Modelo lineal mixto de clases latentes con respuesta ordinal y su aplicación en la medición de la religiosidad
Files
Date
2020-01-16
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Pontificia Universidad Católica del Perú
Abstract
Los modelos lineales mixtos de clases latentes desarrollados por Proust-Lima, Philipps y Liquet (2017) son útiles para analizar el aspecto dinámico y la naturaleza multidimensional de un fenómeno de interés en poblaciones no necesariamente homogéneas. Estos permiten identificar las posibles clases latentes en la población bajo estudio y cómo un conjunto de covariables afecta en cada clase a la variable respuesta de interés. En esta tesis se desarrolla el modelo lineal mixto de clases latentes con variable respuesta latente y variable mani-fiesta ordinal, a través de sus dos componentes: el sub-modelo estructural y el sub-modelo de medición, que son complementados con un modelo logístico multinomial para analizar la probabilidad de pertenencia a una clase latente. El modelo se aplicó a un conjunto de datos pertenecientes al Estudio Nacional de Juventud y Religión (NSYR por las siglas en inglés “National Study of Youth and Religion”), con el fin de encontrar clases latentes en el constructo religiosidad y describir su evolución. Como resultado, se identificaron tres clases latentes con trayectorias distintas para cada caso.
Latent class linear mixed models developed by Proust-Lima, Philipps y Liquet (2017) are useful to analyze the dynamic aspect and the multidimensional nature of a phenomenon of interest in populations not necessarily homogeneous. These allow to identify the possible latent classes in the population under study and how a set of covariates affects the response variable of interest in each class. In this thesis, the latent class linear mixed model with latent response variable and ordinal manifest variable is developed, through its two components: the structural sub-model and the measure sub-model, which are complemented with a mul-tinominal logistic model to analyze the probability of belonging to a latent class. The model was applied to a dataset from the National Study of Youth and Religion (NSYR), in order to find latent classes in the religiosity construct and to describe their evolution. As a result, three latent classes were identified with different trajectories for each case.
Latent class linear mixed models developed by Proust-Lima, Philipps y Liquet (2017) are useful to analyze the dynamic aspect and the multidimensional nature of a phenomenon of interest in populations not necessarily homogeneous. These allow to identify the possible latent classes in the population under study and how a set of covariates affects the response variable of interest in each class. In this thesis, the latent class linear mixed model with latent response variable and ordinal manifest variable is developed, through its two components: the structural sub-model and the measure sub-model, which are complemented with a mul-tinominal logistic model to analyze the probability of belonging to a latent class. The model was applied to a dataset from the National Study of Youth and Religion (NSYR), in order to find latent classes in the religiosity construct and to describe their evolution. As a result, three latent classes were identified with different trajectories for each case.
Description
Keywords
Modelos lineales (Estadística), Variables latentes, Religión--Juventud--Modelos estadísticos
Citation
Collections
Endorsement
Review
Supplemented By
Referenced By
Creative Commons license
Except where otherwised noted, this item's license is described as info:eu-repo/semantics/openAccess