Safeguarding Podcast – The Central Drama of Life with Prof Yvonne Kelly

By Neil Fairbrother

In this safeguarding podcast we discuss the relationship between social media use and adolescent mental health with Professor Yvonne Kelly of University College London. Based on 11,000 14 year olds from the Millennium Cohort study, the results of this analysis show clear pathways from social media usage to depressive symptoms of mental health. We also discuss how this can be averted, what industry at large can do and how new forms of algorithms could help.

http://traffic.libsyn.com/safetonetfoundation/SafeToNet_Foundation_podcast_-_the_Central_Drama_of_Life_with_Prof_Yvonne_Kelly.mp3

There’s a lightly edited for legibility transcript below for those that can’t use podcasts, or for those that simply prefer to read.

Neil Fairbrother

Welcome to another edition of the SafeToNet Foundation’s safeguarding podcast where we talk about all things to do with safeguarding children in the online digital context.

The online digital context comprises three areas, technology, law and ethics or culture with child safeguarding right in the centre of this Venn diagram, and it encompasses all stakeholders between the child using a smartphone and the content or the person online that they are interacting with.

Today’s podcast is focusing on the effect that social media usage can have on the mental health of children, and to guide us through this topic I’m joined by Professor Yvonne Kelly, Director of the ESRC International Centre for Life Course Studies in Society and Health of the University College London.

Thank you for inviting me over Yvonne. Could you provide us with a brief resume of yourself please and tell us a little bit about what your team does?

Yvonne Kelly

I am Professor of Life Course Epidemiology here at UCL, and within this Department of Epidemiology and Public Health, I run a research group, a team called ICLS that you’ve just introduced. What we do in that group is we look at all stages of this thing called the “Life Course”. We follow people really from cradle to grave, so we have different points in the Life Course that we focus on. We look at children and young people, we look at people of working age and we look at people post retirement.

The work that we’re going to talk about today is part of a large program of work that myself and colleagues are doing, looking at young people and different aspects of their mental health, the physical growth and the learning, their educational attainment.

Neil Fairbrother

Okay. Thank you for that. Now, earlier this year you published a study that you worked on with some of your colleagues; Afshin Zilanawala, Cara Booker and Amanda Sacker. The report was called “Social Media Use and Adolescent Mental Health: Findings from the UK Millennium Cohort Study”. What is the UK Millennium Cohort study? That sounds very interesting.

Yvonne Kelly

This is one of the UK’s real jewels in the crown, as they’re known. It’s one of the birth cohorts studies. There are several of these, in this one all participants were born in the years spanning 2000 to 2002 and on the whole there’s upwards of 19,000 participants in that study. They’d been seen every couple of years, so they were seen when they were babies and then ages three, five, seven, 11, and 14. And there are new data, which are going to become available to us next year from when participants were aged around 17 years.

And those data are very rich. They collect information on a wide variety of aspects of the young person’s life. And you know, because we have the information on the same individuals right throughout this early part of their life course, they’re really very powerful. They’re a very rich source of data for any number of research topics including the use of social media in relation to mental health.

Neil Fairbrother

Okay. Now, you mentioned that this was one of the UK’s crown jewels. That implies this at least unusual, if not unique. Are there other countries doing similar things?

Yvonne Kelly

There are these birth cohort studies in other parts of the world. So the US, Australia, different European countries particularly Norway, Denmark, there are a wide range of settings globally in which these longitudinal studies are taking place.

The crown jewels reference comes from the fact that the first of these birth cohorts was set up in the UK in 1946 and then there was another one in 1958 and another in 1970. So the UK has this long track record going back over several decades in this kind of study design and the kinds of research that can be done on studies of this calibre. This particular study, the Millennium Cohort Study is run out of the Institute of Education by the Centre for Longitudinal Studies, which is also part of UCL.

Neil Fairbrother

Okay. So let’s cut to the quick. Is there a relationship between social media use and adolescent mental health?

Yvonne Kelly

Well, there is an association between the amount of time that people spend online and some markers of their mental health. We have to be really careful there and I’m guessing that you’ve heard of this before around cause and effect because when we look at information as a sort of a “snapshot” in time, we look at people’s reported social media use and their mental health at the same time, we don’t know which way around that association is going.

