavrinides, Georgiou, & Theofanous, 2010).

To date, most researchers have investigated the current situation in terms of theprevalence and frequency of bullying to provide data as a reference for preventionand intervention, including anti-bullying policies. The revised Olweus Bullying–Victim Questionnaire (OBVQ) (Olweus, 1996) is the most commonly used researchinstrument in this domain. The questionnaire includes two global items and severalspecific questions with five response options: ‘never’, ‘only once or twice’, ‘2–3times per month’, ‘once per week’ and ‘several times per week’. Students who

*Corresponding author. Email: [email protected]

Educational PsychologyVol. 32, No. 2, March 2012, 169–182

ISSN 0144-3410 print/ISSN 1469-5820 online� 2012 Taylor & Francishttp://dx.doi.org/10.1080/01443410.2011.633495http://www.tandfonline.com

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

endorse 2–3 times per month or more on two global items are regarded as self-reported victims or bullies (Solberg & Olweus, 2003). Many researchers haveadopted Olweus’s definition and global question in investigations of the prevalenceof school bullying (Black & Jackson, 2007; Sapouna, 2008; Totura et al., 2009;Vaillancourt et al., 2008). The survey conducted by the World Health Organization(WHO) on Health Behaviour in School-aged Children (HBSC) in 39 countriesfound that 15, 14 and 10% of students aged 11, 13 and 15, respectively, were vic-tims of bullying and that 9, 12 and 12% of students aged 11, 13 and 15, respec-tively, had engaged in bullying behaviour (Currie et al., 2008).

Other researchers have investigated the prevalence of bullying with the specificquestions from the OBVQ, and students who endorsed 2–3 times per month on atleast one of the items were regarded as self-reported victims or bullies (Scheithauer,Hayer, Petermann, & Jugert, 2006; Theriot, Dulmus, Sowers, & Johnson, 2005).Alternatively, other researchers have added the scores, with higher total scores indi-cating more frequent or more serious experiences of bullying or being bullied(Georgiou, 2008; Georgiou & Stavrinides, 2008; Smith & Gross, 2006). Severalresearchers have also been concerned about the relative frequency of bullyingbehaviours or categories of bullying (Bradshaw, Sawyer, & O’Brennan, 2007;Cheng, Chen, Liu, & Chen, 2011).

Current bullying research has focused on frequency, which serves as the inter-cept point not only for estimating the prevalence of bullying (Solberg & Olweus,2003) but also for identifying the most widespread forms of bullying in attempts toset priorities for prevention and intervention.

However, the frequency of bullying and being bullied cannot be equated withits severity; students do not necessarily consider the most frequent bullying behav-iours to be more serious and the less frequent bullying behaviours to be less seri-ous. Delfabbro et al. (2006) found that physical bullying occurred much lessfrequently than did other types of bullying, but Maunder, Harrop, and Tattersall(2010) found that students considered physical bullying to be more serious thanother types. These findings have revealed the frequency of bullying is not propor-tional to the perceptions of its severity.

Thus far, although researchers have studied the perceived severity of differentbullying behaviours perpetrated in school settings, most studies have focused onteachers’ perspectives. Teachers have considered physical bullying to be more seri-ous than verbal or relational bullying (Bauman & Del Rio, 2006; Ellis & Shute,2007; Hazler, Miller, Carney, & Green, 2001; Mishna, 2004; Yoon & Kerber,2003). Importantly, judgements made by teachers about the severity of bullyinghave determined whether they will intervene in situations (Mishna, Scarcello,Pepler, & Wiener, 2005). Currently, few studies have been conducted from the stu-dents’ perspective. Maunder and colleagues (2010) included student viewpoints inthe assessment of the seriousness of school bullying but did not provide evidenceof the validity of the scale used. We need to develop a reliable and valid scale toprovide more information about the seriousness of bullying among secondaryschool students as well as about how to prevent these behaviours.

The definition and categories of school bullying

We adopted Olweus’ (1993) definition of bullying in this study: ‘A student is beingbullied or victimised when he is exposed repeatedly and over time to negative

170 L.M. Chen et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

actions on the part of one or more other students’ (p. 9). Olweus also identifiedthree prominent characteristics of bullying behaviour: negative actions, repetitionand power imbalance.

