Using state space grids to display, describe, quantify, and analyze synchronized time series or event sequences

T. Hollenstein

Human Development and Applied Psychology, University of Toronto, Toronto, ON, Canada

Advances in computing have accelerated the growth in observational methodologies and have allowed researchers to capture complex sequences of events as they unfold in time. This increase in observational data has created a demand for methods that can visually display, describe, quantify, and analyze two or more synchronized time series or event sequences. The current paper describes a new methodology based on dynamic systems principles, state space grids, which is well-suited for the kind of data obtained via observational programs such as The Observer^® (Noldus Information Technology bv, The Netherlands) and others.

State space grids are two-dimensional grids constructed from synchronous sequences of two ordinal or categorical data streams. Each axis represents mutually-exclusive categories along a single dimension, such as low-medium-high levels of one variable or discrete behavioral states of one subject. The cells on the grid represent joint states of categories along the two dimensions. For example, on a 3x3 grid, the x-axis may represent a mother’s behavior (negative, neutral or positive) and the y-axis represent a child’s behavior (negative, neutral, or positive) and each cell represents the dyadic state (mom positive, child negative). Behavior is plotted on these grids as a sequence of events that move from state to state – each time the behavior changes a point is plotted in the new state and a line is drawn that connects the old and new state. In this way, the trajectory of behavior along 2 dimensions can be viewed.

The freely available software program GridWare (www.statespacegrids.org) will be used to demonstrate how synchronized time series or event sequences can be depicted on a state space grid. GridWare also allows for the selection of multiple trajectories based on grouping characteristics (i.e. gender, treatment group), derives measures describing the pattern of behavior across the whole grid or within a cell, and can export these measures and images of the plots for use in other programs. Examples from data coded with The Observer will be shown.

Paper presented at Measuring Behavior 2005 , 5^th International Conference on Methods and Techniques in Behavioral Research, 30 August - 2 September 2005, Wageningen, The Netherlands.

© 2005 Noldus Information Technology bv