Detection of unobserved heterogeneity with growth mixture models

Latent growth curve models as structural equation models are extensively discussedin various research fields (Duncan et al., 2006). Recent methodological and statisticalextension are focused on the consideration of unobserved heterogeneity in empiricaldata. Muth´en extended the classical structural...

Descripción completa

Autores Principales: Reinecke, Jost, Mariotti, Luca
Formato: Artículo
Idioma: Español
Publicado: 2015
Acceso en línea: http://revistas.ucr.ac.cr/index.php/matematica/article/view/1416
http://hdl.handle.net/10669/12944
Sumario: Latent growth curve models as structural equation models are extensively discussedin various research fields (Duncan et al., 2006). Recent methodological and statisticalextension are focused on the consideration of unobserved heterogeneity in empiricaldata. Muth´en extended the classical structural equation approach by mixture components,i. e. categorical latent classes (Muth´en 2002, 2004, 2007).The paper will discuss applications of growth mixture models with data from oneof the first panel studies in Germany which explore deviant and delinquent behavior ofadolescents (Reinecke, 2006a, 2006b). Observed as well as unobserved heterogeneitywill be considered with growth mixture models using the program Mplus (Muth´en& Muth´en, 2006). Special attention is given to the distribution of the substantivedependent variables as a count measures (Poisson distribution, zero-inflated Poissondistribution, cf. Nagin, 1999). Different model specifications with respect to substantivequestions will also be emphasized.Keywords: Panel data, growth mixture models, heterogeneity, Poisson distribution.