Children and Youth Services Review 35 (2013) 2091–2100

Contents lists available at ScienceDirect

Children and Youth Services Review

j ourna l homepage: www.e lsev ie r .com/ locate /ch i ldyouth

Agreement among students', teachers', and parents' perceptions ofvictimization by bullying

Michelle K. Demaray ⁎, Christine K. Malecki, Stephanie M. Secord, Kelly M. LyellPsychology Department, Northern Illinois University, DeKalb, IL 60115, USA

⁎ Corresponding author at: Department of PsychologyBuilding, Rm. 400, Northern Illinois University, DeKalb, IL7077.

E-mail address: mkdemaray@niu.edu (M.K. Demaray)

0190-7409/$ – see front matter © 2013 Elsevier Ltd. All rihttp://dx.doi.org/10.1016/j.childyouth.2013.10.018

a b s t r a c t

a r t i c l e i n f o

Article history:Received 19 June 2013Received in revised form 14 October 2013Accepted 16 October 2013Available online 28 October 2013

Keywords:BullyingVictimizationSchool violencePerceptions

Bullying is a growing problem in many schools today, and accurate perceptions of bullying and victimization inschools are necessary in order for programs aimed at intervention for bullying behaviors to be effective. Thecurrent study examined agreement among students', teachers', and parents' perceptions of victimization acrossgender and grade level by surveying 137 students in grades 3–8, and their parents and teachers using a commonmeasure of bullying. Overall, students reported the highest levels of victimization, and teachers reported the low-est levels of victimization. This patternwas consistent across gender, but inconsistent across grade level. Studentsand parents had moderate agreement correlations on levels of victimization, whereas teachers and students hadlow agreement correlations on levels of victimization. Overall, when students and parents or teachers disagreed,the disagreement was an underestimate, rather than an overestimate on the adult's part.

© 2013 Elsevier Ltd. All rights reserved.

1. Introduction

From a social-ecological perspective, bullying and victimization areinfluenced bymultiple contexts that the child is embedded in, includinghome, school, peers, and the community (Espelage & Swearer, 2010).While many bullying prevention and intervention programs aim toinclude these multiple influences, few studies have assessed the degreeto which multiple sources are aware of and agree on the extent ofbullying occurring at the child's school. While perfect agreement ofperceptions may be an idealistic goal, it is important for research toinvestigate patterns and underlying reasons for discrepant perceptionsof victimization. This is essential for the effective development,implementation, and evaluation of bullying prevention and interven-tion programs. The current study adds to the literature on bullying byexamining school-wide agreement among perceptions of bullying andvictimization in a unique sample that consists of parallel ratings onvictimization from parents, teachers, and students for each participantin an entire school.

1.1. Overview of bullying in youth

Bullying is becoming an increasingly significant concern in manyschools, with as many as 30–35% of elementary and middle schoolstudents reporting being involved in bullying as either the bully, victim,or both (Bradshaw, Sawyer, & O'Brennan, 2007; Nansel et al., 2001).Furthermore, it is commonly reported that both bullies and victims

, Psychology-Computer Science60115, USA. Tel.: +1 815 753

.

ghts reserved.

experience poor psychosocial adjustment (Nansel et al., 2001). It iswell established in the literature that victims of bullying tend to reportlow self-esteem, poor peer relations, poor academic achievement, andgreater levels of internalizing and externalizing distress (e.g., anxiety,depression, aggression) than their peers (Hawker & Boulton, 2000;Hunter, Boyle, & Warden, 2007; Lopez & Dubois, 2005; Paul &Cillessen, 2007). Bullies may also be vulnerable to a myriad of issues.For example, significant relations among bullying behavior and anger,misconduct, depression, impulsivity, and maladaptive social skillshave been found (Bosworth, Espelage, & Simon, 1999).

While there are some inconsistencies in the way that bullying andvictimization are defined, many agree that bullying can be defined asproactive aggression where the behavior is repetitive, intended to beharmful, and within a disproportionate power relationship (Olweus,2010). Some researchers define bullying as physical (e.g., hitting, kicking,pushing) or relational (e.g., teasing, spreading rumors, excluding), whileothers define bullying as direct (i.e., physical or verbal confrontation) orindirect (less overt behaviors, such as spreading rumors or excluding)behaviors (Sveinsson&Morris, 2007). Bullying behaviors can be initiatedby an individual student or a group of students, just as the victim can bean individual or a group of students (Elinoff, Chafouleas, & Sassu, 2004).

1.2. Agreement rates in bullying and victimization

The different assessment methods utilized in bully research tend toresult in a wide range of prevalence rates for bullying and victimization(Swearer, Siebecker, Johnsen-Frerichs, & Wang, 2010). However, fewstudies have investigated the relations between who is reporting thebullying behavior and prevalence rates. Even fewer studies haveexamined convergence and divergence across sources within the samestudent. The current study utilized a unique sample that included

2092 M.K. Demaray et al. / Children and Youth Services Review 35 (2013) 2091–2100

school-wide participation consisting of students, their teachers, and themajority of their parents.

The few researchers that have investigated agreement rates inbullying have generally found that teachers report lower estimates ofbullying than students (Houndoumadi & Pateraki, 2001; Stockdale,Hangaduambo, Duys, Larson, & Sarvela, 2002). For example, Bradshawet al. (2007) examined the discrepancy between school staff andstudent perceptions of bullying behavior and bully-related attitudes ina large-scale study that included over 15,000 elementary, middle, andhigh school students and over 1500 school staff members in a publicschool district. Bradshaw and colleagues found that school staffunderestimated the number of students involved in bullying, with themajority of staff reporting bullying rates of less than 10% and studentsreporting bullying rates between 20% and 30%. Furthermore, the resultsof this study indicated that the discrepancy between teacher-reportedand student-reported bullying rates was larger in elementary school,with less than 1% of elementary school staff reporting bullying ratessimilar to elementary school students. Similar results were found in astudy examining agreement rates among teachers and students inseven rural elementary schools in the United States (Stockdale et al.,2002), as well as among a British sample in an inner-city secondaryschool (Pervin & Turner, 1994). A recent meta-analysis investigatingpredictors of bullying behaviors found that informant source countedfor significant variation in the reported prevalence rates. Cook,Williams, Guerra, and Kim (2010) found that prevalence rates reportedby peer nominations were significantly lower than student and teacherreports (parent report was excluded from the meta-analysis due to thelimited number of studies using parent report). Interestingly, the re-searchers found that prevalence rates reported by teachers and studentswere not significantly different. These results differ from the findingsgenerally reported in the literature that find teachers report lower levelsof bullying than students.

Research examining parent reports of bullying has generally foundthat parents also underestimate the extent of bullying behavior; howev-er, to a lesser extent than teachers. The difference in discrepancybetween student and parent reports as opposed to student and teacherreports may be due to the fact that children are more likely to telltheir parents rather than their teachers when they are victimized.Houndoumadi and Pateraki (2001) examined teacher and parentawareness of bullying from the students' perspectives in a sample of3rd through 6th grade students. Students reported that parents weremore aware of and have talked to their children about being victimized,however teachers were more aware of and have talked to studentsabout bullying others. This is consistent with prior research that hasfound that bullies are more likely to have damaged relationships withparents and perceive lower levels of parental support, and thereforeare less likely to talk to their parents about their experienceswith bullying (Demaray & Malecki, 2003; Rigby, 1993). Additionally,Houndoumadi and Pateraki (2001) found that girls and youngerstudents were more likely to tell their parents about being victimized.In a recent study examining parent and child concordance regardingteasing among 5th grade students and their parents, only 5% of the sam-ple indicated disagreement between the child being teased (i.e., thechild reported being teased, but the parent did not report their childhad been teased) (Holt, Kantor, & Finkelhor, 2009). Only 2% of the sam-ple agreed on the child teasing peers (i.e., the child reported they hadteased peers and the parent reported their child teased peers). Howev-er, for the majority of the sample (78%), both parents and childrenagreed that the child had not teased others. In a study examining parentand student perspectives of bullying at seven rural elementary schools,Stockdale et al. (2002) found that students and parents reported similarrates regarding verbal bullying, however parents' estimates of physicalbullying and victimization in general were significantly lower thanstudents' reports.

