|
Abstract
|
OBJECTIVES: To evaluate CADRI's configural and metric structures and explore its performance from the perspective of the respondents as perpetrators and victims. In the process, we present shorter versions for both roles.
METHODS: The sample consisted of 561 adolescents aged 15 and 19, enrolled in public and private schools in Rio de Janeiro, Brazil. Initially, confirmatory factor analyzes (CFA) tested the originally proposed dimensional structure. Since this model was rejected, the second step analyzed the data via Exploratory Structural Equation Models (ESEM) to reassess the instrument's structure. Applying a decision algorithm on a new dimensional structure, we proceeded to reduce the number of items to establish separate sets for victimization and perpetration. These final models were then analyzed via CFA to assess their psychometric properties.
RESULTS: The final solutions contained 15 items comprising four and three factors for victimization and perpetration, respectively. Thirteen items were common to both models, but two were unique to their respective sets. Regardless of their small differences, the final solutions fitted adequately, held factorial item pertinence and unambiguity, contained reliable and non-redundant items, and sustained factor-based convergent and discriminant validities.
CONCLUSION: The four and three-factor models were within the dimensional bounds proposed in the original CADRI, the shorter scales still succeeding in partially capturing what was envisaged in the first version of the instrument. However, the shorter versions suggested here are still tentative, thus requiring further examination.
Language: en |