Analysis of intensive categorical longitudinal data
pp. 241-253
in: Jaan Valsiner, Peter Molenaar, Maria Lyra, Nandita Chaudhary (eds), Dynamic process methodology in the social and developmental sciences, Berlin, Springer, 2009Abstrakt
Intensive longitudinal data are defined as data that come from more than the usual three or four observation points in time yet from fewer than the 100 or more required for time series analysis (Walls & Schafer, 2006). Consider, for example, a clinical design with 20 repeated observations. Data from this design are hard to analyze. Unless the sample is very large, 20 observations are too many for structural modeling. For repeated measures ANOVA with polynomial decomposition, polynomials of up to the 19th order would have to be estimated (which is the easy part) and interpreted (which is the hard part). This applies accordingly to hierarchical linear models of this design. For longitudinal, P-technique factor analysis, 100 observation are needed. In brief, data that are intensive in the sense that more observations are made over time than usual pose specific analytic problems.