Some studies have found that agreement in reported bullying ratesmay differ by type of bullying. In a study previously described,

Bradshaw et al. (2007) found that students and staff generally agreedon the frequency of types of bullying, with direct verbal forms of bullying(e.g., name-calling, teasing) being the most frequently reported type ofbullying by elementary, middle, and high school staff and students,followed by physical forms of bullying (e.g., pushing, hitting). In a recentstudy, Waasdorp, Pas, O'Brennan, and Bradshaw (2011) examinedperceptual differences between students, staff, and parents regardingschool safety, school belonging, and bullying among a sample of 11,674students and 1027 staff at 44 elementary, middle, and high schools.Waasdorp and colleagues found that school level and type of bullyingsignificantly impacted perceptions of bullying. The discrepancy betweenstudent and staff reports ofwitnessing bullyingwas smaller in elementarythanmiddle and high schools. This finding differs from a study previouslydescribed that found that the discrepancies between teacher and studentreports of bullying were larger in elementary than middle and highschools (Bradshaw et al., 2007). Surprisingly, Waasdorp et al. (2011)also found that as students reported higher rates of indirect bullying,teachers reported witnessing more bullying than students. This could bedue to differing opinions between teachers and students on what“witnessing” bullying means. Students may not think that helping tospread a rumor or exclude a peer would represent a bullying situation.

In addition to type of bullying, discrepant viewpoints may be due todiffering opinions on definitions of bullying. Research has found thatteachers are less likely to consider indirect behaviors (e.g., exclusion,spreading rumors, etc.) as bullying and include repetitive behavior intheir definitions (Mishna, Scarcello, Pepler, & Wiener, 2005). Parentsand teachers maymistake verbal and social forms of bullying as playfulbehavior between friends and may not report it as bullying (Bauman &Del Rio, 2006). Characteristics of teachers, such as previous experiencewith bullying, may also affect reports of bullying. For example,Waasdorp et al. (2011) found that school staff members who reportedbeing bullied in the past weremore likely to report witnessing bullying.Other variables, such as the size and climate of school, the age andgender of students, and the social status of the bullies and victims mayinfluence teachers' perceptions of bullying (Holt et al., 2009). There islimited research investigating variables that influence parents' percep-tions of bullying, although it is thought that similar variables such asparents' previous experience with bullying, and age and gender of thechild may influence parents' perceptions as well (Holt et al., 2009).

Discrepant viewpoints among students, teachers, and parents onbullying have important implications for prevention and early interven-tion services. Many theories and much of the research on bullying be-haviors support the notion of a social-ecological framework, whichsuggests that bullying and victimization are reciprocally influenced bythe individual, family, peer group, school, community, and society(Espelage & Swearer, 2010). According to this perspective, the most ef-fective bully prevention and intervention practices are those programswhich take a multi-systematic perspective (Holt et al., 2009). Many ofthese programs (e.g., Bully-Proofing Your School, 2004) require collab-oration among school staff, students and the community to appropriate-ly implement the program and alter bullying behaviors and attitudes atschool. Therefore, accurate assessment and interpretation of student,parent, and teacher reports of bullying is essential for these programsto be developed, implemented, and evaluated effectively.

While these studies have made important contributions to the liter-ature, a major limitation is that the teachers completed questionnairesasking them to make global ratings of students' victimization asopposed to rating each individual student's level of victimization. Thecurrent study is unique in that the teachers filled out a questionnaireabout each of their students. Furthermore, parents filled out a question-naire about each of their children. This allows the researchers to exam-ine exact agreement among students, their teachers, and their parentson perceptions of victimization. In addition, prior research has investi-gated agreement on total scores, which may be misleading. Forexample, a teacher and child may have similar total scores for levels ofvictimization but have endorsed different items. The current study

Table 1Participants.

Grade Student-completed % Parent-completed

Total sample 137 100 103Boys 75 54.7 55Girls 62 45.3 48

3rd 25 18.2 20Boys 13 52.0 10Girls 12 48.0 10

4th 26 19.0 21Boys 12 46.2 10Girls 14 53.8 11

5th 26 19.0 21Boys 17 65.4 13Girls 9 34.6 8

6th 17 12.4 13Boys 11 64.7 10Girls 6 35.3 3

7th 19 13.8 13Boys 10 52.6 6Girls 9 47.4 7

8th 24 17.5 15Boys 12 50.0 6Girls 12 50.0 9

2093M.K. Demaray et al. / Children and Youth Services Review 35 (2013) 2091–2100

includes item-level data and analyses to further investigate the agree-ment among different sources on specific victimization experiences.

1.3. Purpose of the present study

The present study investigated agreement among parents, teachers,and students in perceptions of victimization frombullying at a small K-8private school. The current study added to the literature on agreementrates in victimization by investigating differences in agreement bygender and grade level, as well as by individual items (e.g., physical,verbal, etc.). To the authors' knowledge, no study to date has examinedparent, teacher, and student perspectives on victimization utilizingparallel forms and comparing agreement on each item (as opposed tototal scores). Therefore, many of the analyses are exploratory in nature.Few studies have been able to obtain cross-informant data from threesources. This is a unique sample because the study obtained school-wide participation, including full participation of the students, theirteachers', and their parents'. The current study was guided by the fol-lowing research questions: (1) What is the level of agreement amongparents, teachers, and students in overall perceptions of victimization?Do total agreement rates vary by gender and grade level across thesesources? (2) What is the level of agreement among parents andstudents and teachers and students at the item level for rates of victim-ization? Do item level agreement rates vary by gender across thesesources? Given that prior research has found that teachers and parentstend to report lower levels of bullying than students (e.g., Bradshawet al., 2007; Stockdale et al., 2002), the researchers expected studentsto report higher levels of bullying than both parents and teachers.Furthermore, based on past research (e.g., Houndoumadi & Pateraki,2001), it was hypothesized that the discrepancy in agreement ratewould be larger between teachers and students than parents andstudents regarding reported rates of victimization. Finally, regardingage and gender differences, it was hypothesized that there would behigher rates of agreement between girls and teachers, and betweenparents and younger students (Houndoumadi & Pateraki, 2001).

2. Method

2.1. Participants

The participants in this study include 137 students (grades 3–8),their parents, and their teachers (n = 6) from a private K-8 school inthe Midwest. Although this sample is relatively small, it is unique andspecialized. All but three students in grades 3–8 participated in thestudy, 78% of the students were rated by their parents, and 100% ofthe students were rated by their teachers. The school is a small Catholicschool that has one classroom for each grade level of no more than 26students per class. It is noteworthy that the 7th and 8th students attendschool on the same campus, however in a separate building. The studentsample was evenly divided between boys (n = 75) and girls (n = 62).The enrollment at this school by ethnicity was 86% white with 5%Hispanic/Latino, 2% Asian or Pacific Islander, 5% Multi-Racial, and 2%“other”. Only 6% of the students in the school are non-Catholic. SeeTable 1 for participant demographics. The school receives federal Title1 funds with 14% of students receiving Title 1 services. The students inthird through seventh grade average at the 69th, 60th, 70th, 64th, and82nd national percentile rank on their most recent Iowa Test of BasicSkills (ITBS). The participating teachers have taught in this school forbetween 9 and 25 years, with most teaching at the school for 11 yearsor more. All teachers were white and female.

2.2. Measures

Students, their parents, and their teachers completed modifiedversions of the Revised Olweus Bully/Victim Questionnaire (OB/VQ;Olweus, 1996) which includes parallel items regarding bullying and

victimization, as well as a definition of bullying to assist the rater withcompleting the survey. The original version of the questionnaire wasdeveloped for students, although the measure was adapted to be usedwith parents and teachers aswell. Overall, there is very limited informa-tion available regarding the psychometric properties of the OB/VQ. Assome evidence of reliability, Solberg and Olweus (2003) used thismeasure with thirty-seven schools with students in grades 5–9 inNorway and found reliability estimates of .88 for students being bulliedand .87 for bullying other students. Lee and Cornell (2010) foundmodest evidence of concurrent validity for the OB/VQ using a sampleof middle school students in the U.S. The researchers found thatself-reported bullying and victimization on the OB/VQ was significantlyrelated to peer nominations for bullying and victimization and academicgrades. Although the measure has limited evidence of reliability andvalidity, it is widely used in both research and applied settings to assessbullying and victimization.

Students are asked to rate the frequency of their experiences in thepast couple of months as bullies/victims by responding to questionson a five point scale: “1 = It hasn't happened to me in the past coupleof months,” “2 = only once or twice a month,” “3 = 2 or 3 times amonth,” “4 = about once a week,” and “5 = several times a week.”The student version of OB/VQ has a total of 39 items. The currentstudy only used the 10 items that assessed victimization. A samplevictimization item is “I was called mean names, was made fun of, orwas teased in a hurtful way.”

The student version of the OB/VQ was adapted for parent use bychanging the wording to apply to “my child” and adding a sixth ratingscale option of 1 for “I don't know.” The parent version had a total of10 items assessing victimization. A sample victimization item on theparent version is “Your child was called mean names, was made funof, or was teased in a hurtful way.”

The student version of the OB/VQ was also adapted for teacher useby inserting the student's name within the question and adding asixth rating scale option of 1 for “I don't know.” Teachers rated eachstudent in their class on a total of 10 items assessing victimization. Asample victimization item on the teacher version is “How much doyou think (student's name in your classroom) was called mean names,made fun of, or teased in a hurtfulway?” Each itemof the teacher surveyincludes a table with all of the students' names in the classroom downthe first column and the six points of the rating scale across the nextsix columns. This way, teachers could efficiently rate each student intheir classrooms on each of the items.

Two items were changed in all three versions of the survey due toconcerns about the content. “I was bullied/bullied him or her with

Table 2Victimization Scores reported by students, teachers, and parents by gender and gradelevel.

Source M SD N Minimum Maximum

Student Total sample 14.83 6.75 134 10 46Gender Boys 14.75 7.24 72 10 46

Girls 14.92 6.18 62 10 31Grade level 3rd/4th 17.43 7.16 49 10 36

5th/6th 13.67 7.14 42 10 467th/8th 13.00 4.81 43 10 31

Grade level/gender

3rd/4th boys 17.96 7.73 23 10 363rd/4th girls 16.96 6.74 26 10 305th/6th boys 14.41 8.45 27 10 465th/6th girls 12.33 3.70 15 10 257th/8th boys 11.82 2.40 22 10 197th/8th girls 14.24 6.28 21 10 31

Teacher Total sample 11.12 3.27 137 9 29Gender Boys 10.73 2.47 75 9 25

Girls 11.60 4.01 62 10 29Grade level 3rd/4th 10.67 1.61 51 10 18

5th/6th 11.09 3.54 43 10 257th/8th 11.70 4.30 43 9 29

Grade level/gender

3rd/4th boys 10.28 .74 25 10 133rd/4th girls 11.04 2.09 26 10 185th/6th boys 10.11 .31 28 10 115th/6th girls 12.93 5.64 15 10 257th/8th boys 12.05 4.27 22 9 257th/8th girls 11.33 4.41 21 10 29

Parent Total sample 13.03 4.58 104 10 30Gender Boys 12.96 4.67 56 10 30

Girls 13.10 4.52 48 10 28Grade level 3rd/4th 13.17 4.16 42 10 26

5th/6th 13.06 5.30 34 10 307th/8th 12.79 4.39 28 10 26

Grade level/gender

3rd/4th boys 13.62 4.88 21 10 263rd/4th girls 12.71 3.33 21 10 235th/6th boys 13.04 5.17 23 10 305th/6th girls 13.09 5.80 11 10 287th/8th boys 11.67 3.06 12 10 207th/8th girls 13.63 5.11 16 10 26

2094 M.K. Demaray et al. / Children and Youth Services Review 35 (2013) 2091–2100

mean names, comments or gestures with a sexual meaning” waschanged to “I was bullied/bullied him or her with mean names,comments or gestures with a bad meaning,” and “I was bullied/bulliedhim or her with mean names or comments about my race or color”was changed to “I was bullied/bullied him or her with mean names orcomments about the way I look.”

The items on the OB/VQ were scored by summing 10 items for vic-timization for students, parents and teachers. If parents and teachersrated an item (“I don't know”), a score of 1 was included in the sum ofthe scores so that an “I don't know” was given the same weight as ascore of 1 = “Never happened.” The rationale is that if a teacher/parentdid not know if a child was victimized that should be reflected in thescore — not treated as missing data. However, because we were inter-ested in agreement at the item level, the “I don't know” responseswere not recoded for the item level analyses and were treated asmissing data. No missing data estimation procedures were conductedto replacemissing values at the item level. Additionally, for the analysesthat required total summed scores, data were only summed if eight ofthe ten items were complete for each scale.

2.3. Procedures

Datawere collected as part of a school-wide social emotional assess-ment using a passive consent process andwere approved for analysis bythe IRB. All students in grades 3–8 (n = 137) were given the OB/VQ tofill out in their homeroom class. The survey was administered by thehomeroom teachers; teachers read the instructions to the surveyaloud at all grade levels and were available to answer questions.Teachers in grades 3 and 4 were encouraged to read the items aloudto their classes. All student surveys were kept confidential by using

unique identification numbers and by having the teachers seal the stu-dents' surveys in an envelope after theywere completed. Parent surveyswere sent homewith each student in grades 3–8 (parents received onesurvey for each child attending the school) with a sealable returnenvelope. One hundred and three of 138 (75.2%) parent surveys werecompleted and returned. Each teacher in grades 3–8 (100%) completeda survey on each student in his or her class.

3. Results

Means and standard deviations of the victimization scores reportedby all sources can be found in Table 2. Descriptive data at the itemlevel are presented in Table 3 (self report), Table 4 (parent report),and Table 5 (teacher report). Effect sizes were calculated (Cohen's d;Cohen, 1988) and correlations were compared using z-scores fromFisher's r-to-z transformation (Fisher, 1921).

3.1. Preliminary analyses

Gender and grade level differences in levels of reported victimizationand bullyingwere analyzed via three ANOVAs. In each ANOVA the victimscore fromOBVQwas entered as the dependent variable andGender andGrade Level (3 and 4; 5 and 6; 7 and 8)were entered as the independentvariables. Grade levels were grouped due to lower n sizes in order toinvestigate developmental trends. See Table 2 for descriptive data.

For parent-reported levels of victimization, the ANOVA was notsignificant by Gender, F (1, 98) = .149, p = .700; Grade Level, F (2,98) = .109, p = .896; or the Gender by Grade Level interaction, F (2,98) = .790, p = .457. For teacher-reported levels of victimization, theANOVA was not significant for Gender, F (1, 131) = 2.942, p = .089,or Grade Level, F (2, 131) = 1.413, p = .247. The Grade Level byGender interaction was significant, F (2, 131) = 3.146, p b .05. Visualanalysis of the plot of themeans indicated that teachers rated victimiza-tion similarly for boys and for girls except that at the 5th/6th grade level,teachers rated girls as being victimized more frequently than boys(Cohen's d = .71, medium).

For student self-reported levels of victimization, the ANOVAwas notsignificant for Gender, F (1, 128) = .036, p = .850, nor was the Genderby Grade Level interaction, F (2, 128) = 1.37, p = .259. However, themain effect of Grade Level was significant, F (2, 128) = 6.62, p b .01.Post hoc comparisons using a Bonferroni correction indicated that stu-dents in 3rd/4th grade reported higher victimization than the 5th/6th(Cohen's d = .53, medium) and the 7th/8th grade students (Cohen'sd = .73, medium).

3.2. Main analyses

Question 1:What is the level of agreement among parents, teachers,and students in overall perceptions of victimization? Do total agree-ment rates vary by gender and grade level across these sources? Thisquestion was answered by conducting repeated measures ANOVAs(RMANOVAs) and correlations among the victim score across sources(parent, teacher, and student). These RMANOVAs and correlationswere conducted on the total sample, by gender, and by grade level(grades 3 and 4; 5 and 6; 7 and 8). The correlationswere also conductedby grade level and gender. See Tables 6.1, 6.2, and 6.3 for descriptiveanalyses and effect sizes comparing rater scores. Additionally, seeTables 7.1 and 7.2 for the correlations of victim scores.

On the total sample, examining the three victimization scores (self,parent, and teacher), the repeated measures ANOVA was significant,Wilks' Lambda = .732, F (2, 100) = 18.268, p b .001, indicating signif-icant differences among students', parents', and teachers' perceptions oflevels of victimization. Follow-up comparisons found that all threescores differed significantly (ps b .001), with students reporting thehighest levels of victimization and teachers reporting the lowest levelsof victimization (see Table 6.1).

Table 3Student rating descriptives by gender and grade level.