Researchers have used different categories for school bullying. The OBVQincludes the following seven types of bullying:

being bullied verbally, being excluded from a group or completely ignored, being bul-lied physically, having false rumours spread, having money and other things takenaway or damaged, being threatened or forced to do things, and being bullied aboutone’s race or colour. (Solberg & Olwues, 2003, p. 247)

Other researchers have divided bullying behaviours into six categories: physical,verbal, relational, property, behavioural bullying and coercive behaviour (Cheng,Chen, Ho, & Cheng, 2011). Thus, categories of bullying differ and often overlapwith one another. Moreover, the impact of the recent emergence of cyber bullyingshould not be underestimated (Shariff, 2009). The present study adopted the classifi-cation developed by Berger (2007) on the basis of her review of 1049 papers in thePsycINFO database: physical bullying (e.g. pushing, kicking, damaging others’homework and damaging others’ property), relational bullying (e.g. manipulatinginterpersonal relationships), verbal bullying (e.g. name-calling) and cyber bullying(e.g. spreading rumours online and sending malicious text messages).

Perceived severity of school bullying

The majority of studies on the seriousness of school bullying have been conductedfrom the teachers’ perspectives. It was found that teachers ranked physical bullyingas the most severe, verbal bullying as the second, and relational bullying as thethird (Ellis & Shute, 2007; Mishna et al., 2005; Yoon & Kerber, 2003). Becauserelational bullying has appeared to be a minor compared with physical and verbalbullying, teachers have tended to let students settle relational issues by themselves.School counsellors have expressed similar opinions about the seriousness of bully-ing, considering physical bullying as more serious than verbal and relational bully-ing (Hazler et al., 2001; Jacobsen & Bauman, 2007).

The involvement of teachers in efforts to stop bullying depends on their judge-ment about the seriousness of different bullying behaviours. Teachers have showngreater sympathy to students who suffered from verbal and physical bullying, andthey have been five times as likely to intervene in instances of verbal and physicalbullying than in those of relational bullying (Yoon & Kerber, 2003). Ellis and Shute(2007) reported that the problem-solving and rules-sanctions approaches adopted byteachers to address school bullying were affected by the perceived severity of thebullying; bullying perceived as more serious was more likely to elicit the simulta-neous use of the two approaches, showing that the likelihood of teacher involve-ment varied as a function of the perceived seriousness of the bullying.

Maunder and colleagues (2010) have explored the perceived seriousness of bul-lying from the perspectives of secondary teachers, staff as well as students. Theyfound that indirect bullying behaviours (e.g. socially excluding someone) wereregarded as less serious than direct bullying behaviours (e.g. threatening or hitting/kicking/punching). These perceptions were held across teachers, staffs and students.Although the research conducted by Maunder and his colleagues was one of thefew studies investigating the severity of school bullying behaviours from multiple

Educational Psychology 171

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

perspectives, they did not provide adequate evidence supporting the validity of theirinstrument.

Purpose of this study

The purpose of this study was to construct a self-report scale focused on the per-ceived severity of school bullying among secondary-school students and to furtherverify its reliability and validity. We also examined the perceived severity of differ-ent bullying behaviours among secondary-school students to determine the consis-tency of these judgements with the perspectives of teachers reported in the relevantliterature.

Method

Participants

This study included two samples. Sample 1 was used to revise the questionnaireitems, and sample 2 was used to confirm the quality of the items. The samples con-sisted of students from southern Taiwan. Sample 1 included students from twojunior high schools (grades 7–9), two senior high schools (grades 10–12), and twovocational high schools (grades 10–12). One class was randomly selected from eachgrade in each school, yielding a sample comprising three classes from each school.We distributed 750 questionnaires, and a total 605 valid surveys were returned, rep-resenting a response rate of 80.6%. The sample consisted of 255 male (42.3%) and348 female (57.7%) students; these included 186 junior high school students(30.7%), 205 high school students (33.9%) and 214 vocational students (35.4%).The ages ranged from 12 to 20 years old (M= 15.79, SD = 1.57); 519 were of Tai-wanese descent (90.6%), 9 were of aboriginal descent (1.6%), 7 were the childrenof foreign-born parents (1.2%) and 38 were of others (6.6%).

Sample 2 included two junior high schools, three high schools and four voca-tional schools; once again, a class was randomly selected from each grade in eachschool. We distributed 1155 questionnaires, and 869 valid surveys were returned,representing a response rate of 75.2%. The sample included 441 male (51.1%) and422 female (48.9%) students; of these, 172 were junior high school students(19.8%), 339 senior high school students (39.0%) and 358 vocational high schoolstudents (41.2%). The ages of the sample ranged from 12 to 20 years old(M= 16.13, SD= 1.42); 766 were of Taiwanese descent (95.4%), 13 were of aborigi-nal descent (1.6%), 4 were children of foreign-born parents (.5%) and 20 were ofothers (2.5%).

