Modeling psychotherapy process by time-series panel analysis (TSPA).

The authors introduce the methodology of aggregated time-series analysis (time-series panel analysis [TSPA]), by which prototypical process patterns are estimated using longitudinal psychotherapy process data. Empirical trajectories of 202 outpatients (15-107 sessions) were available. Presession questionnaires provided measures of patient's well-being and patient's therapy motivation. TSPA was contrasted with growth curve modeling. Fixed effects were estimated in both methods. Unbalanced longitudinal data considering multiple levels can be analyzed. Using Granger causality derived from time-lagged associations, the TSPA pattern revealed feedback relationships between well-being and therapy motivation. Growth curve analysis highlighted logarithmic increases of well-being trajectories. In particular, TSPA can illuminate change mechanisms in psychotherapy field data by its nonexperimental approximation to an analysis of causal dynamic structures.