Item Group N M SD Range

Been bullied in past Total 128 1.92 1.34 1–5Boys/girls 69/59 1.84/2.02 1.30/1.40 1–5/1–53–4/5–6/7–8 43/42/43 2.51/1.67/1.58 1.56/1.16/1.07 1–5/1–5/1–5

Been called names, made fun Total 134 1.84 1.21 1–5Boys/girls 72/62 1.75/1.94 1.18/1.24 1–5/1–53–4/5–6/7–8 49/42/43 2.14/1.57/1.74 1.38/1.16/1.07 1–5/1–5/1–5

Been left out, excluded Total 133 1.62 1.06 1–5Boys/girls 72/61 1.58/1.66 1.07/1.06 1–5/1–53–4/5–6/7–8 48/42/43 1.85/1.55/1.42 1.17/1.13/.82 1–5/1–5/1–5

Been hit, kicked, pushed, shoved Total 134 1.45 1.00 1–5Boys/girls 72/62 1.53/1.35 1.11/.85 1–5/1–53–4/5–6/7–8 49/42/43 1.98/1.24/1.05 1.32/.79/.21 1–5/1–5/1–2

Others told lies, spread rumors Total 134 1.48 .92 1–5Boys/girls 72/62 1.46/1.50 .86/.99 1–5/1–53–4/5–6/7–8 49/42/43 1.78/1.29/1.33 1.12/.71/.74 1–5/1–4/1–5

Money taken or things damaged Total 133 1.25 .68 1–5Boys/girls 72/61 1.26/1.23 .67/.69 1–4/1–53–4/5–6/7–8 49/41/43 1.49/1.22/1.00 .94/.57/.00 1–5/1–4/1–1

Been threatened or forced Total 132 1.27 .67 1–5Boys/girls 71/61 1.34/1.20 .79/.48 1–5/1–33–4/5–6/7–8 49/41/42 1.49/1.27/1.02 .71/.84/1.54 1–4/1–5/1–2

Called names about looks Total 133 1.57 1.03 1–5Boys/girls 72/61 1.51/1.64 1.02/1.05 1–5/1–53–4/5–6/7–8 48/42/43 1.81/1.45/1.42 1.23/.92/.85 1–5/1–5/1–5

Called names or gestures Total 133 1.46 .88 1–5Boys/girls 71/62 1.46/1.45 .86/.90 1–4/1–53–4/5–6/7–8 48/42/43 1.54/1.40/1.42 .90/.83/.91 1–5/1–4/1–5

Been bullied on cell or internet Total 131 1.15 .60 1–5Boys/girls 70/61 1.16/1.15 .58/.63 1–4/1–53–4/5–6/7–8 48/41/42 1.27/1.10/1.07 .82/.49/.34 1–5/1–4/1–3

Note. 1 = it hasn't happened to me in the past couple of months; 2 = only once or twice; 3 = 2 or 3 times a month; 4 = about once a week; 5 = several times a week.

2095M.K. Demaray et al. / Children and Youth Services Review 35 (2013) 2091–2100

For boys and for girls, examining the three victimization scores (self,parent, and teacher), the repeated measures ANOVAs were significantfor both boys and girls, Wilks' Lambda = .744, F (2, 52) = 8.94,p b .001, and Wilks' Lambda = .691, F (2, 46) = 10.31, p b .001,

Table 4Parent rating descriptives by gender and grade level.

Item Group “Don't know” N

Been bullied in past Total 9Boys/girls 3/63–4/5–6/7–8 5/2/2

Been called names made fun Total 13Boys/girls 7/63–4/5–6/7–8 6/5/2

Been left out, excluded Total 14Boys/girls 8/63–4/5–6/7–8 8/4/2

Been hit, kicked, pushed, shoved Total 7Boys/girls 5/23–4/5–6/7–8 2/4/1

Others told lies, spread rumors Total 17Boys/girls 7/103–4/5–6/7–8 6/8/3

Money taken or things damaged Total 9Boys/girls 5/43–4/5–6/7–8 5/3/1

Been threatened or forced Total 9Boys/girls 4/53–4/5–6/7–8 6/2/1

Called names about looks Total 11Boys/girls 6/53–4/5–6/7–8 4/4/3

Called names or gestures Total 14Boys/girls 7/73–4/5–6/7–8 3/6/5

Been bullied on cell or internet Total 11Boys/girls 5/63–4/5–6/7–8 5/4/2

Note. 1 = it hasn't happened in the past couple of months; 2 = only once or twice; 3 = 2 or

respectively indicating significant differences among students', parents',and teachers' perceptions of levels of victimization. Follow-up compar-isons for both boys and girls again found that all three scores differedsignificantly (ps b .001), with both students reporting the highest levels

N M SD Range

93 1.77 1.08 1–552/41 1.77/1.78 1.17/1.13 1–5/1–537/31/25 1.95/1.74/1.56 1.20/1.18/1.00 1–5/1–5/1–591 1.79 1.12 1–549/42 1.67/1.93 1.11/1.13 1–5/1–536/29/26 1.83/1.90/1.62 1.08/1.29/.98 1–5/1–5/1–590 1.61 1.00 1–548/42 1.60/1.62 1.03/.99 1–5/1–534/30/26 1.50/1.67/1.69 .62/1.15/1.23 1–3/1–5/1–596 1.19 .69 1–550/46 1.22/1.15 .74/.63 1–4/1–539/30/27 1.31/1.07/1.15 .83/.25/.77 1–4/1–2/1–586 1.29 .80 1–549/37 1.29/1.30 .82/.78 1–5/1–536/26/24 1.28/1.31/1.29 .70/.88/.86 1–4/1–5/1–595 1.01 .10 1–251/44 1.00/1.02 .00/.15 1–1/1–237/31/27 1.00/1.00/1.04 .00/.00/.19 1–1/1–1/1–295 1.01 .10 1–252/43 1.02/1.00 .14/.00 1–2/1–136/32/27 1.03/1.00/1.00 .17/.00/.00 1–2/1–1/1–193 1.42 .85 1–550/43 1.44/1.40 .97/.69 1–5/1–438/30/25 1.42/1.50/1.32 .89/1.01/.56 1–5/1–5/1–389 1.37 .91 1–548/41 1.40/1.34 .98/.82 1–5/1–539/27/23 1.33/1.44/1.35 .87/1.01/.88 1–5/1–5/1–593 1.04 .33 1–451/42 1.00/1.10 .00/.48 1–1/1–437/30/26 1.00/1.00/1.15 .00/.00/.61 1–1/1–1/1–4

3 times a month; 4 = about once a week; 5 = several times a week.

Table 5Teacher rating descriptives by gender and grade level.

Item Group “Don't know” N N M SD Range

Been bullied in past Total 84 53 1.60 1.15 1–5Boys/girls 46/38 29/24 1.48/1.75 .95/1.36 1–4/1–53–4/5–6/7–8 48/23/13 3/20/30 2.00/1.45/1.67 .00/1.23/1.15 2–2/1–5/1–5

Been called names made fun Total 67 70 1.50 1.06 1–5Boys/girls 39/28 36/34 1.33/1.68 .79/1.68 1–4/1–53–4/5–6/7–8 42/24/1 9/19/42 2.22/1.42/1.38 .44/1.26/1.01 2–3/1–5/1–5

Been left out, excluded Total 25 112 1.26 .67 1–5Boys/girls 13/12 62/50 1.16/1.38 .45/.85 1–3/1–53–4/5–6/7–8 25/0/0 26/43/43 1.38/1.21/1.23 .64/.67/.68 1–3/1–5/1–4

Been hit, kicked, pushed, shoved Total 1 136 1.07 .44 1–4Boys/girls 1/0 74/62 1.12/1.00 .60/.00 1–4/1–13–4/5–6/7–8 1/0/0 50/43/43 1.00/1.00/1.21 .00/.00/.77 1–1/1–1/1–4

Others told lies, spread rumors Total 61 75 1.23 .67 1–5Boys/girls 36/25 38/37 1.05/1.41 .32/.86 1–3/1–53–4/5–6/7–8 29/13/19 22/30/23 1.32/1.20/1.17 .65/.76/.58 1–3/1–5/1–3

Money taken or things damaged Total 26 111 1.00 .00 1–1Boys/girls 12/14 63/48 1.00/1.00 .00/.00 1–1/1–13–4/5–6/7–8 26/0/0 25/43/43 1.00/1.00/1.00 .00/.00/.00 1–1/1–1/1–1

Been threatened or forced Total 1 136 1.00 .00 1–1Boys/girls 1/0 74/62 1.00/1.00 .00/.00 1–1/1–13–4/5–6/7–8 0/0/1 51/43/42 1.00/1.00/1.00 .00/.00/.00 1–1/1–1/1–1

Called names about looks Total 86 51 1.31 .99 1–5Boys/girls 51/35 24/27 1.08/1.52 .41/1.28 1–3/1–53–4/5–6/7–8 26/40/20 25/3/23 1.00/4.67/1.22 .00/.58/.74 1–1/4–5/1–4

Called names or gestures Total 86 51 1.33 .79 1–5Boys/girls 51/35 24/27 1.29/1.37 .62/.93 1–3/1–53–4/5–6/7–8 25/41/20 26/2/23 1.12/3.00/1.43 .33/.00/.99 1–2/3–3/1–5

Been bullied on cell or internet Total 137 0Boys/girls 75/62 0/03–4/5–6/7–8 51/43/43 0/0/0

Note. 1 = it hasn't happened in the past couple of months; 2 = only once or twice; 3 = 2 or 3 times a month; 4 = about once a week; 5 = several times a week.

Table 6.2Agreement rates by gender and rater.

Gender Rater M SD N Comparison Cohen's d Label

Girl Parent 13.10 4.52 48 Parent/teacher .37 SmallTeacher 11.58 3.72 48 Parent/student .41 SmallStudent 15.40 6.45 48 Teacher/student .73 Medium

Boy Parent 13.07 4.72 54 Parent/teacher .70 MediumTeacher 10.50 2.18 54 Parent/student .28 SmallStudent 14.85 7.77 54 Teacher/student .76 Medium

2096 M.K. Demaray et al. / Children and Youth Services Review 35 (2013) 2091–2100

of victimization and teachers reporting the lowest levels of victimiza-tion (see Table 6.2).

