Longitudinal research studies with panel data are often applied to analyze processes of stability and change in individuals or groups. Working on this kind of data allows to explore individual differences and changes of patterns in variables across time. While methods as analysis of variance (ANOVA) and of covariance (ANCOVA) are based on the assumption of homogeneous covariance matrix across the levels of the between-subjects factors and same covariance patterns for the repeated measures, on the basis of the structural equation modelling methodology is possible to analyze longitudinal data with a new paradigm: the latent growth curve models (e.g. Muthén & Asparaouhov, 2006; Nylun, Asparouhov, & Muthén, 2007; Green, 2014). This statistical approach models heterogeneity by classifying individuals into groups with similar patterns, called latent classes. Indeed, in Latent Growth Modeling approaches, such as Latent Class Growth Analysis (LCGA) and Growth Mixture Modeling (GMM) repeated measurements of observed variables are used as indicators of latent variables that describe specific characteristics of individual’s change. In this methodology intercept and slope are considered two latent variables (called also random coefficients), which respectively represents the level of the studied variable when time is equal to zero, and the rate of change in the same variable across time. Strengths of this approach are the flexibility in modelling changes across time and the availability of indexes to adequately evaluate these models. The present study shows an application of growth mixture models considering unobserved heterogeneity within the framework of Mplus (Muthen and Muthen, 2001, 2004). A daily survey was conducted on a sample of 108 first- and second-year undergraduate students enrolled at a Psychology degree. During the study week they were asked to answer to an on-line questionnaire focused on several measures, among which Positive and Negative Affect. Alternatives models were compared modelling different latent classes, random effects and covariates to better model changes over time of Positive and Negative Affect.
|Titolo:||Positive and Negative Affect over time: an application of Latent Class Growth Analysis (LCGA) and Growth Mixture Modeling (GMM)|
|Data di pubblicazione:||2016|
|Appare nelle tipologie:||04.02 Abstract in Atti di convegno|