Publication: El efecto de regresión a la media en evaluaciones longitudinales a través de un estudio de simulación Monte Carlo
Loading...
Files
Official URL
Full text at PDC
Publication Date
2019-05-27
Authors
Advisors (or tutors)
Castro Morera, Maria
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad Complutense de Madrid
Citations
Exportar
Abstract
Las evaluaciones educativas cada vez están adquiriendo una mayor importancia como medio de control y mejora de la calidad de los sistemas educativos. A nivel metodológico existen diferentes tipos de evaluaciones, de estatus, basadas en un único momento temporal, evaluaciones de ganancia, caracterizadas porque se estudia el cambio en términos de rendimiento entre dos mediciones realizadas a un mismo estudiante, y evaluaciones basadas en modelos de crecimiento, que son las más adecuadas, pues el aprendizaje implica una evolución a lo largo del tiempo, y, por lo tanto, debería analizarse esa trayectoria académica de un mismo estudiante en diferentes ocasiones de medida. A pesar de ser esta última evaluación la más idónea por la gran información que proporciona, es escaso el número de evaluaciones existentes de este tipo, debido a la complejidad y el coste que suponen, y menor es aún el número de estudios que analizan la fiabilidad existente en la estimación de las puntuaciones de los estudiantes obtenidos a lo largo del tiempo, así como la incidencia de posibles artefactos estadísticos. Uno de los destacados, propio de este tipo de diseños longitudinales, y que se aborda en este trabajo es el Efecto de Regresión a la Media...
Educational assessments are acquiring increasing importance as a means of monitoring and improving the quality of education systems. On the methodological level, there are different types of assessment: status assessments, which are based on a single point in time; progress assessments, which study the change in terms of performance between two measurements applied to the same student; and assessments based on growth models, which are the most suitable, because if learning implies an evolution, then a student’s academic progress should be analysed over a period which encompasses more than two measurements.Despite the fact that this latter assessment is the most suitable due to the vast amount of information which it provides, very few assessments of this type exist due to their complexity and cost, and there are even fewer studies which analyse the reliability of estimates of student scores obtained over time or the impact of possible statistical artefacts. One of the most notable of these statistical artefacts is the Regression-to-the-Mean Effect, which is typical of this type of longitudinal design and is addressed in this work...
Educational assessments are acquiring increasing importance as a means of monitoring and improving the quality of education systems. On the methodological level, there are different types of assessment: status assessments, which are based on a single point in time; progress assessments, which study the change in terms of performance between two measurements applied to the same student; and assessments based on growth models, which are the most suitable, because if learning implies an evolution, then a student’s academic progress should be analysed over a period which encompasses more than two measurements.Despite the fact that this latter assessment is the most suitable due to the vast amount of information which it provides, very few assessments of this type exist due to their complexity and cost, and there are even fewer studies which analyse the reliability of estimates of student scores obtained over time or the impact of possible statistical artefacts. One of the most notable of these statistical artefacts is the Regression-to-the-Mean Effect, which is typical of this type of longitudinal design and is addressed in this work...
Description
Tesis de la Universidad Complutense de Madrid, Facultad de Educación, Departamento de Investigación y Psicología en Educación, leída el 30-11-2018