Finally, the same analysis was conducted for each grade level thathad ratings from all three sources (3rd/4th, 5th/6th, and 7th/8th). Therepeated measures ANOVAs were not significant for 5th/6th grade and7th/8th grade students, Wilks' Lambda = .865, F (2, 32) = 2.50,p = .098, Wilks' Lambda = .850, F (2, 26) = 2.29, p = .121, respec-tively. The repeated measures ANOVA was significant for 3rd/4thgrade, Wilks' Lambda = .458, F (2, 38) = 22.47, p b .001. Follow-upcomparisons for the 3rd/4th grade followed the same pattern as previ-ous analyses with all three scores differing significantly (ps b .001),with students reporting the highest levels of victimization and teachersreporting the lowest levels of victimization (see Table 6.3).

Correlations indicate that on the total sample for third througheighth grade, parents and students had moderate agreement (.47,p b .01) on overall levels of victimization. The correlation betweenteacher and student scores was not significant, but the correlationbetween parent and teacher scores was (.33, p b .01). A correlation of.60, p b .01 indicated that parents of boys in grades three througheight had strong agreement in their reports of the boys' victimization.No such relation was found for teachers and boys' ratings nor betweenparent and teacher ratings for boys. For girls, the relations betweengirls' and their parents' and teachers' reports of victimization were notas strong. Interestingly, parents and teachers showed strong agreementabout their reports of girls' victimization (r = .54, p b .01).

Table 6.1Agreement rates by rater.

Rater M SD N Comparison Cohen's d Label

Parent 13.09 4.60 102 Parent/teacher .53 MediumTeacher 11.01 3.04 102 Parent/student .34 SmallStudent 15.11 7.15 102 Teacher/student .75 Medium

When examining results overall by grade level, the 3rd/4th graders'reports of victimization were strongly correlated with the reports oftheir parents (r = .59, p b .05) and their teachers (r = .53, p b .01).Parents and their 5th/6th graders showed moderate agreement(r = .36, p b .05) with no relation between teachers and their 5th/6thgraders' reports of victimization. There was moderate relation between7th/8th grade student reports with their parents (r = .49, p b .05). It isinteresting that at the 3rd/4th and 5th/6th grade level parents andteachers agreed overall at a moderate level (r = .40 and .46, p b .01,respectively).

Looking even more specifically at these relations by grade level andgender, thereweremoderate to strong relations between3rd/4th graders'reports and those of their parents and teachers. At the 5th/6th grade level,

Table 6.3Agreement rates by grade level and rater.

Grade level Rater M SD N Comparison Cohen's d Label

3rd/4th Parent 13.33 4.20 40 Parent/teacher .81 LargeTeacher 10.73 1.77 40 Parent/student .68 MediumStudent 17.33 7.16 40 Teacher/student 1.27 Large

Note. There were no significant differences in reported victimization levels by rater in5th/6th and 7th/8th grade levels.

Table 7.1Correlations and z-scoresa between student, parent, and teachers' victim total scores.

Parent and studentvictim score correlation

Teacher and studentvictim score correlation

Raterz-scoreb

Total sample .465⁎⁎ .123 2.85⁎⁎

GenderBoys .597⁎⁎ −.024 3.86⁎⁎

Girls .280 .260⁎ .11Gender z-score 1.96⁎ 1.64

Grade level and gender3rd/4th .592⁎ .533⁎⁎ .39

Boys .769⁎⁎ .610⁎⁎ .92Girls .327 .651⁎⁎ 1.39

Gender z-score 1.97⁎ .225th/6th .362⁎ .061 1.32

Boys .471⁎ .174 1.11Girls .046 .382 .78

Gender z-score 1.11 .647th/8th .485⁎ .116 1.62

Boys .575 −.050 1.74Girls .459 .236 .70

Gender z-score .37 .88

a Fisher r-to-z transformation.b Rater z-score is the comparison of the parent/student correlation and the teacher/

student correlation.⁎ p b .05.⁎⁎ p b .01.

2097M.K. Demaray et al. / Children and Youth Services Review 35 (2013) 2091–2100

parents' and boys' ratings and teachers' and girls' ratingsweremoderatelyrelated. Parents' and teachers' ratings were related for girls but not forboys. Finally, 7th/8th grade girls' and boys' ratings were related to theirparents' ratings at a moderate level, but not so with their teachers.See Tables 7.1 and 7.2 for these specific correlations and comparisons ofthese correlations. Note that there are several significant z-scorescomparing correlations that will be noted in the Discussion section.

Question 2: What is the level of agreement among parents andstudents and teachers and students at the item level for rates of victimiza-tion? Do item level agreement rates vary by gender across these sources?In order to look at agreement rates among the sources at the item level,responses at the item level were recoded to be dichotomous with1 = did not experience victimization and consisted of items ratedas “1 = It hasn't happened” or it happened “2 = only once or twice amonth” and 2 = experienced victimization if the item was rated it

Table 7.2Correlations and z-scoresa between parent and teachers' victim total scores.

Parent and teacher victimscore correlation

Total sample .331⁎⁎

GenderBoys .069Girls .542⁎⁎

Gender z-score 2.65⁎⁎

Grade level and gender3rd/4th .401⁎⁎

Boys .700⁎⁎

Girls .517⁎

Gender z-score .895th/6th .464⁎⁎

Boys −.125Girls .875⁎⁎

Gender z-score 3.54⁎⁎

7th/8th .086Boys .111Girls .205

Gender z-score .22

a Fisher r-to-z transformation.⁎ p b .05⁎⁎ p b .01.

happened “3 = 2 or 3 times a month” “4 = about once a week” or“5 = several times a week”.

In order to investigate item level agreement rates across the sources,each item was considered an agreement if both the adult (parent orteacher) and the child/student rated the item a 1 (did not experiencevictimization) or both rated it a 2 (experienced victimization). If theadult rated the item “I don't know” this was considered missingdata. Table 8 presents agreement data across the four possible options:1) both adult and student rated the child as having experienced victim-ization; 2) both the adult and student rated the child as not experienc-ing victimization; 3) the adult rated the child as victimized but the childrated themselves as not victimized (adult overestimate); and 4) theadult rated the child as not victimized but the child rated themselvesas victimized (adult underestimate). The table also includes the totalagreement rate (sums of items 1 and 2 above) and the total disagree-ment rate (sums of 3 and 4 above). Lastly, Cohen's Kappa data arepresented in the last column of the table. Cohen's Kappa is an estimateof the agreement between two sources and adjusts for the levels dueto chance alone. A Kappa of 0 would indicate chance agreement and aKappa of 1.0 would indicate perfect agreement. According to Viera andGarrett (2005), Kappa values can be interpreted as follows: slightagreement = less than 0.20; fair agreement = 0.20 to 0.40; moderateagreement = 0.41 to 0.60; substantial agreement = 0.61 to 0.80; andalmost perfect agreement = 0.81 to 1.00.

First, the item level agreement between adult and child sources(parent–child and teacher–student) was examined on the overallsample of 3rd through 8th grade students (see Table 8). Teachers andstudents demonstrated very little agreement, with Kappas for mostitems in the “slight agreement” or “less than chance” range. Percentageagreement on the items where Kappa could be calculated ranged from70% to 90%. Kappas were not able to be calculated on three items be-cause of lack of variability in responses, specifically having things stolen,being threatened or forced to do something, and being bullied on theinternet or cell phone. Teachers rated the first two of these items asnot happening with little to no variability (with 94 and 95% agreementwith their students that it had not happened) and on the cell/internetitem, they uniformly answered “don't know” resulting in missing data.Overall, when teachers and students did not agree, the disagreementwas an underestimate rather than an overestimate on the teachers'part on all items but one (“others told lies, spread rumors”). Thepercentage of teachers' underestimate of students' victimization rangedfrom 4% to 21% of the time.

Parents and their children had better agreement with most Kappavalues in the “fair agreement” range. Percentage of total agreement forthe items where Kappa could be calculated ranged from 79% to 96%.Again, Kappas were not calculated for the items regarding having thingsstolen and being threatened or forced to do something due to lack of re-sponse variability. Percent agreement for those two items was 95 and96%, respectively. When parents and their children disagreed, the dis-agreementwas an underestimate rather than an overestimate on the par-ents' part on all items but one (where there was equal under- andoverestimate; “been left out, excluded”). The percentage of parents' un-derestimate of students' victimization ranged from 4% to 14% of the time.

