Associations between socioeconomic position and changes in children’s screen-viewing between ages 6 and 9: a longitudinal study

Objectives To explore socioeconomic differences in screen-viewing at ages 6 and 9, and how these are related to different media uses. Design Longitudinal cohort study. Setting Children recruited from 57 state-funded primary schools in Southwest England, UK. Participants 1299 children at ages 5–6, 1223 children at ages 8–9, including 685 children at both time points. Outcome measures Children’s total screen-viewing time (parent-reported) and time spent using multiple screen devices simultaneously (multiscreen viewing), for weekdays and weekends. Methods Negative binomial regression was used to model associations between socioeconomic variables (highest household education and area deprivation) and total screen-viewing at age 6 and the change from age 6 to 9. We additionally adjusted for child characteristics, parental influences and media devices in the home. Multiscreen viewing was analysed separately. Results Household education was associated with children’s screen-viewing at age 6 with lower screen-viewing in higher socioeconomic groups (21%–27% less in households with a Degree or Higher Degree, compared with General Certificate of Secondary Education: GCSE). These differences were explained by the presence of games consoles, parental limits on screen-viewing and average parent screen-viewing. Between ages 6 and 9, there were larger increases in screen-viewing for children from A level and Degree households (13% and 6%, respectively, in the week) and a decrease in Higher Degree households (16%), compared with GCSE households. Differences by household education remained when adjusting for media devices and parental factors. Conclusions Children’s screen-viewing patterns differ by parental education with higher levels of viewing among children living in households with lower educational qualifications. These differences are already present at age 6, and continue at age 9. Strategies to manage child sedentary time, and particularly screen-viewing, may need to take account of the socioeconomic differences and target strategies to specific groups.

Conclusions: Children's screen-viewing patterns differ by parental education with higher levels of viewing among children living in households with lower educational qualifications. These differences are already present at age 6, and continue at age 9, even when accounting for differences in baseline screen-viewing and device ownership. Strategies to manage child sedentary time, and particularly screen-viewing, may need to take account of the socio-economic differences and target strategies to specific groups.

METHODS
parents completed a questionnaire about personal and family characteristics while 'second' parents completed a shorter questionnaire.

Screen-viewing data
In both years, parents were asked about the number of hours they and their child typically spent engaging in specific screen-viewing behaviours on weekdays and at weekends: TV, computer, games consoles and, in Year 4 only, multiscreen viewing [26]. These were recorded as either 'None' or in hourly categories from '0-1 hours' up to '4 hours or more', and recoded based on midpoints to give the average number of minutes spent in each type of screen-viewing on weekdays and weekends. These were summed to form the average total number of minutes spent on any type of screen-viewing. Where two parents completed questionnaires, parental screen-viewing was taken as the average (approximately 27% of respondents in each year).

Other measurements
Child height and weight were measured, and body mass index (BMI) was calculated and converted to an age-and sex-specific standard deviation score based on UK reference curves [27,28]. Accelerometer data were processed using Kinesoft (v3.3.75; Kinesoft, Saskatchewan, Canada), and were included if participants provided at least three days of valid data (including at least one weekend day), where a valid day was defined as at least 500 minutes of data after excluding intervals of ≥60 minutes of zero counts, allowing up to two minutes of interruptions. Minutes spent in moderate-to-vigorous-intensity physical activity (MVPA) and mean sedentary time per day were derived using population-specific cut points for children [29]. The first parent was asked whether they limited the time their child spent engaging in three different types of screen-viewing (TV, computer and video games) with responses from 1 'Strongly disagree' to 4 'Strongly agree'. The average of these was used to capture parental limits on screenviewing. We also asked about the number of media devices in the home: TVs, computers (desktop or laptop), tablet computers and games consoles (including handheld consoles).