Instrument

This study revised the Victim Scale of the School Bullying Scales (Cheng, Chen,Liu et al., 2011), which used a 5-point Likert scale to rate the frequency of differentbehaviours. We selected this scale because it uses Olweus’ definition of school bul-lying, and specified the four characteristics of bullying: negative behaviours, repeti-tion, power imbalance and intentionality. Second, the scale includes 14 itemsaddressing the most common physical, verbal and relational bullying types identi-fied by Kyriakides, Kaloyirou, and Lindsay (2006), as well as the newer cyber bul-lying behaviours. Third, the scale was validated with Rasch analysis, and the resultsshowed good evidence of reliability and validity.

172 L.M. Chen et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

On the other hand, this scale has fewer items about relational and cyber bulling,and it assesses only the frequency of being bullied without providing informationabout the perceived severity of bullying. Thus, this study revised the original scaleby adding 10 items to address a greater number of the bullying behaviourscommonly observed in schools; eight, five, six and five items addressed physical,verbal, relational and cyber bullying, respectively, yielding a total of 24 items.Second, we revised the response options to measure the severity rather than thefrequency of behaviours. We used a 5-point Likert scale from 1 (less serious) to 5(more serious).

Procedures

First, we randomly selected administrative regions of Kaohsiung City in Taiwan.Second, we randomly selected schools in the selected administrative regions andtelephoned the principals of these schools for permission to conduct the study.Third, following these schools’ standard agreement policies, we invited each princi-pal to convene a meeting with school staff and teachers to decide whether or not toparticipate in this study. After obtaining consent from these schools, we then mailedthe questionnaires, survey instructions, a gift (a pencil) for each student and astamped return envelope to a designated agent at each school. The survey was com-pleted by each class as a unit. Several statements were specified on the front pageof the questionnaire, regarding the purpose of research, participants’ rights and con-tact information for inquiries about research questions. The principles of anonymityand confidentiality were explained prior to the survey to ensure that the interests ofstudents, teachers and schools would not be compromised. Students were free toparticipate or withdraw from the survey. The agent at each school collected thequestionnaires and returned them by mail.

Data analysis

This study used the Rasch technique for data analysis to enable ordinal scales to beconverted into interval scales in logit units. In this way, personal traits andresponses to items were expressed by logit units along one continuum (Bond &Fox, 2007), which enabled comparisons of the perceived severity of different bully-ing behaviours on the same scale. Second, Rasch measurements include indices thatreflect model–data fit to examine whether each item fits its potential construct andserve as a reference for future revisions of the items.

This study used the partial credit model (PCM) (Masters, 1982), which enablesanalysis of data collected via Likert scales while permitting each item to have a dif-ferent threshold and structure. This diversity is closer to actual situations in whichparticipants may have different interpretations of options. We used ConQuest soft-ware (Wu, Adams, & Wilson, 2007) to estimate the parameters in the PCM modelwith the marginal maximum likelihood method.

The data were analysed in two stages. First, data from sample 1 were used toexamine the quality of items. Statistics reflecting fit served as indices to enable theidentification of problematic items. Infit mean square (MNSQ) statistics wasadopted because it was more sensitive to unexpected behaviour affecting responsesto items near the person’s measure level and have greater threat to validity (Linacre,2006). An item with infit MNSQ above 1.4 or below .6 (Wright & Linacre, 1994)

Educational Psychology 173

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

would be deleted as it had 40% more or less variation in the observed data than theRasch model predicted. Additionally, Rasch–residual-based principal-componentsanalysis (PCAR) was executed by Winsteps software (Linacre, 2006) to examinethe dimensional structure of the SBSS. Differential item-functioning (DIF) analysesby gender were performed and served as references for item revision or deletion. Adifference of .5 logits in the overall difficulty of items across groups was regardedas a sign of substantial DIF (Bond & Fox, 2007; Wang, 2008). The data collectedfrom sample 2 were used to validate the quality of the items.

A variety of evidence was used to assess the validity (Wolfe & Smith, 2007) ofthe SBSS. In terms of content, we examined item–measure correlations as well asthe fit indices of items. In terms of structure, PCAR was employed to check unidi-mensionality (i.e. whether the latent trait measured by the scale belonged to a singleconstruct). In terms of generalisability, DIF was analysed by sex to determinewhether the scale could be applied to different groups. Person separation reliabilitywas calculated as a group estimate of reliability (Schumacker & Smith, 2007). Interms of substantive concerns, we examined whether the rankings of the perceivedseverity of bullying behaviours were identical in both samples. In terms of interpret-ability, we examined the person–item map to understand the distribution of personmeasures and item estimates.

