Predictores asociados a variaciones en puntajes Simce en la región del Biobío
Predictores asociados a variaciones en puntajes Simce en la región del Biobío
Authors
Orellana Olivares, Ricardo
Merino Escobar, José Manuel
Merino Escobar, José Manuel
Authors
Date
2014-09-07
Datos de publicación:
10.7770/CUHSO-V23N1-ART445
Keywords
Análisis estadístico de regresión lineal múltiple - Región del Biobío - Prueba Simce - Modelo óptimo trivariado - Simce
Collections
Abstract
El creciente interés por contar con investigaciones que expliquen
diferencias en resultados académicos en el Sistema de Medición de los Resultados
del Aprendizaje (Simce) que vayan más allá de la mera construcción
de un «ranking educativo», es el fundamento del propósito principal
de este estudio que busca identificar posibles predictores de tipo social, cultural
o económico, que incidan de manera estadísticamente significativa en
variaciones en los resultados de la prueba mencionada aplicada el año 2009
a Cuarto básico. La población comprendida corresponde a la totalidad de
la región del Biobío, Chile. Como esta población resulta demasiado extensa
para el análisis de nivel individual, se trabajó con las medias comunales del
Simce en las 54 comunas que comprenden la mencionada región. El método
de recolección de datos consistió en una búsqueda exhaustiva de la mayor
cantidad de predictores disponibles en bases de datos confiables como el
Censo, Casen, Sinim y Simce. Los datos obtenidos fueron ordenados, analizados
y correlacionados a través del programa de análisis estadístico SAS.
El análisis estadístico de regresión lineal múltiple permitió establecer un modelo
óptimo trivariado para la variable dependiente, conformado por las
variables independientes: evaluación docente, crecimiento anual comunal de
la población y porcentaje de acceso a computador. Finalmente, este modelo explicó un 57% de la variación total de los resultados Simce 2009 con un
error alfa de 1 por 10.000.
The increased interest of having investigations that explain academic differences on the academic results in the system of measuring the quality of education (Simce) beyond of a simple makeup of «educative ranking », emerges as a purpose of this study to identify possible predictors of social, cultural and economic type, that impact in a statistically significative way on variations of results of the Simce evaluation applied in 2009 to the 4 grade of elementary level. The population studied was the total of the Biobío region, Chile. As this population is too large to be analyzed, it was decided to work with the average Simce score at the 54 communes-level which belong to the Biobio region. The data collection method consisted on an exhausted research of the majority of the variables available in the aggregated data such as Censo, Casen, Sinim and Simce. The obtained results were organized, analyzed and correlated, across the program of statistical analysis labeled SAS. Multiple linear regression analysis it was possible to establish an optimal model composed by three independent variables: teachers evaluation, annual increased of the population and percentage of access to computer. Finally, this model, explained 57% out of total variations of the 2009 Simce results, with an alpha error of 1 per 10.000.
The increased interest of having investigations that explain academic differences on the academic results in the system of measuring the quality of education (Simce) beyond of a simple makeup of «educative ranking », emerges as a purpose of this study to identify possible predictors of social, cultural and economic type, that impact in a statistically significative way on variations of results of the Simce evaluation applied in 2009 to the 4 grade of elementary level. The population studied was the total of the Biobío region, Chile. As this population is too large to be analyzed, it was decided to work with the average Simce score at the 54 communes-level which belong to the Biobio region. The data collection method consisted on an exhausted research of the majority of the variables available in the aggregated data such as Censo, Casen, Sinim and Simce. The obtained results were organized, analyzed and correlated, across the program of statistical analysis labeled SAS. Multiple linear regression analysis it was possible to establish an optimal model composed by three independent variables: teachers evaluation, annual increased of the population and percentage of access to computer. Finally, this model, explained 57% out of total variations of the 2009 Simce results, with an alpha error of 1 per 10.000.