Written evidence submitted by the University of Sheffield (SMH0121)
THE EFFECT OF SOCIAL MEDIA USE ON CHILDREN’S WELLBEING
Emily McDool, Philip Powell*, Jennifer Roberts, Karl Taylor
Department of Economics, University of Sheffield, UK
1 Summary
1.1 Our evidence addresses the following questions: (1) What evidence there is on the effects of social media and screen-use on young people’s physical and mental well-being – for better and for worse – and any gaps in the evidence; and (4) the physical/mental harms from social media use and screen-use, including: safety online risks, the extent of any addictive behaviour, and aspects of social media/apps which magnify such addictive behaviour.
1.2 The research we refer to in this piece of evidence was generated as part of an Engineering and Physical Sciences Research Council (EPSRC) funded project on ‘digital empathy’. Contributors to this evidence are Dr Emily McDool, Dr Philip Powell, Professor Jennifer Roberts, and Professor Karl Taylor from the Department of Economics, University of Sheffield, UK.
1.3 We focus on one important aspect of children’s internet use, the use of social media or social networking. We estimate the effect of children’s online social networking on their subjective wellbeing. We use a large sample of 10-15 year olds from the UK Household Longitudinal Study, and estimate the effect of time spent chatting on social websites on a number of outcomes that reflect how happy these young people feel about different aspects of their lives, specifically: school work, school attended, appearance, family, friends, and life as a whole.
1.4 We use data measured over time and a special statistical method called ‘instrumental variables’, using matched data on local broadband connection speeds from Ofcom, to produce causal estimates of the effect of chatting online on happiness, not merely associations.
1.5 Our results suggest that spending more time on social networks reduces the happiness that children feel with all aspects of their lives, and that the effects are worse for girls than boys. Our most stringent statistical model suggests that a 1 standard deviation increase in the time spent using social media leads to a 0.38 standard deviation reduction in how happy children feel about their life overall. This effect is not trivial, being approximately half the size of the effect on young peoples’ happiness of living in a single parent household. Reporting reduced happiness on our indicators is associated with greater socio-emotional difficulties in young people[1].
1.6 Considering the different aspects of life measured in the study, the largest effects are for happiness with family and school attended and the smallest effects are for appearance and school work.
1.7 We explore three explanations for our results. We find particularly strong detrimental effects for those on a lower income, who report being bullied, and those who report undertaking a greater number of additional activities.
1.8 Our results suggest that policy initiatives designed to reduce the time young people spend on social media would have positive effects on wellbeing.
2 Background
2.1 Childhood circumstances and behaviours have been shown to have important persistent effects in later life. One aspect of childhood that has changed dramatically in the past decade is the advent of social media, or online social networking. Young people are heavy adopters of social media; today’s teenagers are the first cohort to have grown up with online social networking. An ONS survey in 2015 revealed that, in the UK, 92% of 16 to 24 year olds use online social networks[2]. Along with teenagers, younger children are also increasingly users of social media; while most sites stipulate a minimum user age of 13, few require any validation, and a survey for the children’s BBC channel (CBBC) found that more than three quarters of 10 to 12 year olds have social media accounts[3].
2.2 A lot of the existing evidence on the effects of social media use comes from small selective samples from outside of the UK, so it is difficult to generalise this to children in the UK. In contrast our evidence comes from a large representative sample (n = 5171) of children from across the UK.
2.3 Much of the existing evidence establishes an association between social media use and wellbeing but is not able to claim that the relationship is causal. Associations can be misleading as they may be driven by other factors. Our method comes as close as possible to establishing a causal effect in the absence of experimental methodology.
3 Where our evidence comes from
3.1 We use a large secondary data source, the UK Household Longitudinal Study (UKHLS), often called the Understanding Society survey. This is secondary data, meaning that it was not collected specifically for the purposes of studying children and social media use.
3.2 UKHLS is a study of 21st century UK life and how it is changing. It captures a wide range of information about people’s social and economic circumstances, attitudes, behaviours and health. It is funded by the Economic and Social Research Council (ESRC) with additional funding from a consortium of government departments.
3.3 UKHLS is a representative sample of over 40,000 households across the UK; the same individuals and households are interviewed in each wave. Six waves of data are available; starting in 2009-2011 (wave 1), which provided data on over 50,000 individuals. In wave 6 (2014-2016), over 45,000 individuals were interviewed. All adult members of each household are interviewed along with children in the households aged 10 to 15 years old.
3.4 Our data is from children interviewed in waves 2 to 6; data comes from a face-to-face interview and a self-completion questionnaire. We also make use of information from the adult interviews so that we can include household and family circumstances in our analysis. Our analysis uses a sample of 5,171 children, providing 10,039 observations between waves 2 to 6 of UKHLS.
3.5 The outcomes are measures of domain satisfaction; children are asked how they feel about different aspects of their life, specifically: school work, school attended, appearance, family, friends, and life as a whole. The possible responses are on a 7-point scale ranging from 1 = not happy at all, through to 7 = completely happy. Average happiness is highest for family (6.32) and lowest for appearance (5.20).
3.6 The main explanatory variable is obtained by firstly asking: Do you belong to a social web-site such as Bebo, Facebook or Myspace? 73% of respondents answer ‘Yes’; they are then asked How many hours do you spend chatting or interacting with friends through a social web-site like that on a normal school day? The responses are: 1 = none (11.6%), 2 = less than an hour (44.8%), 3 = 1-3 hours (32.8%), 4 = 4-6 hours (7.8%), and 5 = 7 or more hours (3.1%).
