Using linear mixed models in longitudinal studies: Application of SAS PROC MIXED

• Autores: Roser Bono Cabré, Jaume Arnau Gras, Nekane Balluerka Lasa
• Localización: REMA, ISSN-e 1135-6855, Vol. 12, Nº. 2, 2007, págs. 15-31
• Idioma: español
• Enlaces
  • Texto completo (pdf)
• Dialnet Métricas: 2 Citas
• Resumen

  • The objectives of this article are twofold: (a) to outline the basic concepts associated with the linear mixed model and (b) to illustrate how this model can be used to analyse systematic interindividual differences in intraindividual change, this being achieved through a longitudinal study of a cohort of children living in Cordoba (Argentina). These objectives will be met by using the PROC MIXED statement of the SAS software. This software fits a wide variety of linear mixed models to longitudinal data, thus enabling valid statistical inferences to be made. Since the choice of covariance structure may influence the values obtained in significance tests for fixed effects, we focus our attention on this aspect. The most common covariance structures for modelling longitudinal data are described and guidelines are proposed for choosing the structure which enables more powerful and more efficient regression parameter estimates to be made.