ABSTRACT
The collection of multivariate time-series and their analysis with mathematical models is necessary if we are effectively and fully to represent process and change. Among the promising applications currently available are variations of the common factor model that integrate factor and time series modeling features in a common analytic framework. We highlight some differences and similarities between two kinds of time series models for common factors: a direct autoregressive factor score (DAFS) model and a white-noise factor score (WNFS) model. Particular specifications of these models are fitted to data reflecting short-term changes in an intensively measured individual's self-reported affect. Results of the model fitting underscore the importance of explicit differences in model specifications that define one's view of the nature of process and change.
