The present research was based on a self-control model as a theoretical approach to understanding health behavior patterns. For substance use, theoretical papers have outlined how behavioral and emotional regulation are important factors in the etiology and prevention of substance use [5, 6]. Recent research has shown self-control measures are related to substance use in adolescence [7, 8] and self-control concepts have been incorporated in theoretical models of adaptation to illness among adults [9]. However, at present there have been few applications of self-control concepts for research on adolescents' diet and exercise behaviors.

Self-control is broadly defined as a set of processes involved in guiding and monitoring behavior [10]. Theory and research suggest that various aspects of self-control are interrelated and are not identical to intelligence [11]. Prior research with adolescents has shown that there are two domains of self-control measures, which are termed good self-control or planfulness, and poor self-control or impulsiveness; these two domains are statistically distinct and have different antecedents and consequences [12]. There are several facets for each type of self-control construct and these may have relevance for diet and physical activity. In the domain of good self-control are indicators such as planning and problem solving, which could be relevant for the planning and implementation of proper dietary choices and exercise patterns. An orientation toward the future may also have relevance for sustaining healthful patterns of behavior. Under the domain of poor self-control are aspects such as distractibility and impatience, which could interfere with making informed decisions, and impulsiveness, which could interfere with attempts at establishing patterns of more healthful behavior.

There has been previous research on the relation of psychosocial factors to adolescents' dietary and physical activity behaviors; however, these studies have mostly been based in the health belief model and theory of reasoned action, which do not explicitly include the concept of self-regulation. Social-cognitive theory [13] includes discussion of self-regulation but studies conducted within this framework have tended to focus on self-efficacy and outcome expectancies [14, 15]. A recent study with church-going adults in a rural Southern area [16] did include a measure of self-control for physical activity, based on items such as setting aside time for activity and choosing to walk instead of driving. This study found that self-regulation was modestly related to self-efficacy, and the self-regulation measure was inversely related to negative outcome expectancies and positively related to physical activity. To our knowledge the Anderson et al. study [16] with adults has been the only one to specifically examine this construct, and there is a need for research with adolescents to study how different aspects of self-control are related to diet and exercise in this age group. While adolescents' diet and exercise may be determined to some extent by family eating patterns and environmental factors, the dynamic of development during adolescence is one of decreasing time with parents and increased time spent with friends and in leisure and recreation activities [17]. From middle school onward, young persons have control over food choices in school cafeterias and in social activities with friends after school, and they have discretion about engaging in physical activity versus maintaining a sedentary lifestyle, both in the school context and at home. Thus there is reason to examine self-control concepts in relation to adolescents' dietary intake and frequency of physical activity.

The present research investigated how a multivariate conceptualization of self-control is related to diet exercise measures in a community sample of adolescents. We obtained standardized measures of low-fat and high-fat food intake and measures of vigorous physical exercise and sedentary behavior. The self-control constructs were related to diet and exercise measures in a structural equation modeling analysis, including covariates so as to control for any correlation of self-control with gender, ethnicity, and parental education. Multiple-group modeling procedures were then performed to determine whether effects for self-control constructs are comparable across gender and ethnic groups.

The results showed that planfulness and impulsiveness had different types of relationships to dietary and exercise behaviors. Persons who scored higher on good self-control had more fruit and vegetable intake and less sedentary behavior, as well as having more participation in sports. A different pattern was noted for persons scoring higher on poor self-control, who had more saturated-fat intake and a lower frequency of engaging in vigorous exercise. These results were based on a diverse sample of adolescents and analyses indicated that effects of self-control were not limited to a particular demographic subgroup, as effects of self-control on dietary and physical activity outcomes were found to be comparable across gender and ethnic groups. The finding of different patterns for the two domains of self-control is consistent with research in other areas [12, 26]. The suggestion is that a higher level of good self-control relates to behaviors that require information seeking, planning, and investment of effort to accomplish, whereas persons who score high on poor self-control are more susceptible to situational pressures and less inclined to resist temptations for undesirable behaviors [11]. Theoretical papers have outlined dual-process models of adolescent behavior positing a more reasoned/reflective pathway together with a more impulsive/reactive pathway [27], and these concepts may be useful for understanding processes in the formation of diet and exercise patterns.

The finding of significant effects for self-control in this study indicates that adolescents do have some choice in regard to their diet and exercise so individual characteristics can play a role in shaping patterns of eating and frequency of physical activity, and they may interact with environmental factors. The results suggest directions for further research on health-related behavior in the context of social-cognitive theory [28]. It is possible that self-control has direct effects on dietary intake and physical activity, not involving any other processes. Research on adolescent substance use, however, suggests that indirect effects should also be considered [11]. It is possible that developing a higher level of good self-control is associated with greater awareness of vulnerability to health risks and a more reasoned approach to making decisions about diet and exercise [29-30]. The component processes underlying a higher level of poor self-control are conducive to a present-oriented perspective in which there is less concern with the future, more inclination to affiliate with teens who are engaging in health risk behaviors [12, 18], and less motivation to persevere in achieving goals for more healthful behavior [31]. These kinds of processes may be explored in further research on relations of self-control processes to dietary and physical activity outcomes.

The results of the present study are consistent with national data in showing gender and ethnicity related to diet and exercise patterns [e.g., 32, 33]. It is recognized that the determinants of dietary intake and physical activity are multifactorial in nature, including variables such as home availability of fruits and vegetables and family support for and modeling of behavior patterns [15, 16, 34]. In addition, social factors act to promote or create barriers to healthful behavioral patterns [35, 36], and environmental factors may affect access to shopping and physical activity opportunities [37, 38]. These kinds of variables may be addressed in efforts to reduce health disparities in the population.

Although relatively simple measures were used to index diet and exercise, the self-control constructs accounted for 5-15% of the variance in the health behavior outcomes controlling for demographics, a contribution that is meaningful from a prevention standpoint in view of the multiple contributions to such behaviors and the fact that self-control characteristics are modifiable. Interventions in clinic and home settings that included components for goal-setting and self-monitoring have shown effects for reducing body-mass index scores [39, 40], and the present research has suggested problem solving, delay of gratification, self-reinforcement and other cognitive skills as also relevant for diet and exercise patterns. Thus we think in addition to family environment, attention to self-control processes is warranted as an avenue of investigation for programs designed to improve health behaviors in childhood and adolescence.