The purpose of this study was to examine the cross-sectional associations between a wide range of characteristics of the family environment and girls' physical activity, television use and dietary behaviors, as well as to explore relationships between family environment factors and girls' BMI and body composition. Across all of the behavioral outcomes, in both the independent and mutually-adjusted analyses, parents' own behavior was associated with their daughters' behavior. The family environment factors observed in this study explained between 4 and 12 percent of the variance in girls' weight-related behaviors beyond the contribution of sociodemographic characteristics. Few associations were observed between the family environment factors and girls' BMI and body composition; however, the number of media resources in the home was positively associated with both girls' BMI and percent body fat.
Previous studies examining the relationship between parents' own PA habits and adolescents' activity have produced mixed results [8, 35]. While in a longitudinal study utilizing parents' report of their physical activity habits, Anderssen et al  found that adolescents whose mothers were active exhibited less of a decline in physical activity through adolescence, most studies of adolescents have not found such a relationship. In their review of environmental correlates of PA in youth, Ferriera et al  noted that studies that utilized parental report of their own activity were more likely to show associations between parental PA and adolescents' activity, a factor that may have contributed to the associations between parental PA and girls' PA observed in the current study. Additionally, as girls who were not regularly physically active were specifically recruited for New Moves, it may be that sedentary girls are likely to have sedentary parents, while there may be a weaker association between parent and girl behavior among regularly-active girls.
The lack of association between family support for PA and girls' PA habits in the mutually-adjusted model was surprising considering a number of previous studies have found that family support and encouragement play an important role in the PA habits of adolescents [8, 21, 36]. However, most of these studies were conducted in primarily white and high SES samples [8, 36]; therefore, familial support may be less influential to the PA habits of adolescents from lower SES and racial and ethnic minority groups. Additionally, few studies examined the combined association of parental modeling of behavior and parental support for behavior on girls' PA, even though often these two family characteristics are associated as parents who enjoy physical activity are likely to encourage their children to participate in activity. Despite previous suggestions that parental modeling may not be influential to adolescents' PA habits in the absence of parental support for PA , findings from the current study suggest that for socio-demographically diverse sedentary adolescent girls, parents' own physical activity habits may be more influential than the methods parents are using to encourage their daughter to be active. Reasons for this lack of an independent relationship may include that adolescent girls perceive parental encouragement as nagging, or are not receptive to parents' encouragement to be active when their parents themselves are not active. These findings support those of Heitzler et al  who observed that while simultaneously evaluating the relationships between family and peer factors and adolescents' PA, parents' PA was significantly associated with adolescents' PA while parental support for PA was not.
Despite previous studies observing multiple family environment predictors of youths' TV use [26, 37], in the current study only parental TV use was associated with girls' TV use. This lack of consistency with other studies may be attributable to the fact that most previous studies utilized samples of grade school children or younger adolescents, whose behavior may be more influenced by parents' restriction of their TV time or by physical resources in the home. Additionally, for the current study girls who were sedentary, did not enjoy physical activity, and were at risk for obesity were actively recruited. These girls' TV use may be influenced by different social and environmental factors as compared to girls who are not sedentary. As the home is the venue at which most adolescents watch excessive TV, and modifying TV use has great potential to influence weight and body composition , identifying methods to intervene on adolescents' TV use is of vital importance.
Consistent with previous studies [13, 15] factors in the family environment including parental intake of soft drinks and FV, home availability of these foods, and frequency of family meals were associated with girls' dietary intake in the independent models. However, only parental intake of soft drinks and FV and soft drink availability in the home remained independent predictors of girls' intake in the mutually-adjusted models. These findings suggest that the positive relationship observed between parents' and girls' intake is not merely due to the greater presence of these foods in the home because both parents and children eat them, but that dietary behavior can be instilled in youth through parental modeling of intake.
Few studies have examined relationships between behavior-specific family environment factors such as support for and modeling of physical activity and dietary intake, and adolescents' weight and body composition. In the current study, relationships were observed between media resources in the home and number of TVs in the home and girls' BMI, and between media resources in the home and girls' percent body fat in the independent models. This relationship was not mediated by girls' TV viewing. A significant inverse association was also found between the frequency of family meals and girls' BMI in the univariate models, a relationship that has been observed in previous cross-sectional studies . This relationship was not mediated by girls' fruit and vegetable intake or by soft drink intake. This lack of mediation by television use and dietary intake suggests that family environment factors may influence weight and body composition via behaviors other than those assessed in this study. For example, the number of media resources and televisions in the home may be associated with girls' weight and body composition because adolescents spend time on the computer or playing video games at the expense of sleeping, an emerging risk factor for obesity .
This study addressed a number of gaps in the literature by examining the role of the family environment in girls' weight-related behaviors among a racially, ethnically, and socioeconomically diverse group of adolescent girls who were either currently overweight or obese or at risk for obesity due to a sedentary lifestyle. To confirm that this was a unique sample of sedentary girls, girls in the current study reported engaging in 3.0 blocks of MVPA per day and 7.9 blocks of sedentary activity (watching TV, listening to music, talking on the phone, using the computer, and hanging around) per day. This is less physical activity and slightly more sedentary activity than reported by a sample of African American and white 9th grade girls who engaged in 3.5 blocks of MVPA and 7.8 blocks of sedentary activity per day . While identifying a population of girls at high risk for overweight and obesity is a study strength, it is also a limitation in that study findings may not be highly generalizabile to other populations of adolescents. An additional strength of this study was the use of parental report of the family environment, which may be more valid than adolescents' report and highlights key areas for interventions aiming to modify parental behavior and the presence of resources in the home. Finally, use of both individual and mutually-adjusted models allowed for a comprehensive exploration of both the total effect of each of the family environment factors, as well as the unique contribution of each of the factors on youths' outcomes. Limitations of the current study include its cross-sectional design, which does not allow for an examination of the temporal relationship between family environment factors and girls' behavior, and the use of self-report measures for family environment factors as well as girls' behavior, which may be subject to reporting bias. Additionally, a limitation of the assessment of the family environment was while there was great breadth in constructs measured, in order to keep the survey at a reasonable length to ensure parent participation, in-depth assessment of some of the family components was sacrificed. Specifically, the lack of multi-item scales to assess some factors in the family environment, such as family support for healthy eating, is particularly limiting in studies such as this one that had a relatively small sample size. Having single-item measures coupled with a small study sample often results in large deviation around the mean, which may contribute to an inability to identify significant relationships when they do exist.