# New ITEM response models: application to school bullying data

## TL;DR

This paper introduces new item response theory models tailored for school bullying data, accounting for the bounded nature of victimization and incorporating covariates to analyze individual and school effects.

## Contribution

It proposes novel IRT logistic models with bounded latent variables and an empirical anchor procedure, extending analysis capabilities for bullying data.

## Key findings

- Models effectively capture bullying victimization levels.
- Joint inference on individual and school factors achieved.
- Application demonstrates model utility on real data.

## Abstract

School bullying victimization is a variable that cannot be measured directly. Taking into account that this variable has a lower bound, given by the absence of bullying victimization, this paper proposes IRT logistic models, where the latent parameter ranges from $0$ to $\infty$ or from $0$ to a positive real number R, defining the IRT parameters and proposing an empirical anchor procedure. As the academic abilities and the school bullying victimization can be explained due to associated factors such as habits, sex, socioeconomic level and education level of parents, IRT regression models are proposed to make joint inferences about individual and school characteristic effects. Results from the application of the proposed models to the Bogot\'a school bullying dataset are presented. The need for testing based in statistical models increases in different fields.