Conferences

Time series involving an autoregressive component and (latent) trend are proving to be challenging to model, as evidenced by discussions about models such as the autoregressive latent trajectory model (ALT) and the continuous autoregressive latent trajectory model (CALT). More fundamentally, the use of models with autoregressive components are still prevalent in psychology as well as many other fields despite the ability to modeling time series using continuous time methods. This paper will examine the modeling of emotion time series using both discrete and continuous time methods. The discrete time models include the application of autoregressive models to intraindividual data which has been altered to mimic differing sampling rates. As is to be expected, differing parameter estimates occur corresponding to the changes in sampling rate. Continuous time methods are then considered, including the method described by Oud & Jansen (2000) for application of the Continuous Time State Equation, and Differential Equation Models as described in articles such as Boker & Nesselroade (2002). These analyses serve the purpose of demonstrating the conflicts that may occur in the literature due to the combination of differing sampling rates and discrete time methods. Consequently, the need for continuous time methods will be highlighted.