Results

Item revision

The data obtained from sample 1 showed that most items had acceptable infitMNSQ, except item 7 (being cursed at: infit MNSQ=1.53) and item 14 (beingcalled an unpleasant nickname: infit MNSQ= 1.45). We found the discriminationindices of these two items were lower than those of the other items, which reflectedmore noise in the data (Wu & Adams, 2007). Accordingly, items 7 and 14 weredeleted to improve the data’s fit with the model. In addition, DIF analyses wereconducted by gender. We found no statistically significant differences betweengenders.

Additionally, Winsteps software was employed to execute PCAR. Linacre(2006) suggested that the variance explained by items should exceed 60% andthat the unexplained variance explained by the first contrast should be lower thanthree. Results showed that 90.9% of the variance was explained by the items,indicating that the dominant dimension was able to explain 90.9% of the variancein the data. The unexplained variance explained by the first contrast was 3.5, indi-cating that the first residual contrast had an eigenvalue of 3.5, which was some-what greater than the strength of three items, the fewest that could be consideredto constitute a dimension. The Q3 statistic (Yen, 1984) indicated that item 19(being isolated) had high residual correlations with item 21 (not giving a damnabout me; r= .63) and item 15 (being ostracised from a group; r= .51). Item 19was deleted to remedy the problem of the high residual correlation amongrelational bullying items.

After elimination of the aforementioned three items, 21 items including eightaddressing physical bullying, three addressing verbal bullying, five addressingrelational bullying, and five addressing cyber bullying were administered to sample2 in the validation study.

174 L.M. Chen et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

Item validation

Content evidence

PCM was employed to examine the model–data fit in sample 2. Table 1 summarisesthe MNSQ indices for the items. It is clear that all items exhibited good fit with theunidimensional PCM model (infit MNSQ ranged from .82 to 1.32). Item–measurecorrelation was used as the other index of the technical quality of items, and theresults showed that scores on each item exhibited a high degree of correlation withthe total scores on the SBSS (correlation ranged from .67 to .86).

Structure evidence

PCAR was utilised to examine the unidimensionality of the measures. The resultsshowed that a substantial amount of the variance was explained by the items(87.9%) and that the unexplained variance explained by the first residual contrastwas 2.8. Both values met Linacre’s (2006) standard. This result supported the unidi-mensionality of SBSS, and we found no shared secondary dimension among theitems.

Generalisability evidence

DIF analyses were conducted for gender. No substantial DIF was found betweenboys and girls. The differences between boys and girls in the overall estimates onall items ranged from .01 to .27 logit. We used person separation reliability as theother index of generalisability. Result showed the person separation reliability ofthe measures from the SBSS was .97.

Table 1. Summary of item quality indices.

Item Estimate Error Infit MNSQ Item-total correlation

1 .69 .03 1.29 .682 .06 .03 .94 .823 .65 .03 1.29 .674 �.05 .03 .97 .815 �.03 .03 1.06 .796 .21 .03 1.08 .767 .02 .03 .94 .818 .39 .03 1.32 .689 �.27 .03 .96 .8210 .12 .03 .89 .8211 �.18 .03 .95 .8112 �.32 .03 1.24 .7513 �.32 .03 .82 .8614 �.64 .03 .99 .8215 �.36 .03 1.07 .8016 .15 .03 .92 .8117 �.16 .03 1.05 .8018 �.06 .03 .99 .8019 .13 .03 .98 .7920 .21 .03 .90 .8121 �.24 .12 .86 .85

Note: Item estimates are in logit units.

Educational Psychology 175

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

Substantive evidenceTo examine the item-calibration invariance across different samples, we assessedthe Pearson correlation of item estimates obtained from samples 1 and 2. The corre-lation between the two measures was .98, which supports the stability of the itemhierarchy.

Interpretability evidence

Figure 1 depicts the person–item map of the SBSS. The left side shows the personmeasures of the perceived severity of bullying behaviours. More positive valuesindicate a greater likelihood of perceiving all kinds of aggressive behaviour assevere bullying. More negative values indicate that students rarely viewed variousaggressive behaviours as severe bullying. The right side of the person–item mapshows the perceived severity hierarchy of bullying behaviours. Increasingly positivevalues for items denote the items were less likely to be rated as severe bullying.Increasingly negative values for items denote the items were more likely to be ratedas severe bullying.