Statistical Analysis
Descriptive summaries of children's screen-viewing on weekdays and weekends and other participant characteristics were produced for age 6 and age 9. To aid interpretation, the screen-viewing variables are reported in summaries as continuous variables. However, the underlying variables are discrete and so in  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   9 the main analyses negative binomial regression models (similar to Poisson models but suitable for overdispersed count data) were used to model weekday and weekend screen-viewing separately. We modelled the association between total screen-viewing at age 9 and socio-economic variables (household education and IMD) adjusting for baseline screen-viewing at age 6, child characteristics (sex and BMI z-score), screen-viewing patterns (the percentage of total screen-viewing time spent on non-TV screen-viewing and minutes spent multiscreen viewing), child physical activity (average MVPA and sedentary time), parental influences (parental screen-viewing and limits on screen-viewing score) and presence of media devices in the home (more than one TV and any computers, tablets or games consoles). As it includes more than one form of screen viewing and therefore may duplicate information, we assessed multiscreen viewing as a separate outcome. Robust standard errors were used to account for clustering of children within schools. We explored using a range of different multiple imputation approaches to account for missing data and thus maximise the sample size, but due to the categorical nature of the variables, we encountered problems with convergence. We have therefore not presented any imputed data in the paper.
All model results are presented as screen time ratios (rate ratios), defined as the exponent of the model coefficients. An increase of one unit in a given predictor variable is associated with a multiplicative increase in screen-viewing, holding the other predictor variables in the model constant. Model assumptions were checked via model diagnostics, and all analysis was done in Stata v15.
There were differences in the number and types of media devices in the home by household education.
Households where GCSE and A Levels were the highest qualification had more TVs and games consoles in the house, while Higher Degree households had more computers. There were small increases in most devices between age 6 and age 9. The average number of tablet devices was the same across all levels of household education, but increased fourfold to an average of 2.2 per household between the two assessment periods. Parental screen-viewing differed with household education by similar amounts to their children, again with higher screen-viewing among those with lower level qualifications.
Parental screen-viewing was also higher at the second time point, especially among the Higher Degree educated households, and differences with education were smaller.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   15   highest in households with A Level and university degree qualifications (increasing by 21-22% and 16-18% respectively), compared to households with GCSE qualifications (Figure 2, left). There was no association with IMD once household education was taken into account. The presence of games consoles and higher multiscreen viewing were both associated with an increase in screen-viewing of 25-27%. There were weaker associations between screen-viewing and baseline screen-viewing at age 6, parental screen-viewing and proportion of non-TV screen-viewing. Lower screen-viewing was associated with parental limits on screen-viewing in the week, but not at weekends, and with the presence of tablet devices in the home. There was no association between children's screen-viewing and physical activity.
Multiscreen viewing was associated with different factors to overall screen-viewing (Table S2). There was no association between multiscreen viewing and households with different education levels.