We don’t know whether people who have low mood or who are feeling unhappy, feeling sad, hopeless, we don’t know if they’re more likely to spend more time online or whether it’s the time that they spent online, which somehow is making them feel miserable and sad. So there’s that whole problem or challenge of causality really at the heart of this, but what we can say is that if we are hypothesizing that more time online is linked to poor mental health than why might that be?

Because, for example, you have one young person who would spend three hours online and would be having an absolutely fantastic time. Think of all the benefits of social media, using social media platforms, they are many fold; information seeking, socializing, contact with people, family who are in different parts of the world, friends all over the world, lots and lots of benefits to be gained from spending time on social media platforms.

So you take one 14 year old who spends three hours a day and they’re having a fantastic time and then the next 14 year old spends three hours a day online might not be having such a good time because they might be encountering things which aren’t so good for their mental health.

So I think for us, one of the main things that we have to always try to ensure as part of what we’re saying about this is that, yes there’s an association between the amount of time, but that is to do with the content. So what people are experiencing, good experiences, not so good experiences and the context of that use. Is it taking time away from other things, like getting enough sleep, spending face-to-face time with people, getting enough exercise and so on?

Neil Fairbrother

Okay. Now you refer to depressive symptoms in your work, what symptoms did you find?

Yvonne Kelly

Young people themselves answered questions about different aspects of how they were feeling, feelings of sadness, hopelessness, difficulty functioning in everyday life and so on. The kinds of things where when people are being assessed in a clinical setting, the sorts of questions that clinicians would be asking them would be trying to assess how they were doing on a number of aspects related to what we think of as depressed mood.

The questions that they were asked were standardized, for the technical people listening, as the “Short Form” of the Moods and Feelings questionnaire, which has been widely validated in many different studies across the world. So this was taken by the Millennium Cohort Study team and included in the questions that they asked of the young people when they were aged 14.

Neil Fairbrother

Just in terms of scale then, the study is just under I think 11,000 14 year olds. The minimum age for being on social media without parental consent is 13. Did the adverse signs you identified actually start when the children were younger than 14 and if so how long does it take for these effects to actually kick in?

Yvonne Kelly

That’s a really difficult question. For these young people, there was information about their depressive symptoms at age 14 and their social media use age 14, but we don’t have information about their social media use in terms of numbers of hours from earlier ages. So with this particular data you can’t get that before and after sense of what’s going on.

We do have information related to depressive symptoms from when they were aged 11. So we were looking at something called “Internalizing Symptoms” and they relate quite closely to feelings of depression and in our study we were able to take account of those in a statistical sense. So we were looking at the association between social media use at age 14, depressive symptoms at age 14, net of how they were feeling three years previously in terms of the sort of depressed mood.

So from this particular study we can’t say when the association between social media use and mood would kick in because we simply don’t have the information going back on their use earlier.

Neil Fairbrother

Nonetheless you were able to go back in time, because it’s a longitudinal study, you are able to go back in time and make use of historic data of when these same children were younger and there’s a correlation at least.

Yvonne Kelly

Yes, there’s a strong correlation between prior mood and current mood for children and for adults, at what we call the population level because poor mental health problems typically emerge during the adolescent period.

Neil Fairbrother

And why is that? Why do poor mental health conditions happen during adolescence? Is it because the adolescent mind is still developing and is more fragile, more delicate than an adult?

Yvonne Kelly

When I’m teaching about adolescents, my opening gambit is that adolescence is the central drama of life. We all remember it. In some ways we are maturing emotionally, maturing physically, physiologically, the hormones are raging, our social relationships are changing and morphing, our relationships with our family. Typically during the adolescent period, we’re becoming more independent of family, more interactive and perhaps dependent on peer relationships. There’s so much going on.

It’s a well-known phenomena that then problems to do with mental health start to emerge in the adolescent period. The precise reasons around that I don’t think of a particularly well understood, but it is certainly a period of great turbulence in the life course. Something like three quarters of lifetime mental health problems will emerge during the period of adolescence.

