In an extension of maladaptive behavior determinism (MBD) theory, which states that ordered behavior patterns over time are suggestive of disease states, we examined the relation between leisure activity and health behavior over time. MBD is derived from complexity or chaos theory. It was hypothesized that, over time, increased activity levels would be related to more randomly occurring health behaviors. For 68 participants, daily self-monitoring of leisure activities and four health indicators (healthy eating, feeling hassled, positive mood, and drinking alcohol) were assessed for five weeks and modeled using multiple time series methods. Results showed some support for the hypothesis, particularly with respect to the health indicator feeling hassled. The findings extend support for MBD, and also suggest that physically very active leisure time might have health benefits that are dynamical and not necessarily immediately apparent.

The increased appreciation of the benefits of leisure activity and active living (Di Bona, 2000) are well known and reflected in an increase in exercise participation and a positive response to public health promotion efforts (Prankish, Milligan, & Reid, 1998). The benefits of leisure activity include alleviation of anxiety (Kaufmann, 1988), increased well-being (Coleman & Iso-Ahola, 1993), identity development (Kleiber & Rickards, 1985), improved physical and mental health (Calclwell, Smith, & Weissinger, 1992; Winefield, Tiggemann, & Winefield, 1992), and stress-coping benefits (Shaw, Caldwell, & Kleiber, 1996). Prior research often does not distinguish type of leisure activity. However, physically active leisure activities (versus sedentary ones) have been shown to be preventive factors for cardiovascular and other major diseases (Folsom et al., 1997; Mensink, Deketh, MuI, Schuit, & Hofmeister, 1996; Schlicht, 2002; Wenger, 1996). Therefore, inspection of the effect of different types of leisure activities upon health appears to be warranted.

In addition to exploring the effect of type of leisure activity, an issue that has yet to be addressed is the dynamical relation between leisure activity and health. (Prankish et al., 1998; Kleiber, Hutchinson, & Williams, 2002; Mahoney & Stattin, 2000; Mota & Esculas, 2002; Zeijl, te Poel, du Pois-Reymond, Ravesloot, & Meulman, 2000). Dynamical modeling represents how a system changes or "behaves" as time passes and requires repeated measures of variables, such as by daily self-monitoring. This is in contrast to static modeling, which only examines variables at a single point in time. Prior research has relied almost exclusively on static modeling using retrospective self-reports of leisure activity. Given that some health indicators are unstable, such as mood (Hill & Hill, 1991; Thayer, 1996), it is important to inspect their relation to leisure activities over time.

For purposes of this study, we define health in the broadest sense to include not only absence of disease, but also overall well-being (e.g., happiness and quality of life). We define health indicators to include not only physical behaviors related to (or indicators of) health, but mood states as well. Our primary objective was to examine the dynamical influences (i.e., over time) of physical leisure activity on health indicators. This was possible by incorporating daily self-monitoring in a prospective manner for the assessment of all variables (leisure activity and health indicators). Additionally, an open-ended format was incorporated for assessing types of leisure activity, avoiding the problem of pre-determined (and possibly insufficient) categories.

This approach allowed for the testing of an important health theory that has been emerging in recent years: maladaptive behavior determinism (MBD), which is based on complexity or chaos theory. This theory originated through studies of behavior involving laboratory-based biological and psychophysiological assessment over time, such as electrocardiograms (Goldberger & West, 1987). Evidence has suggested that adaptive conditions exhibit a high degree of randomness, whereas maladaptive conditions exhibit more deterministic, and possibly chaotic, characteristics (Babloyantz & Destexhe, 1986; Ehlers, Havstad, Garfinkel, & Kupfer, 1991; Goldberger et al., 1987). However, in studies of behavior requiring self-report outside of the laboratory, comparable evidence has been developing only very gradually (Barton, 1994; Melancon, Joanette, & Belair, 2000).

Nevertheless, the above-cited findings provide support for MBD, which states that adaptive behavior is characterized by a dominant random component because it adjusts to exogenous factors in the environment that occur independently of time (Skarda & Freeman, 1987). Maladaptive behavior is characterized by greater determinism because it responds relatively more to endogenous factors, which do occur as a function of time and perseverate even in the presence of environmental change. So, according to MBD, when an individual is healthy, measures of behavior (indicators) are characterized by unstructured (i.e., random) fluctuations over time. When the individual is unhealthy, these same measures exhibit structured (i.e., cyclic or periodic) patterns over time (Goldberger et al, 1987; Heiby, Pagano, Blaine, Nelson, & Heath, 2003; Skarda et al., 1987). A cycle is simply a pattern of change that repeats itself over time. For example, the rise and fall of the tides follows a cyclic pattern.