The television is ubiquitous in most developed countries, and evidence suggests that many young people exceed recommendations for television viewing [9-11]. During the time spent watching television, adolescents expend little energy [12], and are exposed to numerous advertisements that can influence the type of food desired, requested and consumed [13]. Furthermore, television viewing behaviours may act as a complement to stimulate eating by potentially causing a distraction resulting in a lack of awareness of actual food consumption or overlooking food cues, which may lead to overconsumption and increased energy intake [14]. Early research suggests that young people may associate television viewing with eating from a young age, if for example, parents place their children in front of the television with a snack or a meal while they do other household chores [15]. Cross-sectional studies have found that television viewing is associated with lower fruit and vegetable intake [16,17], higher energy-dense beverage consumption [18,19], and higher energy-dense snack consumption [20,21] in adolescents. Research examining longitudinal associations between television viewing and eating among adolescents has shown that television viewing is predictive of lower fruit and calcium intakes and higher sugar-sweetened beverage and fast food intakes after five years [22-24]. While such research begins to explore the television-diet paradigm of adolescents, the extent to which specific mechanism(s) are driving the association have not been investigated [22]. The tendency for television viewing and eating behaviours to aggregate has important implications for nutrition promotion, highlighting the need for a greater understanding of the mechanisms (mediators) underpinning the association between television viewing and eating behaviours. Mediators have been described as processes that intervene between input and output [25]. Understanding the mediators of the associations between television viewing and eating behaviours may help in developing nutrition promotion interventions targeting, for example, high television viewers. However, little is known about the mediators of the associations between television viewing and eating behaviours among adolescents.

A potential explanation for the association between television viewing and eating behaviours among adolescents can be inferred from the existing literature on the determinants of dietary behaviour. There is evidence that snacking while watching television is associated with a higher consumption of energy-dense foods and drinks, and a lower consumption of fruit, vegetables and milk [13,26], and that snacking while watching television differs according to time spent watching television [27-29]. Researchers have proposed that foods eaten in front of the television may be a primary underlying mechanism driving the relationship between television viewing and eating behaviours [30,31]. A further potential explanation can be inferred from behavioural choice theory (BCT). BCT draws on principles of behavioural economics [32] in which the decisions to be active or sedentary (or eat certain foods) are based in part on differences in their accessibility, and on the motivation to engage in the activities, which can be conceptualised as the reinforcing value of the activities [33]. Television viewing has a high reinforcing value for many young people, and is associated with the choice to spend more time watching television [33]. Moreover, television has become the most effective communication medium and as young people spend a large amount of time viewing television it has become an important contributor to their education and socialisation [34,35], which increases it's reinforcing value. It is plausible that the reinforcing value of television could be associated with eating behaviours. For example, studies have shown that the reinforcing value of television is associated with increased frequency of children's attempts to influence the purchasing of certain foods at the supermarket [34], and with increased preferences for foods viewed on television [36].

Linear regression analyses, adjusted for adolescent gender and year level, revealed that TV viewing was positively associated with snacking while watching TV (β = 0.01, 95% CI 0.005-0.006, p < 0.001) and with perceived value of TV viewing (β = 0.01, 95% CI 0.008-0.01, p < 0.001). Further linear regression analyses revealed that snacking while watching TV and perceived value of TV viewing were positively associated with energy-dense snack and energy-dense drink consumption, and negatively associated with fruit consumption (Table ​(Table22).

Tables ​Tables3,3, ​,44 and ​and55 show the results of the mediation analyses procedures applied to the prediction of T2 energy-dense snacks, energy-dense drinks, and fruit consumption, respectively. Prediction of energy-dense snack consumption by TV viewing declined significantly after controlling for both mediators, suggesting that variations in energy-dense snack consumption by TV viewing were mediated by these constructs (Table ​(Table3).3). Based on the magnitude of the t-values, eating snacks while watching TV played the greatest mediating role (t(1727) = 8.96, p < 0.001).

