Back in March 2023, I published a knee-jerk blog post in response to the latest round of media attention on the alleged negative consequences of social media use for teens. Nearly two years later, I am halfway through my PhD, which aims to showcase improvements in how we investigate the mental health impacts of social media. I (spurred on by my impending upgrade viva) decided it was time to pause and reflect on how the field has developed in the interim, and how my project can help us to build a comprehensive evidence base together.
What has held us back?
The term ‘social media’ covers a huge range of online platforms and activities. It also isn’t immediately clear where to draw the line between social media and, say, video-sharing platforms such as YouTube, or instant messaging services like WhatsApp. Of course, if researchers mean different things by the term ‘social media’ in the first place, then they will have very different approaches to investigating their relationship with mental health. But even if we agree what ‘social media’ means, the wide-ranging uses of social media still present a problem. After all, would we expect the mental health influence of browsing thirst traps on Instagram to be the same as that of watching baking videos on TikTok?
Historically, research about social media was limited by precisely this problem: studies large enough to allow us to detect an effect only asked questions about social media in the aggregate, e.g. “How often do you visit a social networking site on the Internet?”. The result is that we could only study the effect of social media as a whole, averaging across everything from puppies to pornography. When this is the only evidence available, there is only one option available for evidence-based intervention, and that is to restrict social media use entirely. Recently, this idea has been floated regularly by journalists, entrepreneurs and policymakers alike. But the evidence on which they base these recommendations is flawed for other reasons, too, and not just because self-reported social media use has been found not to be a good measure of actual social media use.
A pervasive issue in psychology, and one that digital mental health research has not escaped, is that people outside the Global North are seldom included. To this day, many headline-making psychological findings are in fact based only on American psychology undergrads who take part in studies in exchange for course credits. While many studies about the relationship between social media use and mental health are representative of a country in the Global North, there is still no reason to believe that their findings are applicable to those living elsewhere. Yet there still seems to be a lack of interest in supporting researchers in the Global South to produce evidence that is relevant to their own countries.
In the preceding paragraphs I have often referred to the “impact”, “effect” or “consequences” of social media use, but really I should not have. This is because very little, if any, of the existing research is actually capable of determining whether social media use is actually a cause of changes in mental health. In the same way that it seems unlikely that excessive margarine consumption is the major driving factor behind divorce in Maine, it is not clear that a correlation between increasing use of social media and worsening adolescent mental health means that the former causes the latter. While we often do adjust analyses to take other factors into account, it is usually impossible to account for all the factors (“confounders”) that could cause the two to appear more related than they are.
In pharmaceutical research, the standard way to get around the issue of spurious correlations is to conduct a randomised controlled trial. However, attempts to export this approach to social media research have their own problems. Let’s say I sign 100 participants up to a study, where 50% will give up social media for a few months and 50% will continue to use it as normal. I can assign the participants to a group at random, so that it’s not just the ones who feel they have a deeply unhealthy relationship with social media electing to give it up. However, the placebo effect is a problem for psychologists, too. Given the regular news headlines about how social media is harmful, those who give it up may feel that their mental health ought to be improving, and often this will result in their mental health improving. In addition, this sense that they ought to be feeling better might influence how they fill in questionnaires that measure their depression, anxiety and other symptoms.
All of these challenges have led to an unclear picture of the actual effect of social media use on adolescent mental health. And when we add the fact that studies reporting statistically significant results are more likely to published, the evidence starts to look very weak indeed.
Is it really a problem?
A question I often hear is “Surely if we just get kids off social media, it won’t do them any harm, so does it matter that the evidence for doing so is a bit weak?” But if there is limited evidence for social media being harmful, there is even less to support the idea that taking social media away is not harmful. Importantly, withdrawing social media is likely to affect most negatively the most vulnerable young people: those who are marginalised, those with unsupportive families, those with existing mental health conditions etc. For those people, the “harmless” intervention of removing access to social media becomes the removal of a lifeline, a means by which they can build a community and a safe space for them to exist as they are. With increasing knowledge about the interactions between social media use and mental health comes the opportunity to suggest more targeted interventions that will allow a safer online environment. The gaps in our understanding of social media are therefore an issue of mental health equity, for underprivileged individuals in the Global North but also for the entire population of the Global South, about whom there is very little research available.
What am I doing about it?
I am not blessed with the power to solve all the challenges in this field on my own. In fact, all of the research in my PhD thesis will commit at least one of the sins I have identified, and probably more. But it will also show how the latest advancements in the ways we investigate social media can address some of these problems.
In my first study, I used data from a survey conducted by UNICEF Innocenti – Global Office of Research and Foresight, asking 12–17-year-old children in 12 countries in the Global South about their online activities. I was particularly interested in whether there was a clear relationship between internet use and internet-mediated communication and mental health symptoms at the time of the survey. There was not (preprint, not yet peer-reviewed), and the results showed how much more research outside the Global North is needed.
I am currently analysing data from the UK’s Millennium Cohort Study for my second study. Here I am interested in whether use of secondary mental health services by adolescents in England was associated with prior use of social media, using an NHS healthcare records linkage. To minimise the bias that creeps into my findings, I will be adjusting for as many potential confounders as possible, and using extremely recent statistical developments (warning: technical content!) to handle cases where participants have missing data.
Still to come: I will be diving into some of the methods that are being developed to move beyond overall usage time as a measure of social media use. These include regular short surveys on a smartphone app that people can complete as they go about their daily lives, apps that take periodic screenshots, eye tracking, touch tracking and many other exciting innovations.
I am extremely fortunate to have had the opportunity to work with a diverse range of researchers and with advisors with lived experience of mental health difficulties, who have guided and will continue to guide me through my PhD and contribute to the research that will go into my thesis. With their help, I hope to be able to make a meaningful contribution to improving our understanding of young people’s relationships with social media.
How do you solve a problem like measuring social media? by Tom Metherell is licensed under CC BY 4.0
