INVESTIGATING METHODS OF INCORPORATING COVARIATES IN GROWTH MIXING MODELING: A SIMULATION STUDY

Simple item page
dc.contributor.advisorHarring, Jeffrey R.en_US
dc.contributor.authorLi, Mingen_US
dc.contributor.departmentMeasurement, Statistics and Evaluationen_US
dc.contributor.publisherDigital Repository at the University of Marylanden_US
dc.contributor.publisherUniversity of Maryland (College Park, Md.)en_US
dc.date.accessioned2015-09-18T05:38:47Z
dc.date.available2015-09-18T05:38:47Z
dc.date.issued2015en_US
dc.description.abstractThe current research aims to evaluate the performance of various approaches for estimating covariates within the latent class membership regression model in the context of growth mixture models. Researchers have been searching for more efficient and accurate estimation methods for incorporating covariate information in mixture modeling in order to clearly differentiate between subjects from different groups and to make interpretation of the growth trajectories more meaningful. However, few studies have considered more complicated models such as growth mixture models where the latent class variable is more difficult to identify. To this end, two Monte Carlo simulations were conducted. In Simulation I, four estimation approaches were investigated to examine parameter recovery, variance and standard error efficacy related to both categorical and continuous covariates that defined the regression model for the latent class membership part of the model. Data generated for Simulation II include three covariates, with one dichotomous variable linked to latent class membership and the other two (one dichotomous and one continuous) associated with measurement part of the growth mixture model. Three estimation approaches were then compared using the population data generation model as well as a misspecified model.en_US
dc.identifierhttps://doi.org/10.13016/M2JS9Z
dc.identifier.urihttp://hdl.handle.net/1903/16943
dc.language.isoenen_US
dc.subject.pqcontrolledEducational tests & measurementsen_US
dc.subject.pquncontrolledcovariateen_US
dc.subject.pquncontrolledgrowth mixture modelingen_US
dc.subject.pquncontrolledlatent classen_US
dc.titleINVESTIGATING METHODS OF INCORPORATING COVARIATES IN GROWTH MIXING MODELING: A SIMULATION STUDYen_US
dc.typeDissertationen_US

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Li_umd_0117E_16371.pdf
Size:
1.61 MB
Format:
Adobe Portable Document Format
Download

Collections

UMD Theses and Dissertations
Human Development & Quantitative Methodology Theses and Dissertations