Rasch measurement aligns person measures and item estimates on the samescale with logit units. When a person measure was equal to a specific item estimate,the person had an approximately 50% chance of perceiving the bullying behaviouras severe. When a person measure was above the item estimate, this chance wasgreater than 50%. For example, when a person measure was above the estimate ofitem 20 (being shoved), the respondent’s likelihood of perceiving this behaviour assevere bullying was higher than 50%.

Of all items, item 1 (being tripped, .69 logit), item 3 (having one’s belongingshidden, .65 logit), item 8 (being forced to do something, .39 logit), item 6 (havingsomething thrown at you, .21 logit) and item 20 (being shoved, .21 logit) wererarely rated as severe bullying. These behaviours may be viewed as playful teasingand thus not perceived as severe bullying. Item 14 (one’s friendship being ruined,�.64 logit), item 15 (being hit and kicked, �.36 logit), item 12 (belongings takenwithout permission, �.32 logit), item 13 (being ostracised, �.32 logit) and item 9(having humiliating photos posted online, �.27 logit) were perceived as the mostsevere bullying behaviours.

Figure 2 displays item calibrations and person measures on the x-axis and itemcategorisations on the y-axis. The logit scale, person measures and item calibrationsdisplayed on x-axis are identical to those in Figure 1. On the y-axis, item categories

Figure 1. Person–item map of the SBSS.Note: Each ‘X’ on the left denotes a group of persons. Right side presents the item number,behavioural description and its logit units.

176 L.M. Chen et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

are arranged from top to bottom according to increasing severity. This sequence,which was derived from the teachers’ perspectives reported in the relevant literature,was treated as the expected pattern of perceived severity of bullying.

The average estimates of physical, verbal, relational and cyber bullying were.16, .02, �.21 and �.07 logit, respectively. Results showed an inconsistencybetween expected and empirical estimates of item categories. Relational and cyberbullying were rated as more severe than verbal and physical bullying, which isinconsistent with the teachers’ views of physical bullying as more severe than ver-bal and relational bullying.

It is worth noting that physical bullying showed wide range across the logitscale of perceived severity. Some physical bullying behaviours such as beingtripped or having one’s belongings hidden may be viewed as playful teasing andthus not be rated as severe bullying. ‘Being forced to do something’, which canhelp victims show conformity and prevent them from severe injury, was also per-ceived as not severe. On the other hand, ‘being hit and kicked’ was viewed as moresevere than most relational and cyber bullying behaviours. In addition, physical typeof bullying may not more serious than verbal and relational type of bullying from

Figure 2. Comparison among different categories of bullying.Note: Each ‘X’ denotes a group of persons. The location of ‘X’ denotes the logit units ofperson measures. Item calibrations on the left side (positive values) denote lower severity ofbullying behaviour. Item calibrations on the right side (negative values) denote higherseverity of bullying behaviour.

Educational Psychology 177

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

students’ point of view, but some of the physical bullying behaviours are more seri-ous than other behaviours. The present study suggested that how bullying categoriesdistributed along the logit scale might depend on the nature of bullying behaviours.The rank orderings of specific bullying categories would be affected by bullyingbehaviours examined in a study.

Discussion

Most studies have investigated the prevalence and frequency of school bullying, butfew researchers have focused specifically on the perceived severity of school bully-ing. Academic and practical considerations render examination of the perceivedseverity of school bullying by secondary school students crucial. This study aimedto develop the SBSS and used Rasch techniques to validate its psychometricproperties.

We provided several kinds of evidence to support the validity of the SBSS. Interms of content, the data collected with the SBSS fit the Rasch PCM model well.Additionally, all items exhibited high item–measure correlations (all above .67),which supported the technical quality of items. In terms of structure, the results ofthe PCAR indicated no shared secondary dimension among items, which supportedthe unidimensionality of the SBSS. With respect to generalisability, the person-sepa-ration reliability coefficient of the SBSS was satisfactory. No substantial DIF wasfound between genders, indicating that this instrument can be applied across mea-surement groups. As to substance, we validated the hierarchy of item estimates insamples 1 and 2, and extremely high Pearson correlation coefficients for item esti-mates supported the hierarchical stability of these estimates. In terms of interpret-ability, we presented the person–item map to depict the rankings of the perceivedseverity of bullying behaviours and the distribution of person measures on the samescale as logit units. All this evidence supports the validity of the SBSS.

