Validation of the Urdu Version of Olweus Bully/Victim Questionnaire (OBVQ)

Journal:
Jan-March 2015 Volume 12(1)

Original Article

Author(s):
Rabia Khawar, Farah Malik, Syeda Shahida Batool
Page No:
24

Validation of the Urdu Version of Olweus Bully/Victim Questionnaire (OBVQ)

Rabia Khawar

Corresponding Author:

Rabia Khawar

                                    Department of Psychology, GC University, Katchehry Road,, 54000 Lahore, Pakistan.

                                    Department of Applied Psychology, GC University Allama Iqbal Road, 38000  Faisalabad, Pakistan.

E-mail - khawarthisend@gmail.com phone: 0923216996017

 

Abstract

Objective:       The study intended to validate the Urdu version of Revised Olweus Bully Victim Questionnaire (OBVQ) in our local settings.

Design:           Exploratory study

Place and Duration of Study: The study was conducted in different public and private schools of Lahore during 2012 to 2014.

Participants and Methods:               Participants were 817 students from grades four to six of 12 public and private schools of Lahore. Additionally, 106 teachers participated in bullying nomination of students. Purposive convenient sampling technique was used.

Results:           Confirmatory factor analysis resulted in a good model fit with acceptable values on indices and parameter estimates supporting the construct validity of the instrument. The reliability coefficients for bullying and victimization were comparable to international studies. Significant relationship between the OBVQ 8 types of victimization and bullying and two corresponding global items demonstrated concurrent validity. Children were involved in bullying roles with larger proportion in bully-victim group. OBVQ-Urdu version demonstrated concordance with the teacher nominations of bullying.

Conclusion:    The Urdu version of OBVQ has good psychometric properties and is appropriate for measuring the involvement of Pakistani school students in a variety of bullying behaviors.

 

 

Introduction

During the last two decades, bullying has been emerged as serious problem affecting many students in schools throughout the world1, 2. Researchers have distinguished bullying from ordinary fight or quarrelling between two parties of equal strength by emphasizing the notion of power imbalance. It’s a combination of certain direct and indirect negative actions performed by one or more students repeatedly over time with an intention to harm weaker students. Victims’ inability to defend him or her is another important feature of Olweus definition3, 4. Bullying can take several forms such as physical (e.g. hitting, punching, kicking, and beating up, pushing, spitting, property damage, and/or theft), verbal (e.g. teasing, mocking, name-calling, nasty comment and humiliation, threats and noxious remarks, extortion, and/or racist, sexist or homophobic insults), social (e.g. gossip, rumor spreading, embarrassment, alienation or exclusion from the group) or cyber (e.g. using the cell phone or internet i.e. email, text messaging, social media)5. All types of bullying causes some sort of psychological damage. Researchers have asserted that bullying may result in academic failure, health issues, loss of social relations and a variety of cognitive, affective and behavioral problems6, 7. Perpetrators of bullying are exposed to disrupted academic and social life at school that may result in delinquency and conduct problems in later life8

. Depending on the nature and design of the study, the participants’ age and the region, the proportion of school children repeatedly victimized ranges from 7.1 to 70.2%.9

The most commonly used measure to assess bully/victim status conflicts is the Revised Olweus Bully/Victim Questionnaire10. Though used widely by researchers and educators in different countries, information about the psychometric properties of OBVQ is not well documented11.  The measure provides two main dimensions (being bullied and bullying others) for classifying students as bullies, victims, bully/victims or uninvolved. Studies have provided evidence for the construct validity of these two dimensions 12,13. One of the most comprehensive analysis using Rasch model supported the conceptual design of both dimensions with acceptable psychometric elements for each scale14. Some recent researches have also shown adequate reliability and validity for the instrument when translated into another language (e.g. Greek) 15,16.

This the first attempt to validate the construct validity of Revised Olweus Bully Victim Questionnaire-Urdu version by using confirmatory factor analysis for the two dimensions, examining the classification of children as bullies, victims, or bully/victims and to determine its convergent validity with teacher nominations of children’s role in bullying.

Method

 Participants

The sample consisted of 817 students studying in 4th, 5th and 6th grades of both public and private sector schools of Lahore. Girls (n = 440) and boys (n = 377) with age range between 9 to12 years (M = 10.86, SD = 1.08) were selected from 12 schools that agreed to participate in the study. A total of 106 teachers participated in the nomination procedure (Mage = 29.79, SD age = 4.8). All the teachers had been teaching the same class for at least last four months.

Measures

Revised Olweus Bully-Victim Questionnaire-Urdu Version

Revised Olweus Bully Victim Questionnaire distinguishes among students for their involvement bullying roles with the help of two global questions i.e. item No. 4 (being bullied) and 24 (bullying others) along with 8 types, each answered on a 5 point scale with ‘‘I haven’t been bullied/bullied other students at school in the past couple of months,’’ coded as 1 and “several times a week” coded as 5. The questionnaire had shown good internal consistency (Cronbach’s α = .80 to .90) and can be easily administered in a group setting17. The present study utilized an Urdu version of the OBVQ developed using standardized translation procedures. Initial empirical investigation of the Urdu version had resulted in acceptable psychometric properties with being bullied and bullying others dimensions yielding Chronbach Alpha of .79 and .82 respectively. The current study further evaluated the reliability and validity of the two sets of items measuring bullying and victimization with a larger sample.

 

Teacher Nominations of Students’ Bullying Roles

A brief measure was developed using definition of bullying from Revised Olweus Bully Victim Questionnaire. Teachers were instructed to read the definition of bullying carefully followed by the description of all the bullying roles: bully, victim, bully/victim and uninvolved. They were then presented with the list of randomly selected students from their classes and were asked to assign them to one of the four bullying roles based on their past three months observation of student’s behavior and complaints reported by peers.