DISCUSSION
The data presented in this paper have shown an association between children's screen-viewing and highest household education level, with higher screen-viewing in lower socio-economic groups. These differences were already present at age 6 and continued to be evident at age 9. Except for TV viewing, the time children spent engaged in screen-viewing increased in all households, regardless of education, between ages 6 and 9 on both weekdays and weekends. Adjusting for baseline screen-viewing at age 6, screen-viewing increased most over the time period for children in households where the highest qualifications were A Levels or a university degree. In A-level educated households, this reflected a "catching-up" with the level of screen-viewing seen in GCSE-educated households. These findings imply socio-economic differences in the type and amount of sedentary time and screen-viewing, and the way in which screen-viewing patterns develop. These differences were related to household education rather than area-level deprivation, suggesting that it is long-term socio-economic aspects such as parental knowledge, ability to engage and communicate with services, and possibly income that are important, rather than the socio-economic conditions of the area, such as availability of local resources.
In contrast, multiscreen viewing is associated with area deprivation rather than education and so could represent a proxy for short-term current income or possibly a neighbourhood effect of what is considered to be 'typical' screen-viewing behaviour. As such, strategies to manage child sedentary time, and particularly different types of screen-viewing, may need to take account of the socio-economic differences and target strategies to specific groups.
Participants in households with more devices engaged in more screen-viewing. For example, a child with one or more games consoles in the home engaged in 20% more screen-viewing on weekdays, the equivalent of 144 minutes per day compared to 120 minutes per day for a child with no games consoles.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   18 The number of media devices in the home differed by household education, with more TVs and games consoles in households with GCSE and A level qualifications, and more computers in households with a university or higher degree. While the number of devices changed between the two assessment points, socio-economic differences were small and did not account for the observed differences in total screenviewing between household education. Findings are therefore consistent with previous studies that have a shown a link between the number of media devices and screen-viewing behaviours [12,24]. However, we extend those findings to show that while access to devices may contribute to baseline screenviewing, it does not entirely explain socio-economic differences in the increase in screen-viewing between ages 6 and 9.
There were weak associations between parent and child screen-viewing, with every 30 minutes of parental screen-viewing associated with a 2% increase in child screen-viewing. This association was stronger (~6%) for TV viewing which, consistent with previous studies[12], may reflect parents and children watching TV together. It is important to note that parental limits on screen-time appeared to only be associated with weekday TV and multiscreen viewing. Interestingly, analysis of interviews conducted with a sub-sample of the parents in the study showed that many parents are uneasy about managing non-TV screen-time and feel that they struggle to keep up with rapid technological change [30]. Collectively, these findings may suggest that parental influence on screen-viewing that is not TV viewing is likely to be limited and that there is a need to help parents to identify effective ways to manage constantly adapting forms of viewing.
Tablet ownership did not vary with household education, but increased greatly overall between ages 6 and 9. A limitation of this project is that we did not assess tablet use as a specific behaviour, although we found tablet ownership strongly associated with increased multiscreen viewing. The widespread ownership of tablets is particularly important for understanding the complexity of screen-viewing, as  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   19   watching TV, playing games and using the internet can all be done via a tablet and, as such, identifying the behaviour engaged in while using a tablet is challenging. Thus, it may be that in the future there is a specific need to assess tablet time. A challenge, however, would be how to use tablet data to help parents work with their children to manage overall screen-time, as parents have different views about the benefits and challenges of different forms of screen-viewing. For example, TV viewing is often considered to be family time, while tablet and game console use can be solitary and isolate child from the family [31]. Thus, monitoring tablet use which captures multiple different activities may not be that helpful on its own and it would be more useful to differentiate the various ways in which tablets are used. One way that this could be achieved is via applications that monitor and limit the time spent on games and other non-educational activities. Parents and children could then work together to agree targets for overall and game time on tablets, and the information acquired through the application could be used as a feedback and monitoring device.
The major strengths of this study are the information on a variety of different types of screen-viewing at two time points which span an important period of change in child viewing behaviours. The data collection period (2012-16) coincided with a period of rapid change in screen-viewing technology which has facilitated the presentation of contemporary data. There are, however, several limitations that need to be considered. Firstly, we did not explicitly ask parents about tablet use. As data collection covers a time when screen use is changing rapidly, (for example, global tablet sales increased from 116 million in 2012[32] to 207 million in 2015[33]) this is an important aspect we have not been able to capture directly. Secondly, the differences in the education variable between Year 1 (first parent's education) and Year 4 (household education) may result in inaccuracies in situations where household composition has changed or where people have obtained additional qualifications between the two time points, although we believe that the number of these cases is likely to be small. Finally, as parental screen-  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   20   viewing increased between 2012 and 2015 this may indicate a general change in viewing patterns, and the increases reported between the two age periods could reflect secular changes as opposed to agerelated differences. As such, it would be important to identify if comparable patterns are evident in other datasets.

Conclusion
Children's screen-viewing patterns differ by parental education with higher levels of viewing among children living in households with lower levels of education. These differences are already present at age 6, and continue at age 9, even when accounting for differences in baseline screen-viewing and device ownership. Socio-economic differences narrow with age as children in households with higher qualifications gain greater access to screen-viewing devices. Strategies to manage child sedentary time, and particularly screen-viewing, may need to take account of the socio-economic differences and target strategies to specific groups.

ACKNOWLEDGEMENTS
We would like to thank all of the families and schools that have taken part in the B-PROACT1V project.
We would also like to thank all current and previous members of the research team who are not authors on this paper.