Results of the present study revealed that the top five severe bullying behavioursrated by secondary-school students were ‘one’s friendship being ruined’, ‘being hitand kicked’, ‘belongings taken without permission’, ‘being ostracised’ and ‘havinghumiliating photos posted online’. In contrast, ‘being tripped’, ‘having one’sbelongings hidden’, ‘being forced to do something’, ‘having something thrown atyou’ and ‘being shoved’ were perceived as the top five least severe behaviours bystudents. These results were inconsistent with those reported by Maunder and col-leagues (2010), who found that secondary students from England rated threats asthe most serious bullying behaviour and social exclusion as the least serious bully-ing behaviour. This discrepancy may have resulted from cultural differences. InAsia, students immersed in a collectivist culture may place greater value on socialinteraction and group harmony, which may have contributed to the greater per-ceived severity of relational bullying, such as one’s friendship being ruined. InJapan, bullying (ijime) is a collective behaviour involving ostracism from peergroups and verbal abuse, and perpetrators bully anyone who deviates from thenorms and rules of their community (Akiba, 2004). In England, students living inindividualistic culture may not attend as closely as Asian students to social relation-ships, which may have contributed to the lower perceived severity of relational bul-lying. The connection between the perceived severity of bullying and culturaldifferences should be further investigated in the future. In addition, the severity ofcyber bullying behaviours were rated as moderate among various bullying

178 L.M. Chen et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

behaviours, both in this study and in Maunder et al. (2010) study (denoted asEmail/text messages).

In terms of bullying categories, secondary students rated the severity of rela-tional to be greater than that of physical and verbal bullying. This finding providedaccumulating evidence to the previous studies which indicate that relational bullyingis more harmful than physical or verbal bullying as it might bring victims moredepressive symptoms and emotional distress (Baldry, 2004; Bauman, 2008).

In the context of research showing that teachers considered the severity of phys-ical bullying to be greater than that of verbal and relational bullying (Ellis & Shute,2007; Hazler et al., 2001; Mishna, 2004; Yoon & Kerber, 2003), our results indi-cated discrepancies between teachers’ and students’ perspectives on the severity ofbullying behaviours. Furthermore, teachers tended to respond to and intervene inphysical bullying but were less likely to intervene in presumably less direct formsof bullying (Menesini, Fonzi, & Smith, 2002; Mishna et al., 2005; Yoon & Kerber,2003). Indeed, while dealing with physical bulling, teachers may ignore other typesof bullying such as relational and cyber bullying that are perceived as more seriousby students. This finding suggests that teachers should be more aware of the pre-vention and intervention of relational and cyber bullying.

There are some limitations to the present study. First, the differences betweenteachers’ and students’ perspectives on the severity of bullying behaviours may bedue to cultural differences as well as respondent differences. Future studies canexamine across countries to clarify whether cultural differences exist among second-ary students’ perceived severity of bullying. One need is to test for hierarchicalinvariance of the ordering of perceived severity in different countries or culturalcontexts. Follow-up studies can also compare the perceptions of secondary-schoolstudents and teachers on severity of school bullying, and explore whether respon-dent differences exist between these two perspectives. Second, the perceived sever-ity of a school bullying incident may be determined by contexts and students’personalities. This study used a well-validated instrument to show the current statusof school bullying severity evaluated by Taiwanese secondary students. Researchcan further explore to understand what contextual factors or personality traits mayinfluence students’ perception of bullying severity. Third, as cyber bullying hasbeen developing rapidly in recent years, it is highly possible that new forms ofcyber bullying will appear among secondary school students. Future studies can addnew items regarding cyber bullying behaviours to examine its perceived severity.

In summary, this study contributes to the academic literature by developing andvalidating the SBSS, which provides researchers with a well-validated instrument forcollecting data regarding the perceived severity of school bullying. Second, by identi-fying discrepancies between teachers’ and students’ perspectives on the severity ofschool bullying, this study contributed to prevention and intervention efforts withrespect to bullying. Researchers and educators should develop prevention and inter-vention programmes directed at school bullying based on the perceived severity ratherthan only on the prevalence and frequency of bullying behaviours. More specifically,relational and cyber bullying among students deserves much greater attention.

References

Akiba, M. (2004). Nature and correlates of Ijime: Bullying in Japanese middle school. Inter-national Journal of Educational Research, 41, 216–236. doi: 10.1016/j.ijer.2005.07.002.

Educational Psychology 179

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

Baldry, A.C. (2004). The impact of direct and indirect bullying on the mental and physicalhealth of Italian youngsters. Aggressive Behavior, 30, 343–355.

Bauman, S. (2008). The association between gender, age, and acculturation, and depressionand overt and relational victimization among Mexican American elementary students.Journal of Early Adolescence, 28, 528–554.