Results

Confirmatory Factor Analysis

To determine the construct validity of the set of items assessing being bullied and bullying others status and types in Revised OBVQ -Urdu, we performed a confirmatory factor analysis (CFA) using Analysis of Moment Structures (AMOS), Version 20.0. A poor fit was suggested by the chi square value being significant, χ2 (103) = 334.57, p < .05. Yet the CMIN/df value (CMIN/DF = 3.24) was found approximately acceptable.  Rest of the indices that have been considered for assessing the degree of fit between the hypothesized two factor model and the sample resulted in a good model fit.

Tucker Lewis Index and the Comparative Fit Index were found to be acceptable having values .93 and .94 respectively (TLI & CFI: >.90 acceptable, >.95 excellent18. Root Mean Square Error of Approximation yielded a score of .05 that had been considered excellent 19.

 

Factor loadings for victimization scale range from .59 to .71, and items on bullying scale also showed high loadings ranging between .54 and .65.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1

Confirmatory Factor Analysis for OBVQ-Urdu Version

 

 

 

Reliability Analysis

Internal consistency reliability of OBVQ-Urdu version, based on Cronbach’s Alpha, was also uniformly high across gender and grade level.

Table 1

Reliability Coefficients for OBVQ-Urdu by Gender and Grade

 

Victimization

Bullying

Total

 

Boys

 

 

.83

 

.82

 

.85

Girls

.84

.83

 

.87

4th

.83

 

.83

 

.85

5th

.86

 

.81

 

.85

6th

.82

 

.82

 

.86

Total

.84

.82

 

.86

 

Furthermore, strong relationships between the 8 types of victimization and bullying, and OBVQ global items for being bullied (r = .40 to r =.65) and bullying others (r = .35 to r =.70) demonstrated adequate concurrent validity respectively with alpha significance at p < .001.

Bully-Victim Status Types

Being bullied

With regard to victims of bullying, about one third of the students (28%) reported not being bullied by others in the past couple of months, 25% reported being bullied “once or twice,” 25.5% reported being bullied “two or three times a month,” 9.9% reported being bullied “about once a week,” and 11.6% reported being bullied once a week “several times a week.” A large number of students (71.9%) reported being bullied; using milder criterion of “only once or twice”, and the standard Olweus cut-off classified nearly 47% students as victims of perpetration.

Bullying others

On the OBVQ, the majority of students (45.9%) reported not bullying others in the past couple of months, about 30% reported bullying others “once or twice,” 17% reported bullying others “two or three times a month,” and 9.1% reported bullying others “about once a week”, while 4.2 % bullied others “several times a week.”  Following the “at least once” criterion. More than half students in the sample (54.1%) were found involved in bullying others, and the Olweus standard criterion identified only 34.23% students reporting bullying others.

Finally, we identified children that were uniquely bullies (n = 141), uniquely victims (n = 158), and bully-victims (n = 235). The rest were grouped together as uninvolved (n = 283). To serve this purpose, we considered (1) the cases of the children that were involved in bullying others for “two to three times a month or more” on global item of bullying and at least one of the 8 subsequent forms of bullying, but had not been victims of bullying (2) the cases of the children that were bullied two to three times a month as measured by global item of victimization along with at least one of the 8 ways, but had not been involved in bullying others, and (3) the cases of the children that had been bullying others and been bullied two to three times a month or more on both global items and at least one of the 8 corresponding items for bullying and victimization. The results of these calculations show that only 35% of the students remain uninvolved, 17.3% of the students were uniquely bullies, 19.3% of the children were uniquely victims, and 28.8% were bully-victims. In total, 65.4% of the Pakistani school students have been found to be involved in bullying and victimization.

Relationship between OBVQ Results and Teacher Nominations

Data from the teacher estimates of students’ involvement in bullying served as a check on the accuracy of the self-report. The estimated number of class bullies (and victims) was compared to the total number of students who reported being bullies (and victims).

Table 2

Bullying Role Classification Following OBVQ and Teacher Nomination Criteria

Teacher Nominations

OBVQ Bullying Status

 

 

Bully

Victim

Bully-Victim

Uninvolved

Total

Bully

132

11

26

18

187

Victim

1

136

7

11

155

Bully-Victim

5

6

198

13

222

Uninvolved

3

5

4

241

253

Total

141

158

235

283

817

 

Total of 87 % of the original grouped cases as a result of self-report (OBVQ-Urdu) has been correctly classified as evidenced by teacher nominations. Cohen’s κ was conducted to determine the level of agreement between teacher nominations and self-reported bullying classification for 817 students. Kappa showed high correspondence between the two classification, κ = .82, p < .001.

The majority of the students (95.3%) labeled as uninvolved as a result of OBVQ received the same status in teacher nominations. High level of agreement (89.1%) was found between teacher nominations of students as bully-victim and self-reports. Most of the self-reported victims were accurately classified (87.7%) by teachers into the same role. Although, substantial number of students nominated as bullies by teachers acknowledged their role in self-report (70.6%), some of them were identified as bully-victims (13.9%), and victims (5.9%). A few (9.6%) reported them as uninvolved.

Subsequently, we performed multinomial logistic regression that can be used to observe the effects of independent variables on a nominal dependent variable, such as the teacher nominated groups of bullies, victims, bully/victims, and uninvolved children in our case. Two global items of OBVQ assessing bullying and victimization were added as covariates.

Table 3

Multinomial Logistic Regression Predicting Accuracy of Self-reported Bullying and Victimization

TN                     OBVQ

 

      B           SE

Wald

OR

    95% CI

L

U

Bully