4.1 We use regression analysis to estimate the effect of the time that children spend chatting on social media on their happiness with the different domains of their life, controlling for other variables that are also expected to influence this happiness.
4.2 The main control variables are: children’s characteristics (age, gender); parent and household characteristics (parental employment and education, single parents, household income, housing tenure, other children in household); and local area characteristics (local unemployment rate, share of females, share of population over 65, share of population aged 16-65).
4.3 Our estimation method is designed to deal with some important methodological issues (see below), so that we get as close as we can to estimating a causal effect of social media use on wellbeing, rather than simply establishing an association.
5.1 Our results show that spending more time chatting on social networks reduces the happiness that young people feel with all aspects of their lives (school work, school attended, appearance, family, friends, and life overall).
5.2 The quantitative interpretation of our models is hard to summarise because our main explanatory variable (time spent chatting on social networks) and our outcome variables (how the children feel about different aspects of their lives) are measured on ordinal scales (a reported rating from 1 – 7). Here we interpret the effect on outcomes evaluated at a 1 standard deviation increase in the level of the explanatory variable.
5.3 A 1 standard deviation increase in the time spent chatting on social networks reduces the happiness reported with life overall by 0.38 standard deviations. This is approximately half the size of the adverse effect on happiness of living in a single parent household, and thus should not be considered a trivial effect.
5.4 Looking at different aspects of life, the largest effects are for happiness with family (1: −0.56 SDs) and school attended (1: −0.52 SDs), while the smallest effects are for appearance (1: −0.12 SDs) and school work (1: −0.06 SDs).
5.5 Looking at boys and girls separately reveals some differences. Girls experience more adverse effects from social media use than boys, particularly in domains such as appearance and friends.
6.1 We explore three theories that help to explain why social media use may have a negative effect on young people’s wellbeing. These theories draw on research from both economics and psychology, and it is likely that they are not mutually exclusive.
6.2 Social comparisons: increased social media use is linked to more frequent social comparisons with others; these comparisons are more likely to be negative in direction, given that the material people choose to present online represents selectively idealised versions of their true lives. We explore the effects of time spent on chatting on social media for children with high vs. low self-esteem (using a psychological measure called the Rosenberg self-esteem scale), and for those with personal income above or below average (relative income has been shown to be important for social comparisons). While there is no evident self-esteem effect, there are more adverse effects for those with below average personal income, which provides some support for the social comparisons theory as those with lower relative income are more prone to make negative social comparisons.
6.3 Finite resources: extensive time spent on social media encroaches on other activities known to be beneficial for wellbeing, such as face-to-face socialising, sports or exercise. We explore the effects of time spent on chatting on social media for children with high vs. low participation in other activities (such as going to the cinema, watching sport, or ‘hanging out’ with friends). Our research found that, contrary to this theory, there are more adverse effects for those with higher involvement in other activities. Young people with greater time commitments elsewhere may experience a more negative effect of increased time chatting online on their happiness, due to increased time pressures.
6.4 Cyberbullying: young people who spend more time on social networks have a greater chance of being the victim of cyberbullying. We explore the effects of time spent on chatting on social media for children who report being bullied (this is the general experience of being bullied, not cyberbullying per se) vs. those who say they are not bullied. There are more adverse effects for those who report being bullied, which provides some support for the cyberbullying theory.
7.1 An important methodological issue relates to the direction of causality of the relationship between wellbeing and social media use. We have stated here that social media use predicts changes in wellbeing, but it can also be argued that causality may go in the opposite direction because children with lower levels of psychological wellbeing may choose to spend more time on social media. It is also possible that there is a ‘third variable problem’; meaning that there are factors not included in our analysis (for example loneliness or introversion) that drive both social media use and wellbeing. Failing to account for these factors may result in misleading estimates of the effect of social media use on wellbeing.
7.2 In this paper we address the ‘third variable problem’ using a rich set of control variables (including child, parent and local area characteristics) and using an estimation technique that controls for time invariant individual characteristics that we cannot observe (like personality traits).
7.3 To deal with the problem of direction of causality, we use an instrumental variables (IV) approach. An instrument is a variable that could only affect the outcome measure (wellbeing) via its influence on the main input (social media use). We use information on local area broadband speeds published by Ofcom. The basic premise here is that the quality of internet connection should have no direct effect on how young people feel about the aspects of their lives considered in the study (once we control for other key influences such as household income and the local economy) but will only affect them via its influence on time spent online; empirical tests support the validity of this IV strategy.
7.4 Our estimation method is also designed to deal with two important features of the data; firstly, the outcome measures (how young people feel about various aspects of their life) are not continuous; they are measured on a 7-point ordinal scale. Secondly, we have repeated measures from some children who appear in more than one wave of the data.
7.5 Initiatives to reduce the length of time young people spend chatting on social media on a typical school day is likely to have a positive impact on wellbeing. This is particularly true for certain groups, including: girls, those with a below average personal income, those with additional time pressures, and those who report being the victim of bullying.
April 2018
Funding Acknowledgement: This research is part of a project funded by the UK EPSRC research grant EP/L003635/1 Creating and Exploring Digital Empathy (CEDE).
[1] http://data.parliament.uk/writtenevidence/committeeevidence.svc/evidencedocument/health-committee/children-and-young-peoples-mental-healththe-role-of-education/written/45633.pdf
[2] www.ons.gov.uk/ons/guide-method/method-quality/specific/business-and-energy/e-commerce-and-ict-activity/social-networking/index.html
[3] www.bbc.co.uk/news/education-35524429