Neil Fairbrother

Given that, and given the law around which this minimum of age of 13 is based, the US COPPA law, is being reviewed at the moment, the US legislative organization is open for suggestions. So should one suggestion be that this minimum age should be increased to 16, for example, when perhaps some of this turmoil of the developing mind has been and gone, so to speak.

Yvonne Kelly

Ah, well, as we understand more about adolescents, actually, the upper age range for it is arbitrarily set between 10 and 25 or 30. So 16 isn’t that much further on! Of course there’s a huge difference between a 14 year old and a 16 year old, but it isn’t all done and dusted by age 16. For a lot of people that isn’t done and dusted in the early twenties, when many people are entering the job market and thinking about longer term plans, perhaps partnerships and family formation and so on.

So this period of adolescence really stretches forward. Brain developments, the key aspects of brain development to do with planning, executive function and so on really aren’t finished until we’re in our twenties.

Neil Fairbrother

Okay. You refer to online harassment in your report, but online harassment is a phrase I think that covers a multitude of sins. What do you mean by online harassment?

Yvonne Kelly

Well, we had an interesting set of discussions around this because these times are quite charged and loaded in many ways. The participants in this study would report either being the victim of negative messaging in the online environment or the perpetrator of that – sending unpleasant messages to people, for example.

So we chose the phrase “online harassment” rather than cyberbullying because it was felt that it was more appropriate. People could be on the receiving end or they could be actioning these kinds of negative interactions themselves, or they could be both on the receiving end and involved in the perpetration.

Neil Fairbrother

Did you find any difference between the senders and receivers? Was there a difference in terms of the volumes of messages being sent and received?

Yvonne Kelly

There were fairly equally balanced I guess, though people who reported both being a victim and the perpetrator of this kind of activity had the worst mental health.

Neil Fairbrother

Now you say that youth mental health is a major public concern, which poses substantial societal and economic burden globally. And that sounds very much like was known as ACE or Adverse Childhood Experiences. Is there a relationship between this study and what’s known as ACE?

Yvonne Kelly

ACEs are a really huge and emerging field of research and concentrate really on many of the adversities that young people could experience in those very early years of life, right throughout the early parts of life. But they would typically be to do with what was happening in the family, in the context of the family. Parents having poor mental health, perhaps substance misuse, parents’ incarceration, parents being in prison, experiences of abuse during childhood. I think this is quite separate from that.

But of course, the whole area of first childhood experiences in relation to young people’s mental health is a very, very live research area. And actually one of member of our team is carrying out her own research program, looking at ACEs in relation to mental health.

Neil Fairbrother

Okay. Well, I look forward to reading that report, and maybe there’s another podcast there! Your study is unusual in that you attempt to simultaneously examine multiple potential pathways of young people’s mental health. Most studies have a controlled environment and you change one variable, but you’re looking at multiple pathways here. So did that pose any particular problems for the research and the analysis?

Yvonne Kelly

I think that was really exciting for us because we had the data available, we had the information on these potential intervening pathways, so we were looking at online harassments, we were looking at sleep, we were looking at self-esteem and perceived body image. There’ve been lots of studies which have looked at mostly one, maybe two of those pathways at a time, but we had the opportunity to look at all four and we wanted to do that because we wanted to see if there was an association between social media use and mental health. Which of those pathways could be the main candidates, if you like, in that association?

From a statistical point of view, a technical point of view, yes you have to obviously be thinking very clearly about what you’re going to test and how you’re going to test it. But that’s the kind of work that we do in our research group. So for us it was, it was reasonably straightforward.

Neil Fairbrother

Okay. Now the statistical analysis section becomes quite technical for a lay person anyway. You talk about using multivariable linear regression models, adding and removing variables in separate blocks of adjustment to try to link markers of depressive symptoms with social media use. Is there any way that you can put that into layman’s language for the man in the street, the man on the London omnibus, to understand?

Yvonne Kelly

Yes. So the multi-variable part, is saying that we’re looking at all of these variables, we’re controlling for many, many variables to do with the child’s background. So we’re looking at different aspects of their socioeconomic profile and things going on with the family, prior mental health problems as we mentioned before. So we’re adjusting for that.

We’re holding all of those things constant in a statistical sense and we’re looking at the association of social media and mental health and they we’re saying, okay, if we take account of how their sleep might be associated in this association, what happens to that relationship between social media use and depressive symptoms? So that’s one block, that’s one battery if you like: sleep.

