The present study was set up with two aims in mind. First, to assess the psychometric characteristics of a peer-report measure of bullying in a Dutch sample, and second, to compare relative and absolute ways of assigning roles in the bullying process. The sample consisted of 242 children (51% boys; mean age approximately 10 years) at T1. Two years later, there were 247 children (49% boys).
We made use of an adaptation of the original Participant Role Scales (PRS) Salmivalli, Lagerspetz, Björkqvist et al., 1996; Salmivalli, Lappalainen and Lagerspetz, 1998) and of the Aggression and Victimization Scale [Perry, Kusel and Perry, 1988]. This scale, called the New PRS, consisted of 32 items in total. Five highly reliable scales were distinguished with the help of CFA, one for leader-like bullying behavior, one for follower-like bullying behavior, and scales for outsider, defender and victim. We computed the roles according to four criteria, that is, z-scores (a relative measure) and three different percentage scores (10%, 15% and 20%; each an absolute measure). Sociometric status was also assessed. The concordance between the various methods was moderate. Test–retest stability was also moderate. Test–retest coefficients for the scale scores were considerably higher. Links were found between the roles and sociometric status, irrespective of the method used for assigning roles. Victims were rejected, as were bullies, but defenders were popular. Gender differences indicated that boys were more often bullies or followers, and girls were more often outsiders or defenders. Absolute methods for assigning bullying roles produced fewer unclassifiable children and brought to light substantial differences between classes. Aggr. Behav. 32:343–357, 2006. © 2006 Wiley-Liss, Inc.