AUTHOR CONTRIBUTIONS
RJ, SJS and JLT were involved in the design of this study and in seeking funding for it. The paper was conceived by RJ and RS, and RS performed all analyses. RS and RJ wrote the first draft of the paper and RJ coordinated contributions from other authors. All authors made critical comments on drafts of the paper.

DATA SHARING STATEMENT
The datasets generated during the current study are not publicly available due as the project is ongoing and data are not ready for archiving. We will consider reasonable requests for access to the data once the project is complete in 2019.

1.
Ekelund   Models are adjusted for baseline screen-viewing at age 6, child characteristics, child physical activity.
Screen viewing additionally adjusted for screen-viewing types at age 9.
Higher Index of Multiple deprivation scores indicate more deprived households      An increase of one unit in a given predictor variable is associated with a multiplicative increase in screen-viewing: e.g., every 30 minutes spent in TV viewing at age 6 was associated with an increase in age 9 multi-screen viewing of 20% on weekdays and 16% on weekends 2 increase per 1 standard deviation (14.0) in IMD 3 increase per 30 minutes screen-viewing

Introduction
Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 5-6 Objectives 3 State specific objectives, including any prespecified hypotheses 6

Study design 4
Present key elements of study design early in the paper 6-7 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, exposure, follow-up, and data Describe any efforts to address potential sources of bias 8-9 Study size 10 Explain how the study size was arrived at 6-7 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, describe which groupings were chosen and why 8-9 (a) Describe all statistical methods, including those used to control for confounding 8-9 (b) Describe any methods used to examine subgroups and interactions 8-9 (c) Explain how missing data were addressed 9 (d) If applicable, explain how loss to follow-up was addressed n/a Statistical methods 12 (e) Describe any sensitivity analyses n/a 28 Methods: Negative binomial regression was used to model associations between socio-economic 29 variables (highest household education and area deprivation) and total screen-viewing at age 6 and 30 the change from age 6 to 9. We additionally adjusted for child characteristics, parental influences and 31 media devices in the home. Multiscreen viewing was analysed separately.

Results
32 Results: Household education was associated with children's screen-viewing at age 6 and with the 33 change in screen-viewing between ages 6 and 9, with higher screen-viewing in lower socio-economic 34 groups. The differences at age 6 were explained by the presence of games consoles, parental limits 35 on screen-viewing, and average parent screen-viewing. Household education differences in the 36 change in screen-viewing between ages 6 and 9 remained.
37 Conclusions: Children's screen-viewing patterns differ by parental education with higher levels of 38 viewing among children living in households with lower educational qualifications. These 39 differences are already present at age 6, and continue at age 9. Strategies to manage child sedentary 40 time, and particularly screen-viewing, may need to take account of the socio-economic differences 41 and target strategies to specific groups. 105 The aim of this paper was to explore socio-economic differences in screen-viewing at age 6, and the 106 change in screen-viewing between ages 6 and 9, and whether any differences can be explained by 107 other factors such as different media devices and parental influences. Developing knowledge in this 108 area will support the creation of targeted behaviour change programmes to reduce screen-viewing.   136 At age 6, the first parent was asked their highest educational qualification, while at age 9, they were 137 asked the highest education qualification of anyone in the household. We combined these to form the  169 However, the underlying variables are discrete and so in the main analyses we used negative 170 binomial regression models (similar to Poisson models but suitable for over-dispersed count data) to 171 model weekday and weekend screen-viewing separately. We considered four main models. Model 1 172 explored cross-sectional associations between socioeconomic position (household education and 173 IMD) and total minutes of screen-viewing at age 6. Model 2 additionally adjusted for possible 174 mediators and confounders: child gender, child BMI, number of devices in the home (more than one 175 TV, and presence of any computers, tablets or games consoles), parental screen-viewing and parental

