A model for assessing the association in the repeated measures of depression among the elderly

The dependence in the outcome variables is a major issue of concern in modeling the correlated data stemmed from the repeated observations. The marginal models such as GEE and the conditional models based on Markov chain have been employed for longitudinal data in the past. However, it has been evident that without addressing the underlying association parameters, the analysis of repeated outcome variables remains far from being resolved. In this paper, a method has been demonstrated to model such data using the underlying dependence in the outcome variables as well as dependence between outcome and explanatory variables. An extension of the regressive model is shown in this paper and a comparison is demonstrated between the existing model (reduced model) and the proposed model (extended model). The models are illustrated for depression among the elderly population in the USA using the Health and Retirement Study data from 1992 to 1998. © International Congress on Modelling and Simulation, MODSIM 2013.All right reserved.

cited By 0