And then we say, okay, let’s move on from sleep to online harassment and again we will focus on that part of the pathway and then so on and so forth.

When we’ve looked at all of those separate pathways and tried to assess the potential importance of each of those pathways for this association between social media and sleep, then we can include all of those different pathways at the same time and then say, okay, which if any, might be associated more or less strongly with social media use and with depressive symptoms than the others?

And then we can end up trying to rank in this association which of these pathways, these potential pathways, might be most important. And for us in our paper, in this particular study, we found that sleep, so that’s the quantity and the quality of sleep and experiences in terms of online harassment, both as a victim and a perpetrator, were the most important potential pathways between social media use and depressive symptoms.

Neil Fairbrother

Okay. So let’s talk the results in a little bit more detail. So just to summarize, I think you say that girls use social media more than boys, twice as much, in round numbers. And maybe that’s not that surprising, for some reason I didn’t find that particularly surprising.

Twice as many boys than girls report not using social media. Girls are more likely than boys to have low self-esteem, but that difference isn’t as much as some people might think because there was only, I think, four percentage points difference in that result. Similarly with appearance, how satisfied they are with their appearance. Again, it’s roughly four percentage point difference, which again was a surprisingly small difference between girls and boys, for me.

Body weight dissatisfaction. The difference is more here I think, with 10 percentage points between the girls and boys, which is probably a reflection of societal norms these days or media norms.

But the big one for me, the big surprise is that girls are more likely to be involved in online harassment as a victim or as a perpetrator with nearly 14 percentage points difference between boys and girls. Why is that such a big factor? Why are girls more likely to be the victim or perpetrator of online harassment?

Yvonne Kelly

I don’t think it’s girls per se, but girls are spending more time online, so the probability of them encountering negative activities both as perpetrator or as a victim just multiplies up.

Neil Fairbrother

So it’s a simple function of time?

Yvonne Kelly

I think if they are more likely to be online, more girls are using social media platforms and more girls than boys are using social media platforms for longer, so there’s something there. I don’t think it’s something particular to girls, like girls are more likely to be bullying people compared with boys for example. But it might be something just to do with the online environment, that girls are spending more time that and so simply more likely to come across things which are maybe not so good for them.

It didn’t surprise us that boys and girls were having this same or similar self-perceptions in terms of their body image. I think we expect that girls care more about the way they look. But actually the experiences of young boys, 14 year old boys because of the norms, quite often we don’t think they are affected by pressures of those sorts but really I think they are, and I think the data show that quite clearly.

Neil Fairbrother

Yes and does this time dimension also explain why on average girls have higher depressive symptom scores compared with boys?

Yvonne Kelly

On average girls have high depressive symptom scores, without looking at social media. If you take social media out of the picture, you’ve got 25% of girls having clinically relevant depressive symptoms and you’ve got about 10% of boys and that difference is staggering. It’s not clear why 14 year old girls are experiencing a more depressed mood compared to boys, but it’s a widely reported phenomena that depressive symptoms are more common in this age group of girls, compared with boys.

The factors behind that are going to be multiple if, and it still is an if, what’s happening to young people online in terms of social media use is a factor. It’s only one of very many things to do with the context that young people are growing up in.

Neil Fairbrother

Okay. Now you do draw attention to some limitations of this report. Could you just expand a little bit about what you see as the limitations of this report?

Yvonne Kelly

Well, we talked already a bit [about this], I suppose the whole challenge of trying to tease out causality, what comes first, the chicken and egg in relation to social media and people’s mental health. So that’s a clear limitation. Also, we’re relying on self-reports of time spent online and time using social media. We know that young people use many different media simultaneously.

Quite often, anecdotally, you’re going into a room and there’ll be a young people using that room and they might have the TV on, they might be sitting in front of a laptop, have an iPad or another mobile device all on the go at the same time, engaging in multiple platforms, multiple media. So trying to assess exactly how much time you spend online, isn’t necessarily that straightforward and those methods of capturing online time are well documented in the wider research literature and are well known. So that is a clear limitation.

Neil Fairbrother

You refer to FOMO, Fear Of Missing Out. Is FOMO real?