METHODS
176 behaviour on limiting screen-viewing. Models 3 (unadjusted) and 4 (adjusted for covariates) 177 explored the longitudinal change between ages 6 and 9; that is, associations between socioeconomic 178 position and total minutes of screen-viewing at age 9, adjusting for baseline screen-viewing at age 6.
179 Finally, we also explored the association between socioeconomic position and multiscreen viewing 180 as a separate outcome. Robust standard errors were used to account for clustering of children within 181 schools, and all analyses are based on complete cases. Model results are presented as screen time 182 ratios (rate ratios), defined as the exponent of the model coefficients. An increase of one unit in a 183 given predictor variable is associated with a multiplicative increase in screen-viewing, holding the

RESULTS
187 Children's screen-viewing differed between households with differing levels of education at both 188 ages 6 and 9, with higher screen-viewing in lower qualified households, decreasing as education 189 increased, with 35-50 minutes difference between the highest and lowest household education 190 qualifications (Figure 1). Time spent screen-viewing was higher at age 9 than age 6, and higher at 191 weekends than during the week among all education groups, with time spent watching TV and 192 playing games consoles nearly doubling at weekends (Table S1 and (Table S3).  219 parental screen-viewing and parental limits on screen-viewing all measured at age 6. These factors 220 accounted for household education differences on both weekdays and weekends. Children's screen-221 viewing was higher in households with games consoles, and lower when parents were more likely to 222 limit screen-viewing (Table S4). There were weak associations with parental screen-viewing and 223 child BMI.
224 Models 3 and 4 examined at the change in screen-viewing between ages 6 and 9 (   271 overall between the two assessment points, most notably with a large increase in tablet devices and a 272 slight decrease in the number of computers, socio-economic differences were small and did not 273 account for the observed differences in total screen-viewing between household education. We found 274 that higher screen-viewing at age 6 was associated with games consoles at the weekends only, but 275 not with multiple TVs or computers, and larger increases in screen-viewing between ages 6 and 9 276 were associated with games consoles and multiple TVs in the week. However, we also saw tablet 277 ownership associated with smaller increases in screen-viewing between ages 6 and 9. This suggests 278 that the relationship between different devices and screen-viewing is complex, and may be changing 279 over time. Thus, we extend previous findings to show that while access to devices may contribute to 280 baseline screen-viewing, it does not entirely explain socio-economic differences in the increase in 281 screen-viewing between ages 6 and 9. 283 There were weak associations between parent and child screen-viewing, with every 30 minutes of 284 parental screen-viewing associated with a 2% increase in child screen-viewing. Parental limits on 285 screen-time were strongly associated with lower screen-viewing at age 6 and less time spent multi-286 screen-viewing at age 9, but associations were stronger for weekdays than weekends and limiting 287 screen time was associated with changes in screen-viewing between ages 6 and 9 on weekdays only.
288 Interestingly, analysis of interviews conducted with a sub-sample of the parents in the study showed 289 that many parents are uneasy about managing non-TV screen-time and feel that they struggle to keep 290 up with rapid technological change [36]. Collectively, these findings may suggest that there is a need 291 to help parents to identify effective ways to manage constantly adapting forms of viewing.
292 293 Tablet ownership did not vary with household education but increased greatly overall between ages 6 294 and 9. A limitation of this project is that we did not assess tablet or smartphone use as a specific 295 behaviour in the questionnaire. This may under-estimate total screen-viewing, especially for those 296 who are heavy users of these devices, and may explain the observed negative association between 297 screen-viewing and tablet ownership. We also note that tablet ownership is strongly associated with 298 higher levels of multiscreen-viewing on weekdays. The widespread ownership of tablets is 299 particularly important for understanding the complexity of screen-viewing, as watching TV, playing 300 games and using the internet can all be done via a tablet and, as such, identifying the behaviour 301 engaged in while using a tablet is challenging. Thus, it may be that in the future there is a specific 302 need to assess tablet time, and to differentiate the various ways in which tablets are used. For 303 example, applications that allow parents and children to agree targets to limit the time spent on 304 games and other non-educational activities, could also be used as a feedback and monitoring device.