Prediction of energy-dense drink consumption by TV viewing also declined significantly after controlling for both potential mediators, suggesting that variations in energy-dense drink consumption by TV viewing were mediated by these constructs (Table ​(Table4).4). Based on the magnitude of the t-values, the perceived value of TV viewing played the greatest mediating role (t(1727) = 6.17, p < 0.001).

Cross-sectional studies have found associations between television viewing and unhealthy eating behaviours in pre-school children [48,49], school-aged children [50,51], adolescents [52,53], and adults [54,55]. Longitudinal studies examining the associations between television viewing and eating behaviours are becoming more prevalent; however, data on adolescents comes predominantly from the US. Data from project EAT has shown television viewing to be predictive of lower fruit and calcium intakes, and higher sugar-sweetened beverage and fast food intakes after five years [22-24]. Other prospective research from the US has shown that for each hour increase in television viewing adolescents consumed additional energy intake of 106 kcal a day [56] and decreased intakes of fruit and vegetables [57]. Findings from the current study support and add to previous studies by showing that adolescents in Australia who watch more television are more likely to have poorer eating behaviours. The present study also supports previous literature showing that television viewing is associated with snacking while watching television [28,29], and that snacking while watching television is positively associated with consumption of energy-dense snacks and drinks, and negatively associated with consumption of fruit [13,26]. For some young people a significant proportion of their daily energy intake is consumed while watching television [30]. Experimental studies have shown that watching television while eating may cause a distraction resulting in a delay in normal mealtime satiation and a reduction in internal satiety signals [58,59], which may lead to overeating. Given that adolescence is a critical developmental period, during which lifelong behaviours are formed [60], the present findings suggest that examining methods to modify television viewing and snacking and eating behaviours in front of the television should be a priority for nutrition promotion.

The present study advances previous findings by demonstrating that variations in snacking while watching television play an important role in mediating the associations between television viewing and adolescents' intakes of energy-dense snacks and drinks, and fruit 2-years later. This is a novel finding, since to the best of our knowledge, no previous studies have examined potential mediating influences of the association between television viewing and eating behaviours in adolescents. Snacking while watching television partially explained (mediated) the positive association between television viewing and consumption of energy-dense snacks and drinks, and the negative association between television viewing and consumption of fruit. While watching television, adolescents are exposed to a larger number of advertisements, which are more often for energy-dense foods compared to fresh foods such as fruits and vegetables [61]. Previous research has shown that television advertising directed at young people influences their preferences, requests and short-term consumption of foods and beverage products advertised on television [13]. It may be, therefore, that those adolescents who watch more television, subsequently prefer, request and consume as snacks more of the types of foods advertised, which would explain the associations observed between these variables. Such findings suggest that reducing television viewing and changing snacking habits while watching television may be important strategies for improving the overall healthfulness of adolescent eating behaviours.

The present study found that the perceived value of television viewing varied by the amount of time spent watching television. Adolescents who watched more television placed a higher value on television viewing than those who watched less. This part of the model was cross-sectional so we cannot rule out reverse causality. For many young people, sedentary behaviours such as television viewing have a high reinforcing value (motivation to participate), so they are more likely to become sedentary and less likely to develop a regular physical activity habit [62]. Furthermore, accessibility of sedentary behaviours is much easier than that for physical activity for most people [33], which increases the amount of motivation people need to choose activity over more sedentary pursuits. Experimental research using positive reinforcement to reduce sedentary behaviours and increase the accessibility of physical activity has had mixed success at increasing physical activity [43,63,64]. Future research should examine the applicability of such an approach in real world settings.

Further analyses showed that the value that adolescents placed on television viewing partially explained (mediated) the association between television viewing and consumption of energy-dense snacks, drinks and fruit. Again, this is a novel finding, since to the best of our knowledge, no previous studies have examined potential mediating influences of the association between television viewing and eating behaviours in adolescents. A potential explanation for such associations is that if adolescents place a high value on watching the television, they are more likely to take on board the messages promoted, which as mentioned above, are likely to encourage the consumption of energy-dense foods and drinks rather than more healthy snacks (e.g. fruit). Future research aiming to improve adolescent eating behaviours should assess the efficacy of methods to address the values that adolescents place on television viewing.