Testing Comparability of Gender Bullying Victimization Assessments Through a Latent Class Approach

Abstract

Bullying is a social and health problem that requires appropriate interventions based on valid and fair evaluations of bullying experiences. The validity of interpretations of bullying victimization scores can be compromised by measurement artifacts or biases that may arise during the assessment process. Boys’ and girls’ bullying experiences could lead to differences in their response processes when they answer the bullying scale items and compromise validity if such differences come from measurement artifacts. The study is intended to illustrate how to obtain validity evidence of response processes for the Students’ Experience of Being Bullied Scale from the Programme for International Student Assessment (PISA) 2018 Student Questionnaire and test measurement invariance across gender by latent class analysis (LCA). The sample was taken from the PISA 2018 study and consisted of 11,599 Spanish high school students (50.3% female, 49.7% male). Response patterns were examined through LCA. Four profiles were found: (a) Not Bullied, (b) Bullied (All Types), (c) Relational Bullying, and (d) Potential Friendly Teasing. Measurement invariance across gender was analyzed by a multigroup LCA. LCA results do not guarantee equivalence of measurements. Class prevalence per group and a multinomial logistic regression were calculated to further examine gender differences across latent classes. Boys were more likely to belong to the Bullied (All Types) class and the Potential Friendly Teasing class, whereas girls were more likely to belong to the Relational Bullying class. These findings illustrate the possibilities LCA can offer to provide validity evidence of response processes and suggest that different bullying experiences of girls and boys could compromise the fairness and validity of comparative interpretations of test scores, which is an important issue to address to develop bias-free classroom interventions.