Screen time and NSST showed opposing patterns in relation to key socio-demographic factors, such as sex, age, weight status, household income, parental education and day type.
The contrasting patterns of NSST tended to compensate (at least partially) for SST across socio-demographic groups. One interpretation of this compensatory effect may be that adolescents have a homeostatic mechanism regulating sitting time, a "sedostat" analogous to the "activitystat" hypothesis proposed by Wilkin et al. . Wilkin and colleagues  have suggested that increasing physical activity in one domain (e.g. school) may result in decreasing physical activity in another (at home). Similarly, it may be that reduction of SST will result in increased NSST. This hypothesis has synergies with the 'compensation effect' described by Biddle et al.  where adolescents switch between sedentary choices rather than accruing additional total sedentary time. Another possible explanation relates to behavioural economics theory, which suggests adolescents often make choices between being physically active or sedentary and are influenced by both reinforcing and constraining environmental and social factors. If such a homeostatic mechanism exists, it is unclear whether it has a physiological, environmental or social basis, or a combination of such influences .
Regardless, SST and NSST should not be seen as qualitatively equivalent or interchangeable. The energetic demand of SST (mean = 1.3 METs) is estimated to be somewhat lower than NSST (mean = 1.5 METs). Additionally, NSST is often considered to be more socially "valuable" than SST. From a use of time point of view, SST is much more "permeable" than NSST, that is, it allows for fragmentation (interspersing bouts of SST with other activities) and time compression (reducing total time commitment by multi-tasking). Furthermore, television time is associated with exposure to food advertising and with increased snacking, and hence perhaps with unfavourable dietary habits . In contrast, time spent in sedentary homework activities has been associated with more favourable dietary habits, such as increased fruit and vegetable intake . Because of these numerous differences, reducing NSST is unlikely to be functionally equivalent to reducing SST in terms of modifying energy balance.
An important question is the extent to which SST can be used as a surrogate for, or a predictor of, total sedentary time, given that SST is the most commonly measured aspect of total sedentary time. The correlation between SST and total sedentary time was moderate (r = 0.53) but highly significant. However, the standard error of estimate (86 minutes/day) was large, suggesting that in 95% of cases total sedentary time could only be predicted from SST to within about 172 minutes/day, or about 30% of mean total sedentary time. Using median splits, SST could only be used to correctly categorise adolescents into high or low total sedentary time in about two thirds of cases. The predictive power of SST varied with age, parental education, household income, and area-level SES, suggesting that it may be less useful in predicting total sedentary time in particular groups, particularly older adolescents from high SES families.
The specific strengths of this study include the large sample size, the random nature of the selection process, the wide age range, and the fact that activity was assessed using a methodology which yielded very high-resolution use-of-time data. This is also the first study to comprehensively describe NSST, and to compare it to SST. The study nonetheless has limitations. The use of time approach did not capture multi-tasking, asking adolescents to nominate the activity they were most focused on, or dividing stretches of time between two activities. The survey was conducted mainly in autumn and winter, and seasonal patterns may affect both SST and NSST. There are also well known limitations to self-report, particularly with younger children, although the instrument used has been shown to have high reliability and good validity. In this study, however, the self-reported duration of daily TST (575 minutes/day) was somewhat larger than objectively measured TST reported in young people of similar ages; i.e. ~364-450 minutes per day [33–35]. This may be because accelerometers are typically not worn throughout the entire waking day (about 15 hours in this sample). Accelerometer data are often included in analyses if the participant provides >8 h of valid recording per day. Accelerometers may be removed prior to bed and not capture time spent lying awake in bed and other sedentary aspects of the pre-bedtime routine. The choice of accelerometer cut-point used to define sedentary behaviour is also likely to influence the measurement of TST.