Next, the item level agreement between parents and their childrenwas examined for 3rd through 8th grade boys and girls separately bygender (see Table 9). Overall, Kappa values indicate that parents andtheir boys showed stronger agreement than parents and their girls.Kappa values on the overall “has your child/you been bullied” itemwas in the “substantial agreement” range for boys (Kappa = .64) andin the “slight agreement” range for the girls (Kappa = .14). Similarly,for parents and their boys, the majority of the Kappas on the remainingitems (where Kappa could be computed) were in the “moderate agree-ment” range;whereas for girls and their parents, Kappasweremostly inthe “less than chance agreement” category. One notable exceptionwas on the cyberbullying item where parents and their girls showed“substantial agreement”.

Table 8Parent and teacher with student percent agreement and Kappa by item level on victim items.

Item Adult rater Yes–yesagreement

No–noagreement

Totalagreement

Yes–nooverestimate

No–yesunderestimate

Totaldisagreement

Kappa

Been bullied in past Parent 9 (10.3%) 60 (69%) 69 (79.3%) 6 (6.9%) 12 (13.8%) 18 (20.7%) .37Teacher 1 (1.9%) 37 (71.2%) 38 (73.1%) 6 (11.5%) 8 (15.4%) 14 (26.9%) −.03

Been called names made fun Parent 9 (10%) 64 (71.1%) 73 (81.1%) 6 (6.7%) 11 (12.2%) 17 (18.9%) .40Teacher 3 (4.3%) 46 (65.7%) 49 (70%) 6 (8.6%) 15 (21.4%) 21 (30%) .06

Been left out, excluded Parent 5 (5.7%) 66 (75.9%) 71 (81.6%) 8 (9.2%) 8 (9.2%) 16 (18.4%) .28Teacher 2 (1.8%) 92 (84.4%) 94 (86.2%) 5 (4.6%) 10 (9.2%) 15 (13.8%) .14

Been hit, kicked, pushed, shoved Parent 2 (2.1%) 82 (87.2%) 84 (89.4%) 2 (2.1%) 8 (8.5%) 10 (10.6%) .24Teacher 0 (0%) 114 (85.7%) 114 (85.7%) 3 (2.3%) 16 (12.0%) 19 (14.3%) −.04

Others told lies, spread rumors Parent 1 (1.2%) 72 (85.7%) 73 (86.9%) 5 (6%) 6 (7.1%) 11 (13.1%) .08Teacher 1 (1.4)% 64 (88.9%) 65 (90.3%) 4 (5.6%) 3 (4.2%) 7 (9.8%) .17

Money taken or things damaged Parent 0 (0%) 88 (94.6%) 88 (94.6%) 0 (0%) 5 (5.4%) 5 (5.4%) naTeacher 0 (0%) 102 (93.6%) 102 (93.6%) 0 (0%) 7 (6.4%) 7 (6.4%) na

Been threatened or forced Parent 0 (0%) 88 (95.7%) 88 (95.7%) 0 (0%) 4 (4.3%) 4 (4.3%) naTeacher 0 (0%) 125 (94.7%) 125 (94.7%) 0 (0%) 7 (5.3%) 7 (5.3%) na

Called names about looks Parent 5 (5.5%) 74 (81.3%) 79 (86.8%) 5 (5.5%) 7 (7.7%) 12 (13.2%) .38Teacher 0 (0%) 37 (75.5%) 37 (75.5%) 5 (10.2%) 7 (14.3%) 12 (24.5%) −.14

Called names or gestures Parent 3 (3.5%) 72 (83.7%) 75 (87.2%) 4 (4.7%) 7 (8.1%) 11 (12.8%) .28Teacher 0 (0%) 35 (72.9%) 35 (72.9%) 5 (10.4%) 8 (16.7%) 13 (27.1%) −.15

Been bullied on cell or internet Parent 1 (1.1%) 85 (94.4%) 86 (95.5%) 0 (0%) 4 (4.4%) 4 (4.4%) .32Teacher na na na na na na na

Note. na = question was not asked or data did not allow computation of Kappa.

2098 M.K. Demaray et al. / Children and Youth Services Review 35 (2013) 2091–2100

Finally, the item level agreement between teachers and their stu-dents was examined for 3rd through 8th grade boys and girls separatelyby gender (see Table 10). Overall, Kappa values indicate that regardlessof gender, teachers showed “slight agreement” or “less than chanceagreement” with their students. One exception was “fair agreement”for teachers and their female students on the item regarding studentsbeing left out or excluded.

4. Discussion

Given the prevalence of bullying and victimization in schools, devel-oping prevention and intervention strategies is critical. One logicalquestion is if the adults in youths' lives are knowledgeable about theirvictimization experiences. In general, students in this study reportedsignificantly more victimization than their parents' reports or theirteachers' reports. Parents reported significantly more victimization fortheir children than did their child's teachers.

Regarding students' reports of their own victimization, no genderdifferences were found and the only grade level difference found wasthat 3rd and 4th grade students reported more victimization than the

Table 9Parent with child percent agreement and Kappa by item level on victim items by gender.

Item Gender Yes–yes agreement No–no agreement Total a

Been bullied in past Boys 6 (12.8%) 36 (76.6%) 42 (89Girls 3 (7.5%) 24 (60.0%) 27 (67

Been called names, made fun Boys 5 (10.4%) 35 (72.9%) 40 (83Girls 4 (9.5%) 29 (69.0%) 33 (78

Been left out, excluded Boys 4 (8.7%) 37 (80.4%) 41 (89Girls 1 (2.4%) 29 (70.7%) 30 (73

Been hit, kicked, pushed, shoved Boys 2 (4.2%) 41 (85.4%) 43 (89Girls 0 (0%) 41 (8.7%) 41 (8.

Others told lies, spread rumors Boys 1 (2.1%) 31 (83.0%) 31 (85Girls 0 (0%) 33 (89.2%) 33 (89

Money taken or things damaged Boys 0 (0%) 46 (93.9%) 46 (93Girls 0 (0%) 42 (95.5%) 42 (95

Been threatened or forced Boys 0 (0%) 46 (93.9%) 46 (93Girls 0 (0%) 42 (97.7%) 42 (97

Called names about looks Boys 4 (8.3%) 39 (81.3%) 43 (89Girls 1 (2.3%) 35 (81.4%) 36 (83

Called names or gestures Boys 3 (6.7%) 37 (82.2%) 40 (88Girls 0 (0%) 35 (85.4%) 35 (85

Been bullied on cell or internet Boys 0 (0%) 46 (93.9%) 46 (93Girls 1 (2.4%) 39 (95.1%) 40 (97

Note. na = data did not allow computation of Kappa.

older students (5th grade and up). This interesting pattern of higherself-reports of victimization at the earliest grades for which we havedata and then dropping off to lower levels may be due to the nature ofthe sample in this small, private school. The majority of the childrenin each grade are together from Kindergarten or 1st grade throughouttheir schooling. Thus, perhaps earlier on in their school career, thestudents are getting to know one another, testing the limits of socialinteractions, along with developmentally learning self-control. Perhapsas they enter the later elementary grades, the students all know oneanother and learn to get along more appropriately, learn to tolerateone another better, or simply grow to appreciate one another so thatbullying is less likely. Because of the closer relationships, studentsmay also interpret bullying behaviors as more playful and less harmful,and therefore not endorse them as behaviors that have happened tothem. Another explanation may be developmental differences in howchildren perceive victimization or bullying. Some research has foundthat younger children tend to perceive any type of victimization as “bul-lying” whereas older children are better able to discriminate betweeninfrequent or isolated incidents of aggression and bullying (Guerin &Hennessy, 2002; Smith, Cowie, Olafsson, & Liefooghe, 2002).

greement Yes–no overestimate No–yes underestimate Total disagreement Kappa

.4%) 3 (6.4%) 2 (4.3%) 5 (10.7%) .64

.5%) 3 (7.5%) 10 (25.0%) 13 (32.5%) .14

.3%) 3 (6.3%) 5 (10.4%) 8 (16.7%) .46

.5%) 3 (7.1%) 6 (14.3%) 9 (21.4%) .34

.1%) 3 (6.5%) 2 (4.3%) 5 (10.8%) .55

.1%) 5 (12.2%) 6 (14.6%) 11 (26.8%) −.004

.6%) 1 (2.1%) 4 (8.3%) 5 (10.4%) .397%) 1 (2.2%) 4 (8.7%) 5 (10.9%) −.04.1%) 3 (6.4%) 4 (8.5%) 7 (14.9%) .14.2%) 2 (5.4%) 2 (5.4%) 4 (10.8%) −.06.9%) 0 (0%) 3 (6.1%) 3 (6.1%) na.5%) 0 (0%) 2 (4.5%) 2 (4.5%) na.9%) 0 (0%) 3 (6.1%) 3 (6.1%) na.7%) 0 (0%) 1 (2.3%) 1 (2.3%) na.6%) 3 (6.3%) 2 (4.2%) 5 (10.5%) .56.7%) 2 (4.7%) 5 (11.6%) 7 (16.3%) .14.9%) 2 (4.4%) 3 (6.7%) 5 (11.1%) .48.4%) 2 (4.9%) 4 (9.8%) 6 (14.7%) −.07.9%) 0 (0%) 3 (6.1%) 3 (6.1%) na.5%) 0 (0%) 1 (2.4%) 1 (2.4%) .66

Table 10Teacher with student percent agreement and Kappa by item level on victim items by gender.

