|
Abstract:
|
Panel or longitudinal analysis of dynamical systems is becoming more common, in part thanks to the increasing availability of data and to recent advances in statistical methodology that aim at dispensing with linearity and Gaussianity assumptions. Some of these existing methodologies avoid the need of application-specific analytical derivations, including iterated filtering algorithms. However, these algorithms have so far been developed with a multivariate time-series framework in mind without explicitly considering individual effects specific to single time series, which are often analyzed within longitudinal settings, which will in general help both disentangle within-unit processes from between-unit associations and deal with parameter bias or weak identifiability, which are statistical issues that can limit the scientific interest of conclusions. To facilitate attacking such issues in nonlinear, non-Gaussian panels, we consider extending the existing methodology to a novel approach of panel iterated filtering and apply it to model infectious disease dynamics of polio and HIV.
|