Bauman, S., & Del Rio, A. (2006). Pre-service teachers’ response to bullying scenarios:Comparing physical, verbal, and relational bullying. Journal of Educational Psychology,98, 219–231.

Bauman, S., Rigby, K., & Hoppa, K. (2008). US teachers’ and school counsellors’ strategiesfor handling school bullying incidents. Educational Psychology, 28, 837–856.

Berger, K. (2007). Update on bullying at school: Science forgotten? Developmental Review,27, 90–126.

Black, S.A., & Jackson, E. (2007). Using bullying incident density to evaluate the OlweusBullying Prevention Programme. School Psychology International, 28, 623–638.

Bond, T.G., & Fox, C.M. (2007). Applying the Rasch model: Fundamental measurement inthe human sciences. Mahwah, NJ: Erlbaum.

Bradshaw, C.P., Sawyer, A.L., & O’Brennan, L.M. (2007). Bullying and peer victimizationat school: Perceptual differences between students and school staff. School PsychologyReview, 36, 361–382.

Cheng, Y.Y., Chen, L.M., Ho, H.C., & Cheng, C.L. (2011). Definitions of school bullying inTaiwan: A comparison of multiple perspectives. School Psychology International, 32,227–243. doi: 10.1177/0143034311404130.

Cheng, Y.Y., Chen, L.M., Liu, K.S., & Chen, Y.L. (2011). Development and psychometricevaluation of the school bullying scales: A Rasch measurement approach. Educationaland Psychological Measurement, 71, 200–216.

Currie, C., Gabhainn, S.N., Godeau, E., Roberts, C., Smith, R., Currie, D., . . . Barnekow, V.(Eds.). (2008). Inequalities in young people’s health: HBSC international report from the2005/2006 survey. Copenhagen: WHO Regional Office for Europe.

Delfabbro, P., Winefield, T., Trainor, S., Dollard, M., Anderson, S., Metzer, J., & Hammar-strom, A. (2006). Peer and teacher bullying/victimization of South Australian secondaryschool students: Prevalence and psychosocial profiles. British Journal of EducationalPsychology, 76, 71–90.

Ellis, A.A., & Shute, R. (2007). Teacher responses to bullying in relation to moral orientationand seriousness of bullying. British Journal of Educational Psychology, 77, 649–663.

Fleming, L.C., & Jacobsen, K.H. (2009). Bullying and symptoms of depression in Chileanmiddle school students. Journal of School Health, 79, 130–137.

Georgiou, S.N. (2008). Bullying and victimization at school: The role of mothers. BritishJournal of Educational Psychology, 78, 109–125.

Georgiou, S.N., & Stavrinides, P. (2008). Bullies, victims, and bully-victims: Psychosocialprofiles and attribution styles. School Psychology International, 29, 574–589.

Gini, G., Pozzoli, T., Borghi, F., & Franzoni, L. (2008). The role of bystanders in stu-dents’ perception of bullying and sense of safety. Journal of School Psychology, 46,617–638.

Goldstein, S.E., Young, A., & Boyd, C. (2008). Relational aggression at school: Associationswith school safety and social climate. Journal of Youth and Adolescence, 37, 641–654.

Hampel, P., Manhal, S., & Hayer, T. (2009). Direct and relational bullying among childrenand adolescents: Coping and psychological adjustment. School Psychology International,30, 474–490.

Hawker, D.S.J., & Boulton, M.J. (2000). Twenty years’ research on peer victimization andpsychosocial maladjustment: A meta-analytic review of cross-sectional studies. Journalof Child Psychology and Psychiatry and Allied Disciplines, 41, 441–455.

Hazler, R.J., Miller, D.L., Carney, J.V., & Green, S. (2001). Adult recognition of schoolbullying situations. Educational Research, 43, 133–146.

Jacobsen, K.E., & Bauman, S. (2007). Bullying in schools: School counselors’ responses tothree types of bullying incidents. Professional School Counseling, 11, 1–9.

Kyriakides, L., Kaloyirou, C., & Lindsay, G. (2006). An analysis of the revised OlweusBully/Victim Questionnaire using the Rasch measurement model. British Journal ofEducational Psychology, 76, 781–801.

180 L.M. Chen et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

Linacre, J.M. (2006). A user’s guide to WINSTEPS Rasch model computer program.Chicago, IL: Winsteps.

Masters, G.N. (1982). A Rasch model for partial credit scoring. Psychometrika, 47,149–174.

