Using Three Reporters to Identify Pre-Adolescent Peer Victims through Latent Profile Analysis

• In Event: Poster Session 05
  In Poster Session: PS 05 Topic - Social Relationships

Abstract

Introduction: A challenge of studying peer victimization is identifying who may be a victim. Several studies have identified victims by combining self and peer reports using Latent Profile Analysis (LPA; e.g., Scholte et al., 2013); however, no studies have also included teacher reports or applied multi-level modeling. The goals of the current study were to use a three-reporter methodology and multi-level LPA to: (a) classify victim groups; (b) evaluate the stability of victim groups over one school year; and (c) examine differences among victim groups’ adjustment.

Method: Participants were 1440 racially/ethnically diverse 4th- and 5th-grade children in 74 classrooms in nine elementary schools (Mage = 10.15; 50% female). At the beginning (T1) and end (T2) of the school year, children and teachers completed the Peer Victimization Scale (Austin & Joseph, 1996), and peers completed the nomination item “Who gets hit by other kids or has mean things said to them by other kids?”. Further, at each time point, we collected data on the adjustment constructs of depression and anxiety using self and teacher report, as well as aggression and peer rejection using peer and teacher report. Across time points, Cronbach’s alpha ranged from adequate (α = .77) to excellent (α = .94) for self-report surveys.

Results: We conducted a multi-level LPA to determine the number of groups that emerged at each time point with self-, teacher-, and peer-reported victimization as indicators. At T1, two groups emerged: non-victims (low across all reporters) and victims (high across all reporters). At T2, four groups emerged: non-victims (low across all reporters), moderate victims (moderate across all reporters), discordant high victims (high on self report, very high on peer report, moderate on teacher report), and concordant high victims (high across all reporters). Follow-up chi-square tests of independence revealed that groups characterized by more victimization were more likely to be composed of male and non-European-American children.

Using a chi-square test of independence, we examined the relation of children’s placement in the victim groups from T1 to T2. The stability of victim groups between time points was largely driven by non-victims; T1 victims dispersed fairly evenly across the four groups at T2 (Table 1).

While accounting for gender and race/ethnicity, we ran a series of ANCOVAs to test the differences between victim groups at each time point on all adjustment variables (Table 2). At T1, non-victims fared better on all adjustment constructs according to each informant. At T2, the three victim groups increased in maladjustment from moderate victims to discordant high victims to concordant high victims. Specifically, all teacher-report variables followed the same pattern, whereby maladjustment increased alongside victimization. Self- and peer-report variables, with the exception of self-reported anxiety, followed this same pattern, although discordant high victims often did not differ from either moderate victims or concordant high victims.

Discussion: These findings support the use of three-reporter assessment and a multi-level LPA approach to identify children victimized by their peers. Implications of these findings in the context of school bullying prevention efforts will be discussed.

Authors

• Zachary Mark Meehan, University of Delaware
  Presenting Author

• Julie Ann Hubbard, University of Delaware
  Non-Presenting Author

• Stevie N Grassetti
  Non-Presenting Author

• Marissa A. Docimo, Core Human Factors
  Non-Presenting Author

• Lauren Elizabeth Swift, University of Delaware
  Non-Presenting Author

• Megan Bookhout, University of Richmond
  Non-Presenting Author