Uso de regressão isotônica suavizada na detecção de funcionamento diferencial do item
This project aims to enrich the Differential Item Functioning (DIF) study. In this way, a bibliographical review of some of the existing methods for detection of items with DIF was done and a non-parametric method, which estimates the item characteristic curves of the two compared groups through iso...
Guardado en:
| Autor principal: | |
|---|---|
| Otros Autores: | |
| Formato: | Dissertação |
| Lenguaje: | portugués |
| Publicado: |
Universidade de Brasília
2024
|
| Materias: | |
| Acceso en línea: | https://hdl.handle.net/20.500.14135/965 |
| Etiquetas: |
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | This project aims to enrich the Differential Item Functioning (DIF) study. In this way, a bibliographical review of some of the existing methods for detection of items with DIF was done and a non-parametric method, which estimates the item characteristic curves of the two compared groups through isotonic regression, was proposed. The presented method was compared with other methods already developed to detect the DIF items, Mantel-Haenzel and logistic regression. I made a simulation study with 100 replications for each test configuration. The settings vary according to the size of DIF, the type of DIF - uniform or non-uniform - the sample size and the proportion of examinees in each group. Considering the results of the simulations, the comparison consisted in observing the proportion of DIF detection in items that in fact presented DIF and the proportion of false DIFs indicated by the methods. In addition, an analysis was made of the impact of these items that presented DIF in a test in the estimation of the proficiency of the examinees. |
|---|