Maunder, R., Harrop, A., & Tattersall, A. (2010). Pupil and staff perceptions of bullying insecondary schools: Comparing behavioural definitions and their perceived seriousness.Educational Research, 52, 263–282.

Menesini, E., Fonzi, A., & Smith, P.K. (2002). Attribution of meanings to terms related tobullying: A comparison between teacher’s and pupil’s perspectives in Italy. EuropeanJournal of Psychology of Education, 17, 393–406.

Mishna, F. (2004). A qualitative study of bullying from multiple perspectives. Children andSchools, 26, 234–247.

Mishna, F., Scarcello, I., Pepler, D., & Wiener, J. (2005). Teachers’ understanding ofbullying. Canadian Journal of Education, 28, 718–738.

Nabuzoka, D., Rønning, J.A., & Handegård, B.H. (2009). Exposure to bullying, reactionsand psychological adjustment of secondary school students. Educational Psychology, 29,849–866.

Novick, R.M., & Isaacs, J. (2010). Telling is compelling: The impact of student reports ofbullying on teacher intervention. Educational Psychology, 30, 283–296.

Olweus, D. (1993). Bullying at school: What we know and what we can do. Oxford: Black-well.

Olweus, D. (1996). The revised Olweus Bully/Victim Questionnaire. Bergen: Research Centerfor Health Promotion.

Samara, M., & Smith, P.K. (2008). How schools tackle bullying, and the use of wholeschool policies: Changes over the last decade. Educational Psychology, 28, 663–676.

Sapouna, M. (2008). Bullying in Greek primary and secondary schools. School PsychologyInternational, 29, 199–213.

Scheithauer, H., Hayer, T., Petermann, F., & Jugert, G. (2006). Physical, verbal, andrelational forms of bullying among German students: Age trends, gender differences, andcorrelates. Aggressive Behavior, 32, 261–275.

Schumacker, R.E., & Smith, E.V. Jr. (2007). Reliability: A Rasch perspective. Educationaland Psychological Measurement, 67, 394–409.

Shariff, S. (2009). Confronting cyber-bullying: What schools need to know to control mis-conduct and avoid legal consequences. New York, NY: Cambridge University Press.

Smith, G.R., & Gross, A.M. (2006). Bullying: Prevalence and the effect of age and gender.Child & Family Behavior Therapy, 28, 13–37.

Solberg, M., & Olweus, D. (2003). Prevalence estimation of school bullying with theOlweus Bully/Victim Questionnaire. Aggressive Behavior, 29, 239–268.

Stavrinides, P., Georgiou, S., & Theofanous, V. (2010). Bullying and empathy: A short termlongitudinal investigation. Educational Psychology, 30, 793–802.

Theriot, M.T., Dulmus, C.N., Sowers, K.S., & Johnson, T.K. (2005). Factors relating toself-identification among bullying victims. Children and Youth Services Review, 27,979–994.

Totura, C.M.W., Green, A.E., Karver, M.S., & Gesten, E.L. (2009). Multiple informants inthe assessment of psychological, behavioral, and academic correlates of bullying and vic-timization in middle school. Journal of Adolescence, 32, 193–211.

Vaillancourt, T., McDougall, P., Krygsman, A., Hymel, S., Miller, J., Stiver, K., & Davis, C.(2008). Bullying: Are researchers and children/youth talking about the same thing?International Journal of Behavioral Development, 32, 486–495.

Wang, W.C. (2008). Assessment of differential item functioning. Journal of AppliedMeasurement, 9, 387–408.

Wolfe, E.W., & Smith, E.V. Jr. (2007). Instrument development tools and activities formeasure validation using Rasch models: Part II-Validation activities. Journal of AppliedMeasurement, 8, 204–233.

Wright, B.D., & Linacre, J.M. (1994). Chi-square fit statistics. Rasch Measurement Transac-tions, 8, 370.

Wu, M., & Adams, R. (2007). Applying the Rasch model to psycho-social measurement: Apractical approach. Melbourne: Educational Measurement Solutions.

Educational Psychology 181

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14

Wu, M.L., Adams, R.J., & Wilson, M.R. (2007). ConQuest [Computer software andmanual]. Camberwell: Australian Council for Educational Research.

Yen, W.M. (1984). Effects of local item dependence on the fit and equating performance ofthe three-parameter logistic model. Applied Psychological Measurement, 8, 125–145.

Yoon, J.S., & Kerber, K. (2003). Bullying: Elementary teachers’ attitudes and interventionstrategies. Research in Education, 69, 27–35.

182 L.M. Chen et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

8:53

19

Nov

embe

r 20

14