Link between teens’ time on digital devices and lower wellbeing is “too small to merit substantial scientific discussion”
A new study aims to set new standards for research into the time that children and young people spend in front of screens.
31 May 2019
My friends and I would often be so hooked on the latest Sega Mega Drive video game that we’d play all day long, breaking only for munchies or when nature called. Our parents would urge (plead with) us to get outside, especially when it was sunny. “The fresh air and exercise will do you good”, they would say, or similar. Fast forward to now, and the anxiety over all the time that children and young people spend in front of screens, be it playing video games, watching TV or using social media, has of course only intensified. Surely it can’t be mentally or physically healthy, can it?
As we look to psychologists to provide an answer, we find a field divided. At one extreme, some experts point to survey data throwing up apparently worrying correlations between increased screen time and increased mental health problems. Yet other experts are sceptical, in part because of what they see as the poor quality of much of the correlational evidence for harm.
In this latter camp are Amy Orben and Andrew Przybylski at the University of Oxford, the authors of a recent paper in Psychological Science, which aims to set new standards for research in this area – including by using time-use diary-based reports of screen time (rather than relying on notoriously unreliable retrospective reports), and by pre-registering their methods and hypotheses, thus guarding against the kind of post-hoc data-mining that they say has plagued the field.
The pair began with an exploratory analysis of two large surveys of thousands of teenagers, one based in Ireland, the other in the USA. These included data on teens’ mental health and wellbeing (including mood, depression and self-esteem) and their screen time (including time on TV, computer, and video-games and smartphones). The screen time measures included commonly used retrospective reports, such as “how much time do you typically spend during a weekday watching TV, playing video games etc?”, but crucially also detailed diary-based measures, in which each participant broke down their precise activities during every 15-minute period of a given day (either completed through that day, in the evening or the next day).
Orben and Przybylski found a handful of statistically significant correlations between screen time and wellbeing, and then they used these to make some specific evidence-based, pre-registered (written down publicly in advance) predictions about the kind of screen time–wellbeing associations they were looking for in a second, “confirmatory study”. This is a robust methodological approach that avoids the pitfalls of scouring a large data set hunting for all and any significant correlations (such post-hoc result hunting carries the risk that any associations that do turn up are flukes rather than meaningful).
The survey data Orben and Przybylski used in their confirmatory study involved over 10,000 British teenagers (aged 14 to 15) and their caregivers, who once again provided retrospective and diary-based information on their screen time and completed wellbeing measures.
The researchers found a few statistically significant correlations, including: between greater self-reported screen time and lower wellbeing; and greater diary-recorded screen time and lower wellbeing. Based on some mixed results from the exploratory work, the pair also looked specifically at screen time prior to bed, but this was not found to be associated with wellbeing. Critically, the significant associations that Orben and Przybylski did find were very weak – in fact, of a magnitude “too small to merit substantial scientific discussion”.
The pair provide a graphic illustration of the average size of the associations between screen time and wellbeing that they found. Assuming that causality flows in the direction of greater screen time to poorer wellbeing (the research is cross-sectional so we don’t know if this is the case), then given the size of the association, teenagers would have to increase their screen time use by over 63 hours per day for it to have an effect on their wellbeing that they would actually notice (this is based on what is known from past research about the kind of wellbeing changes that are subjectively noticeable). Of course such an increase is a practical impossibility, thus showing the practical insignificance of the documented correlations.
While aiming to set new standards for research in this area, the researchers admit their study has its own limitations – among them that the data on screen time and wellbeing were not collected at the same time, potentially reducing the chance to identify meaningful relations between the two. Also, some may wonder whether total screen time (however carefully it is measured) is the right metric – perhaps it is too crude and it is the nature of one’s relationship with digital devices that is more relevant to wellbeing, such as why they are used, whether the screen time displaces other meaningful activities, and if there is a compulsive quality to the usage or not.
—‘There are wolves in the forest…’ Professor Andrew Przybylski picks three myths around screen time.