Resumo

ARNAU, Jaume  e  BONO, Roser. Longitudinal studies: Desing and analysis models. Escritos de Psicología [online]. 2008, vol.2, n.1, pp.32-41. ISSN 1989-3809.

The models that traditionally have been used to analyse repeated measure data are linear and follow an approach based on analysis of variance. Their main drawback is that they require balanced data, something that is difficult to achieve in applied contexts. Therefore, alternative models such as the study of growth curves have been developed, which in turn have been used to derive a large number of methods. These methods model both between- and within-individual variation and do not require balanced data. Today, linear mixed models are applied as a general analytical alternative. Mixed models estimate both the expected values of observations (fixed effects) and the variances and covariances of the observations (random effects). So what distinguishes the linear mixed model from the general linear model is the calculation of covariance parameters which allow the analysis of longitudinal data (correlated, incomplete, and with non-constant intervals between observations).

Palavras-chave : Repeated measures designs; longitudinal data; repeated measures ANOVA; MANOVA; GMANOVA; linear mixed model.

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )