May 7, 2019

expert reaction to social media use and life satisfaction in adolescents

A study published in PNAS analyses possible links between social media use and life satisfaction in adolescents.

Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“This is a very interesting study. Many researchers have previously looked for associations between social media use and measures of life satisfaction, happiness, and so on in young people, but no overall clear picture has emerged. This new research throws considerable light on why that might be. It shows that the picture is complicated, certainly can’t be summarised by saying that too much social media use makes young people miserable, that the findings depend to some extent on the choices made in analysing the data, and that more data and more research are required to make sense of all this subtlety. The researchers also argue strongly that researchers need access to much better data on social media use from social media companies. I couldn’t agree more.

“Almost all previous investigation of these questions has been based on observational data, where social media use is recorded in some way in some group of people, their well-being or happiness is also recorded, and the researchers see to what extent these might be associated. This method of study has many potential problems. There are likely to be many other differences, on average, between young people who use social media a lot and those who don’t, and these other differences (known as confounders or control variables) might be what’s driving any observed association, and not the social media use itself. So these studies can’t establish clearly what causes what. It’s possible to adjust the results statistically to try to allow for these other differences, but that can be done only for measurements where the researchers have data, and there are choices to be made about what to take into account and how.

“The new research has two very important strengths. First, unlike most previous work, it uses longitudinal data that followed up young people over time. So it can show whether a change in social media use came before or after a change in life satisfaction. That tells us a little more about what causes what – if a change in life satisfaction happens some time before a change in social media use, then the social media use can’t be causing the change in life satisfaction. Second, rather than making just once choice about which confounders to adjust for, the researchers investigated a wide range of choices of adjustments and other aspects of the statistical analysis. They found that these choices do make a difference to the findings, but that there were no very clear strong associations between social media use and life satisfaction, whichever choices were made. Instead the picture was more subtle. There was some evidence of social media use predicting life satisfaction at a later time, and also some evidence of life satisfaction predicting something about social media use later. But these possible effects were all small or even negligible, and often there was some evidence of differences between males and females.

“The use of longitudinal data helps, but the data are still observational, so it still doesn’t get over the issue that you can’t establish what causes what. For instance, in females, there was evidence that those who used more social media had slightly lower life satisfaction (in several respects) at a later time. But again, this doesn’t show that it was the social media use that caused decreases in life satisfaction. Perhaps something else tended to cause girls to use social media more, and later caused them to have lower life satisfaction, in a way that would have happened even if they hadn’t used more social media. Perhaps this other difference wasn’t recorded in the data. But, in my view, if there was any very clear and consistent causal relationship between social media use and life satisfaction, it would have shone clearly through this impressive new data analysis – and there simply isn’t any sign of anything like that. I very much agree with the researchers when they say that things are more subtle, and that any associations are weak and, quite possibly negligible.

“An unavoidable limitation of this research, and indeed with almost all previous research, is that it uses the participants’ own reports of their use of social media. That might differ from their actual use, possibly in a way that depends on life satisfaction.  But nobody can get past that restriction without more accurate, less subjective data on media use, and that can only come from the social media companies. I hope that this new research can persuade them to get independent researchers involved in working with their data – in appropriately ethical ways, of course.”

Dr Max Davie, Officer for Health Improvement at the Royal College of Paediatrics and Child Health (RCPCH), said:

“At a time when there is limited high quality research in this area, this latest study is very welcome. But as the researchers themselves say, it is the first small step and more data specifically from industry is needed, if we are to learn more.

“This paper suggests that social media has limited effect on teenage life satisfaction. However, there are still issues around screen time more generally, and a risk that screen time may interfere with other important activities like sleep, exercise and spending time with family or friends.  We recommend that families follow our guidance published earlier this year and continue to avoid screen use for one hour before bed since there are other reasons beside mental health for children to need a good nights’ sleep.”

‘Social media’s enduring effect on adolescent life satisfaction’ by Amy Orben et al. was published in PNAS at 20:00 UK time on Monday 6th May.

Declared interests

Prof Kevin McConway: Prof McConway is a member of the SMC Advisory Committee, but his quote above is in his capacity as a professional statistician.

None received.