Item Gender Yes–yes agreement No–no agreement Total agreement Yes–no overestimate No–yes underestimate Total disagreement Kappa

Been bullied in past Boys 0 (0%) 22 (78.6%) 22 (78.6%) 3 (10.7%) 3 (10.7%) 6 (21.4%) −.12Girls 1 (4.2%) 15 (62.5%) 16 (66.7%) 3 (12.5%) 5 (20.8%) 8 (33.3%) .00

Been called names, made fun Boys 0 (0%) 26 (72.2%) 26 (72.2%) 3 (8.3%) 7 (19.4%) 10 (27.7%) −.13Girls 3 (8.8%) 20 (58.8%) 23 (67.6%) 3 (8.8%) 8 (23.5%) 11 (32.3%) .16

Been left out, excluded Boys 0 (0%) 51 (86.4%) 51 (86.4%) 2 (3.4%) 6 (10.2%) 8 (13.6%) −.05Girls 2 (4.0%) 41 (82.0%) 43 (86.0%) 3 (6.0%) 4 (8.0%) 7 (14.0%) .29

Been hit, kicked, pushed, shoved Boys 0 (0%) 58 (81.7%) 58 (81.7%) 3 (4.2%) 10 (14.1%) 13 (18.3%) −.07Girls 0 (0%) 56 (90.3%) 56 (90.3%) 0 (0%) 6 (9.7%) 6 (9.7%) na

Others told lies, spread rumors Boys 0 (0%) 33 (94.3%) 33 (94.3%) 1 (2.9%) 1 (2.9%) 2 (5.8%) −.03Girls 1 (2.7%) 31 (83.8%) 32 (86.5%) 3 (8.1%) 2 (5.4%) 5 (13.5%) .21

Money taken or things damaged Boys 0 (0%) 57 (91.9%) 57 (91.9%) 0 (0%) 5 (8.1%) 5 (8.1%) naGirls 0 (0%) 45 (95.7%) 45 (95.7%) 0 (0%) 2 (4.3%) 2 (4.3%) na

Been threatened or forced Boys 0 (0%) 66 (93.0%) 66 (93.0%) 0 (0%) 5 (7.0%) 5 (7.0%) naGirls 0 (0%) 59 (96.7%) 59 (96.7%) 0 (0%) 2 (3.3%) 2 (3.3%) na

Called names about looks Boys 0 (0%) 19 (86.4%) 19 (86.4%) 1 (4.5%) 2 (9.1%) 3 (13.6%) −.07Girls 0 (0%) 18 (66.7%) 18 (66.7%) 4 (14.8%) 5 (18.5%) 9 (33.3%) −.20

Called names or gestures Boys 0 (0%) 16 (76.2%) 16 (76.2%) 2 (9.5%) 3 (14.3%) 5 (23.8%) −.13Girls 0 (0%) 19 (70.4%) 19 (70.4%) 3 (11.1%) 5 (18.5%) 8 (29.6%) −.16

Note. na = data did not allow computation of Kappa.

2099M.K. Demaray et al. / Children and Youth Services Review 35 (2013) 2091–2100

Parents' ratings of their children's victimization did not differ bychild's gender or grade level. Teachers reported similar levels of victim-ization for boys and girls except at the 5th/6th grade level, they ratedgirls as being victimized more frequently than boys.

4.1. Teachers and their students

Teachers and students have been shown generally to have non-matching perceptions of the overall victimization levels in schools(Bradshaw et al., 2007; Houndoumadi & Pateraki, 2001; Stockdale et al.,2002). The current study assessed this specifically on an individual level(each student reporting their own victimization and the teacher ratingperceived victimization for each student) and on an individual itemlevel. The results support past research indicating that teachers andstudents do not share the same perceptions on an individual studentlevel, and may differ in their perceptions of what bullying behavioris (Vaillancourt et al., 2008). As predicted, teachers are typicallyunderestimating the victimization that their students report experienc-ing. No trends were found in the type of bullying influencing theaccuracy of teachers' perceptions. The teachers did not have good agree-ment with their students on any type of victimization. Teachers' ratingsof victimizations across all grade levels, however, were meaningfullyrelated to the girls' self-ratings of victimization, with a significantrelation only at the 3rd and 4th grade levels for teachers and girls. No sig-nificant or meaningful relations were found between teachers' and boys'ratings of victimization at any grade level. Girls may explicitly report vic-timization situations to their teachers more often, teachers may havemore opportunities to observe social interactions in their classrooms, orgirls may be more open about discussing events in the classroom, evenif not directly with the teacher. It is important to note, however, that ingeneral, these relations were small to moderate at best. Teachers hadless than chance agreement with both, the boys and girls in their classes.The only exceptions included slight agreement regarding girls beingcalled names and fair agreement regarding girls being excluded or leftout.

4.2. Parents and their students

Parents and their children have been found to have discrepant per-ceptions of levels of bullying behavior in the schools, but less discrepantreports than between teachers and students (Holt et al., 2009;Houndoumadi & Pateraki, 2001). Data from the current study providefurther evidence of this phenomenon. There was an overall moderaterelation between parents' reports of victimization and those of theirchildren, with a stronger relation between parents and their boys than

between parents and their girls. Regarding grade level, boys and theirparents had moderate relations in their victimization reports at eachof the grade levels with more weak relations between parents andtheir girls at the 3rd/4th grade level.

Examining the agreement data, parents, like teachers, are under-estimating their child's victimization, but to a lesser degree. Parents'reports better match those of their children with most victimizationitem reports having parent child agreement in the fair agreementcategory. This means that parents are agreeing with the level of vic-timization that their children are reporting better than chance wouldpredict, at a fair level. These results are further illuminated whenexamining the agreement data by gender. Parents and their boyshad substantial agreement on levels of overall victimization, andmoderate agreement on several other items. This compares to slightor less than chance agreement for most of the items between parentsand their girls. These data suggest that it is boys in particular thatseem to be talking to their parents about any victimization thatthey are experiencing. Parents are still underestimating, but theyseem to know more than teachers about their boys' social experi-ences. These results are contrary to earlier studies and our predictionthat girls and their parents would agree more than boys and theirparents. One explanation is that boys' victimization may occur onthe playground more often. If a playground monitor sees theseevents, they are referred directly to the principal who willlikely call the parents of the boys involved. Girls' victimization maybe more subtle and they may bring the situation back to the class-room where perhaps the teacher handles it but does not call theparents.

4.3. Parents and teachers

In the current study, parent and teacher ratings of victimizationwere moderately related. When comparing ratings by gender, parentsand teachers had a higher level of agreement for girls than for boys,specifically at the 5th/6th grade level: teachers and parents showedstronger agreement about their report of girls' victimization than theydid about their report of boys' victimization at this grade level. One ex-planation for teachers not agreeing with students, students agreeingwith parents, but having some agreement between teachers andparents may be that some of the ratings adults do may have more todo with an overall perception of a child and how much they believethat child “would be” or “could be” victimized versus rating exactfrequency counts of bullying incidents. Perhaps the adults can agreeon a general tendency, but parents and students are more accuratelyagreeing at a detailed level.

2100 M.K. Demaray et al. / Children and Youth Services Review 35 (2013) 2091–2100

4.4. Limitations

While the current study clearly benefits from examining multi-informant data, there are some important limitations to consider. Thedata were collected from a small private school where the same groupsof students are together year after year. This is a unique social experi-ence where cohort effects may overshadow real developmental trends(e.g., this group of 4th graders may not be representative of 4th gradersin general). Due to the number of students enrolled at the school, thesample size itself was small. However, this was a unique and specializedsample in that the researchers obtained full participation of students,their teachers, and the majority of their parents. Furthermore, themajority of the students involved were white (86%) and, since theschool is a private Catholic school, only 6% of students were non-Catholic. All teachers participating in the study were white and female.This lack of diversity may result in limited generalizability of results.Finally, the unique data analyses utilized in the study required parentsto provide their child's name on the surveys, thus surveys were notanonymous and social desirability may have influenced responses.Additional research on victimization with larger, more diverse samplesexamining convergent and divergent data across sources may be animportant next step in the literature on rater agreement.