ASSOCIATIONS BETWEEN SOCIO-ECONOMIC POSITION AND CHANGES IN CHILDREN'S SCREEN-VIEWING BETWEEN AGES 6 AND 9: A LONGITUDINAL STUDY
Ruth Salway, Lydia Emm-Collison, Simon J. Sebire, Janice L. Thompson and Russell Jago  Table S6-Associations between socioeconomic position and multi-screen-viewing at age 9 Figure S1: Number of minutes spent multi-screen viewing by household education at age 9, for weekdays (left) and weekends (right).

Introduction
Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 5-6 Objectives 3 State specific objectives, including any prespecified hypotheses 6

Study design 4
Present key elements of study design early in the paper 6-7 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, exposure, follow-up, and data Describe any efforts to address potential sources of bias 8-9 Study size 10 Explain how the study size was arrived at 6-7 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, describe which groupings were chosen and why 8-9 (a) Describe all statistical methods, including those used to control for confounding 8-9 (b) Describe any methods used to examine subgroups and interactions 8-9 (c) Explain how missing data were addressed 9 (d) If applicable, explain how loss to follow-up was addressed n/a Statistical methods 12 (e) Describe any sensitivity analyses n/a 28 Methods: Negative binomial regression was used to model associations between socio-economic 29 variables (highest household education and area deprivation) and total screen-viewing at age 6 and 30 the change from age 6 to 9. We additionally adjusted for child characteristics, parental influences and 31 media devices in the home. Multiscreen viewing was analysed separately.

Results
32 Results: Household education was associated with children's screen-viewing at age 6 and with the 33 change in screen-viewing between ages 6 and 9, with higher screen-viewing in lower socio-economic 34 groups. The differences at age 6 were explained by the presence of games consoles, parental limits 35 on screen-viewing, and average parent screen-viewing. Household education differences in the 36 change in screen-viewing between ages 6 and 9 remained.
37 Conclusions: Children's screen-viewing patterns differ by parental education with higher levels of 38 viewing among children living in households with lower educational qualifications. These 39 differences are already present at age 6, and continue at age 9. Strategies to manage child sedentary 40 time, and particularly screen-viewing, may need to take account of the socio-economic differences 41 and target strategies to specific groups. 105 The aim of this paper was to explore socio-economic differences in screen-viewing at age 6, and the 106 change in screen-viewing between ages 6 and 9, and whether any differences can be explained by 107 other factors such as different media devices and parental influences. Developing knowledge in this 108 area will support the creation of targeted behaviour change programmes to reduce screen-viewing.

RESULTS
187 Children's screen-viewing differed between households with differing levels of education at both 188 ages 6 and 9, with higher screen-viewing in lower qualified households, decreasing as education 189 increased, with 35-50 minutes difference between the highest and lowest household education 190 qualifications (Figure 1). Time spent screen-viewing was higher at age 9 than age 6, and higher at 191 weekends than during the week among all education groups, with time spent watching TV and 192 playing games consoles nearly doubling at weekends (Table S1 and (Table S3).  218 adjusts for possible mediators: number of devices, child BMI, parental screen-viewing and parental 219 limits on screen-viewing all measured at age 6. These factors accounted for household education 220 differences on both weekdays and weekends. Children's screen-viewing was higher in households 221 with games consoles, and lower when parents were more likely to limit screen-viewing (Table S4).
222 There were weak associations with parental screen-viewing and child BMI.
223 Models 3 (unadjusted) and 4 (adjusted) examined the change in screen-viewing between ages 6 and 9 224 (   271 overall between the two assessment points, most notably with a large increase in tablet devices and a 272 slight decrease in the number of computers, socio-economic differences were small and did not 273 account for the observed differences in total screen-viewing between household education. We found 274 that higher screen-viewing at age 6 was associated with games consoles at the weekends only, but 275 not with multiple TVs or computers, and larger increases in screen-viewing between ages 6 and 9 276 were associated with games consoles and multiple TVs in the week. However, we also saw tablet 277 ownership associated with smaller increases in screen-viewing between ages 6 and 9. This suggests 278 that the relationship between different devices and screen-viewing is complex, and may be changing 279 over time. Thus, we extend previous findings to show that while access to devices may contribute to 280 baseline screen-viewing, it does not entirely explain socio-economic differences in the increase in 281 screen-viewing between ages 6 and 9. 283 There were weak associations between parent and child screen-viewing, with every 30 minutes of 284 parental screen-viewing associated with a 2% increase in child screen-viewing. Parental limits on 285 screen-time were strongly associated with lower screen-viewing at age 6 and less time spent multi-286 screen-viewing at age 9, but associations were stronger for weekdays than weekends and limiting 287 screen time was associated with changes in screen-viewing between ages 6 and 9 on weekdays only.
288 Interestingly, analysis of interviews conducted with a sub-sample of the parents in the study showed 289 that many parents are uneasy about managing non-TV screen-time and feel that they struggle to keep 290 up with rapid technological change [36]. Collectively, these findings may suggest that there is a need 291 to help parents to identify effective ways to manage constantly adapting forms of viewing.

