Fingerprinting time series: Dynamic patterns in self-report and performance measures uncovered by a graphical non-linear method

Two simple non-linear techniques are shown to be useful for understanding the dynamics of affect, symptoms, social interaction experience and cognitive performance. The techniques are justified by arguments derived from chaos theory, and demonstrated using data from an intensive time sampling study in which 30 subjects completed a set of self-ratings and a memory task on a hand-held computer every two hours during waking hours for 14 days. The data were pooled and two types of Poincaré plot were constructed for each variable. The first was a plot of each value against its predecessor, and the second was a plot of each change in value from one interval to the next against the previous change. These plots are particularly suitable for uncovering asymmetric state-dependent changes in control between time points. The plots showed a number of distinctive ‘fingerprints’ for the different variables. Altogether, the results suggest that the plots are a novel and useful method for understanding psychological variables in terms of their dynamic control.