4.5. Implications

The results of the current study have important implications for bullyprevention and early intervention services. While it is clear that parentsaremore aware of their child being victimized than teachers, both parentsand teachers underestimated the extent that victimization occurs atschool. This was especially apparent regarding parents and boys. Itcould be that the type of victimization that boys are more likely to expe-rience may occur outside of the classroom (e.g., cafeteria, playground)where an administrator is more likely to intervene and call parents.

The current investigation helps researchers and school practitioners tounderstand patterns in discrepant viewpoints. An important next stepwould be to investigate reasonswhy these discrepancies exist. It is impor-tant that teachers, students, and parents have open lines of communica-tion and that all adults in the child's life are notified when victimizationoccurs. While perfect agreement among sources may not be the goal,accurate perceptions of victimization are imperative for effective develop-ment, implementation, and evaluation of bullying prevention and inter-vention programs. Research suggests that the most effective bullyprograms are those that take a multi-systematic perspective and requirecollaboration among school staff, students, and parents (Holt et al.,2009). These programs may not be as effective if school staff and parentsare not aware of the extent of bullying occurring at the school. Future re-search should investigate ways that will encourage children to talk to theadults in their life about their experienceswith victimization so that prop-er intervention services can be put in place.

References

Bauman, S., & Del Rio, A. (2006). Preservice teachers' responses to bullying scenarios:Comparing physical, verbal, and relational bullying. Journal of EducationalPsychology, 98, 219–231, http://dx.doi.org/10.1037/0022-0663.98.1.219.

Bosworth, K., Espelage, D., & Simon, T. (1999). Factors associated with bullying behavior inmiddle school students. The Journal of Early Adolescence, 19, 341–362, http://dx.doi.org/10.1177/0272431699019003003.

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

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.)Hillsdale,NJ: Lawrence Erlbaum Associates.

Cook, C., Williams, K., Guerra, N., & Kim, T. (2010). Variability in the prevalence of bullyingand victimization: A cross-national and methodological analysis. In S. R. Jimerson,S. M. Swearer, & D. L. Espelage (Eds.), Handbook of bullying in schools: An internationalperspective (pp. 347–362). New York, NY: Taylor & Francis.

Demaray, M., & Malecki, C. (2003). Perceptions of the frequency and importance of socialsupport by students classified as victims, bullies, and bully/victims in an urbanmiddle school. School Psychology Review, 32, 471–489.

Elinoff, M., Chafouleas, S., & Sassu, K. (2004). Bullying: Considering for defining and inter-vening in school settings. Psychology in the Schools, 41, 887–897, http://dx.doi.org/10.1002/pits.20045.

Espelage, E., & Swearer, S. (2010). A social-ecological model for bullying prevention andintervention: Understanding the impact of adults in the social ecology of youngsters.In S. R. Jimerson, S. M. Swearer, & D. L. Espelage (Eds.),Handbook of bullying in schools:An international perspective (pp. 347–362). New York, NY: Taylor & Francis.

Fisher, R. A. (1921). On the “probable error” of a coefficient of correlation deduced from asmall sample. Metron, 1–32.

Guerin, S., & Hennessy, E. (2002). Pupils' definitions of bullying. European Journal ofPsychology of Education, 17(3), 249–261.

Hawker, D., & Boulton, M. (2000). Twenty years' research on peer victimization andpsychosocial maladjustment: A meta-analytic review of cross-sectional studies.Journal of Child Psychology and Psychiatry, 41, 441–455, http://dx.doi.org/10.1111/1469-7610.00629.

Holt, M., Kantor, G., & Finkelhor, D. (2009). Parent/child concordance about bullying in-volvement and family characteristics related to bullying and peer victimization.Journal of School Violence, 8, 42–63, http://dx.doi.org/10.1080/15388220802067813.

Houndoumadi, A., & Pateraki, L. (2001). Bullying and bullies in Greek elementary schools:Pupils' attitudes and teachers'/parents' awareness. Educational Review, 53, 19–26,http://dx.doi.org/10.1080/00131910120033619.

Hunter, S., Boyle, J., &Warden, D. (2007). Perceptions and correlates of peer-victimizationand bullying. British Journal of Educational Psychology, 77, 797–810, http://dx.doi.org/10.1348/000709906X171046.

Lee, T., & Cornell, D. (2010). Concurrent validity of the Olweus Bully/VictimQuestionnaire.Journal of School Violence, 9, 56–73, http://dx.doi.org/10.1080/15388220903185613.

Lopez, C., & Dubois, D. (2005). Peer victimization and rejection: Investigation of an inte-grative model of effects on emotional, behavioral, and academic adjustment inearly adolescence. Journal of Clinical Child and Adolescent Psychology, 34, 25–36,http://dx.doi.org/10.1207/s15374424jccp3401_3.

Mishna, F., Scarcello, I., Pepler, D., & Wiener, J. (2005). Teachers' understanding of bullying.Canadian Journal of Education, 28, 718–738 (Retrieved from http://www.csse-scee.ca/CJE/)

Nansel, T., Overpeck, M., Pilla, R., Ruan, W., Simons-Morton, B., & Scheidt, P. (2001). Bullyingbehaviors among U.S. youth: Prevalence and associations with psychosocial adjustment.Journal of the American Medical Association, 285, 2094–2100, http://dx.doi.org/10.1001/jama.285.16.2094 (and Psychiatry, 35, 1171–1190. doi: http://dx.doi.org/10.1111/j.1469-7610.1994.tb01229.x)

Olweus, D. (1996). The revised Olweus Bully/Victim Questionnaire. Bergen, Norway:Mimeo, Research Center for Health Promotion (HEMIL), University of Bergen.

Olweus, D. (2010). Understanding and researching bullying. In S. R. Jimerson, S. M.Swearer, & D. L. Espelage (Eds.), Handbook of bullying in schools: An internationalperspective (pp. 9–33). New York, NY: Taylor & Francis.

Paul, J., & Cillessen, A. (2007). Dynamics of peer victimization in early adolescence: Re-sults from a four-year longitudinal study. In J. E. Zins, M. J. Elias, & C. A. Maher(Eds.), Bullying, victimization, and peer harassment (pp. 29–48). Binghamton, NY:The Haworth Press, Inc.

Pervin, K., & Turner, A. (1994). An investigation into staff and pupils' knowledge, atti-tudes, and beliefs about bullying in an inner city school. Pastoral Care in Education,12, 16–22 (Retrieved fromhttp://www.napce.org.uk/)

Rigby, K. (1993). School children's perceptions of their families and parents as a functionof peer relations. The Journal of Genetic Psychology, 154, 501–513 (Retrieved fromhttp://www.tandf.co.uk/journals/titles/00221325.asp)

Smith, P. K., Cowie, H., Olafsson, R. G., & Liefooghe, A. P. D. (2002). Definitions of bullying:A comparison of terms used, and age and gender differences, in a fourteen-countryinternational comparison. Child Development, 73(4), 1119–1133.

Solberg, M., & Olweus, D. (2003). Prevalence estimation of school bullying with the OlweusBully/Victim Questionnaire. Aggressive Behavior, 29, 239–268, http://dx.doi.org/10.1002/ab.10047.

Stockdale, M., Hangaduambo, S., Duys, D., Larson, K., & Sarvela, P. (2002). Rural elementa-ry students', parents', and teachers' perceptions of bullying. American Journal of HealthBehavior, 26, 266–277 (Retrieved from http://www.ajhb.org/)

Sveinsson, A., & Morris, R. (2007). Conceptual and methodological issues in assessmentand intervention with school bullies. In J. E. Zins, M. J. Elias, & C. A. Maher (Eds.),Bullying, victimization, and peer harassment (pp. 9–28). Binghamton, NY: The Haw-orth Press, Inc.

Swearer, S., Siebecker, A., Johnsen-Frerichs, L., & Wang, C. (2010). Assessment ofbullying/victimization: The problem of comparability across studies and acrossmethodologies. In S. R. Jimerson, S. M. Swearer, & D. L. Espelage (Eds.), Handbook ofbullying in schools: An international perspective (pp. 305–328). New York, NY: Taylor& Francis.

Vaillancourt, T., McDougall, P., Hymel, S., Krygsman, A., Miller, J., Stiver, K., et al. (2008).Bullying: Are researchers and children/youth talking about the same thing?International Journal of Behavioural Development, 32(6), 486–495.

Viera, A. J., & Garrett, J. M. (2005). Understanding interobserver agreement: The Kappastatistic. Family Medicine, 37, 360–363 (Retrieved from http://www.stfm.org/fmhub/)

Waasdorp, T., Pas, E., O'Brennan, L., & Bradshaw, C. (2011). A multilevel perspectiveon the climate of bullying: Discrepancies among students, school staff, andparents. Journal of School Violence, 10, 115–132, http://dx.doi.org/10.1080/15388220.2010.539164.