Introduction
Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 5-6 Objectives 3 State specific objectives, including any prespecified hypotheses 6

Results
179 behaviour on limiting screen-viewing. Models 3 (unadjusted) and 4 (adjusted for covariates) 180 explored the longitudinal change between ages 6 and 9; that is, associations between socioeconomic 181 position and total minutes of screen-viewing at age 9, adjusting for baseline screen-viewing at age 6.

238
239 Multiscreen viewing was associated with both IMD but not with household education, with 240 children's multiscreen viewing higher in families living in more deprived areas (Table S6 and 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  274 overall between the two assessment points, most notably with a large increase in tablet devices and a 275 slight decrease in the number of computers, socio-economic differences were small and did not 276 account for the observed differences in total screen-viewing between household education. We found 277 that higher screen-viewing at age 6 was associated with games consoles at the weekends only, but 278 not with multiple TVs or computers, and larger increases in screen-viewing between ages 6 and 9 279 were associated with games consoles and multiple TVs in the week. However, we also saw tablet 280 ownership associated with smaller increases in screen-viewing between ages 6 and 9. This suggests 281 that the relationship between different devices and screen-viewing is complex, and may be changing 282 over time. Thus, we extend previous findings to show that while access to devices may contribute to 283 baseline screen-viewing, it does not entirely explain socio-economic differences in the increase in 284 screen-viewing between ages 6 and 9.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   14   285 286 There were weak associations between parent and child screen-viewing, with every 30 minutes of 287 parental screen-viewing associated with a 2% increase in child screen-viewing. Parental limits on 288 screen-time were strongly associated with lower screen-viewing at age 6 and less time spent multi-289 screen-viewing at age 9, but associations were stronger for weekdays than weekends and limiting 290 screen time was associated with changes in screen-viewing between ages 6 and 9 on weekdays only.
291 Interestingly, analysis of interviews conducted with a sub-sample of the parents in the study showed 292 that many parents are uneasy about managing non-TV screen-time and feel that they struggle to keep 293 up with rapid technological change [36]. Collectively, these findings may suggest that there is a need 294 to help parents to identify effective ways to manage constantly adapting forms of viewing.

Introduction
Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 5-6 Objectives 3 State specific objectives, including any prespecified hypotheses 6

Study design 4
Present key elements of study design early in the paper 6-7 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, exposure, follow-up, and data Describe any efforts to address potential sources of bias 8-9 Study size 10 Explain how the study size was arrived at 6-7 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, describe which groupings were chosen and why 8-9 (a) Describe all statistical methods, including those used to control for confounding 8-9 (b) Describe any methods used to examine subgroups and interactions 8-9 (c) Explain how missing data were addressed 9 (d) If applicable, explain how loss to follow-up was addressed n/a Statistical methods 12 (e) Describe any sensitivity analyses n/a  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46 Participants 13* (a) Report numbers of individuals at each stage of study-eg numbers potentially eligible, examined for eligibility, confirmed eligible, included in the study, completing follow-up, and analysed Discuss the generalisability (external validity) of the study results 14-15

Other information
Funding 22 Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on which the present article is based 16 *Give information separately for cases and controls in case-control studies and, if applicable, for exposed and unexposed groups in cohort and cross-sectional studies.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46