ABSTRACT
Models considered up to now assume that measurements are obtained simultaneously at a given point in space/time. They take no account of ordering among the variables, such as ordering in time or ordering across space. For example, a common factor model ordinarily has the assumption that subjects are measured on all variables simultaneously, so there is no ordering to be given among them representative of anything in the world. (They may be given numbers but these are not representative of anything other than that one variable is different from another.) But when measurement variables are ordered in time or space, this changes how we model their relationships, because then we must take into account this ordering. We will call models that take into account ordering of variables in space or time “longitudinal models.” The field of longitudinal modeling is now quite well developed and encompasses many approaches to modeling. SEMs are only one approach to longitudinal modeling. We can thus only refer briefly to some representative longitudinal models.
