Publication: Análisis secuencial de datos observacionales en investigación educativa (y II):
Perspectiva multivariante con modelos log-lineales y logit
Authors
Tójar Hurtado, Juan Carlos ; Serrano Angulo, José
item.page.secondaryauthor
item.page.director
Publisher
Universidad de Murcia
publication.page.editor
publication.page.department
DOI
item.page.type
info:eu-repo/semantics/article
Description
Abstract
Los modelos lag-lineales y logit proporcionan una alternativa multivariante a las técnicas
clásicas de análisis de datos categóricos secuenciales procedentes de la observación sistemática.
En este trabajo se muestra la utilidad y la adecuación de estos modelos matemáticos para
representar fenómenos de interacción observados en sus contextos de origen. Se muestra el
procedimiento completo incluyendo la organización de los datos registrados en tablas de contingencia
multidimensionales, la construcción, la evaluación y la interpretación de los modelos así
como la estimación de sus correspondientes parámetros. Se apuntan además otras posibilidades
diferentes de análisis secuencial de datos observacionales mediante modelos lag-lineales a
desarrollar en futuras investigaciones (modelos de cuasi-independencia, simetría y cuasi-simetría,
modelos con datos ordinales y relaciones con los modelos causales). Por último se concreta
una aplicación de los modelos logit en la evaluación de la calidad de los datos registrados
(concordancia secuencial).
Log-linear and logit models provide a multivariant choice to the traditional techniques of analysis of categorial and sequential data derived from systematic observation. Throughout this report it is shown the usefulness and fitting of these mathematical models to represent phenomenons of interaction observed in their original contexts. The whole process is shown including the organization of data recorded in tables of multidimensional contingencies, the construction, assessment and interpretation of the models, as well as the estimation of its corresponding parameters. Other different possibilities from the sequential analysis of observational data are pointed out by means of log-linear models to be developed in later researches (e. g. models of quasi-independence, symmetry and quasi-symmetry, models with ordinal data and relations with causal models). Finally, it is established an application of logit models to assess at the recorded data quality (sequential concordance).
Log-linear and logit models provide a multivariant choice to the traditional techniques of analysis of categorial and sequential data derived from systematic observation. Throughout this report it is shown the usefulness and fitting of these mathematical models to represent phenomenons of interaction observed in their original contexts. The whole process is shown including the organization of data recorded in tables of multidimensional contingencies, the construction, assessment and interpretation of the models, as well as the estimation of its corresponding parameters. Other different possibilities from the sequential analysis of observational data are pointed out by means of log-linear models to be developed in later researches (e. g. models of quasi-independence, symmetry and quasi-symmetry, models with ordinal data and relations with causal models). Finally, it is established an application of logit models to assess at the recorded data quality (sequential concordance).
publication.page.subject
Citation
item.page.embargo
Ir a Estadísticas
Este ítem está sujeto a una licencia Creative Commons. http://creativecommons.org/licenses/by-nc-nd/4.0/