Yvonne Kelly

Well another anecdote might be useful. If you are perhaps unwell and you don’t feel like going on your phone and you’re off school and you really don’t feel that well and your eyes hurt, you’ve got a bad headache, maybe you’ve got a fever and you know, proper not well, and not able to engage on your phone for a couple of days. And then you pick the phone up and it’s got a thousand posts from Instagram, WhatsApp, Snapchat any number of different apps, I’m not picking on those particular platforms, there are tens and tens of these, they pop up and they emerge and then they close down almost on a weekly basis. But if you’ve got a thousand messages on that and there may be a feeling that you missed out on something.

Many of those messages might be totally extraneous, they might just be a series of smiley faces and emojis on a group chat and that might be 50 or 60 on a particular group chat. But you know, you might have missed out on some event or some conversation which could be important to you.

So much of young people’s time is spent online in beneficial ways with lots and lots of positive outcomes for young people, but I think that’s the fear of missing out. If people spend lots of time communicating online with each other and talking about things they’re going to do talking about plans and so on, I think there’s a real potential for young people to feel like they’re missing out if they’re not connected in that way.

Neil Fairbrother

What recommendations can you make or have you made based on the findings of your research?

Yvonne Kelly

That’s quite tricky. I think we go back almost to the common sense thing around sleep. If we think that there’s an association between using any kind of digital media and sleep, then clearly you want to minimize the impacts that the digital media use might have on sleep. So keeping devices out of the bedroom, maybe having a curfew of use. I know there are guidelines coming out of the States and Canada for example, which have recommendations about hours of use.

But for me, I think I would go back to that whole thing around content and context of use, so maybe hours of use or limiting the number of hours of use per day isn’t something that we would necessarily support because it’s about what’s going on online.

Keeping phones out of the bedroom or out of the sleeping space would be a very simple thing to do and maybe having, if it’s in a family setting, having agreed times for when phones all mobile devices, may or may not be used.

I often worry about double standards. If there’s a parent or a carer potentially criticizing their young person, why have you been on YouTube? I’m going to say, again, I’m not singling out YouTube, but why have you been on YouTube for the last three hours? What are you doing? You’re wasting your time! I think that can be not necessarily a kind of a dialogue that is beneficial for the young person, especially if at the same time that the parents or the carer is scrolling through their own social media feed on Twitter, looking at Facebook, looking at who’s got married or who’s just got divorced and so on. So I think in a family setting, there has to be an honesty and an open dialogue around the use of these kinds of platforms.

Neil Fairbrother

There is an interesting sentence that you have put into the abstract of your report in the interpretation section where you say “…findings are highly relevant for the development of guidelines for the safe use of social media and calls on industry to more tightly regulate hours of social media use”. Which industry you are referring to there?

Yvonne Kelly

Since we’ve read the report, many new applications are popping up in terms of nudging people, saying you’ve been online now for this amount of time, there are more checks and balances. I think that industry has a real opportunity here to develop algorithms which would be able to detect different types of use. So we say hours of use, but again, it goes back to this really important question around content and the context of use, if we can detect all sorts of nefarious activities online. So if for example, people are planning things related to terrorist activities, we know that those sorts of conversations and that kind of data that’s generated can be detected very easily online.

So if we can transplant that into those sorts of algorithms into other aspects of young people’s lives, maybe around online experiences; it’s the exchange of messages to each other, or the exchange of visual images or to do with mood, so if young people are expressing difficulties or they are searching particular websites, looking at various images which might be to do with self-harm. I think if these things can be detected by industry, then many more applications could be developed in terms of safeguarding young people.

Neil Fairbrother

Do some other industries that aren’t part of social media though, have a responsibility here? So for example, the fashion industry, the food industry, the cosmetics industry, they all use hyper-real images. We know the association between, for example, anorexia and the fashion industry. Do they have some responsibility here as well?

Yvonne Kelly

Yes, all of those industries and as a society. The obsession with idealized body image predates social media of course and as far as I can remember those debates going back into the 70s, the 80s, those debates have been very live. So I think as a society, there are multiple stakeholders and multiple parts of society that are relevant here.

Neil Fairbrother

Okay. Yvonne, I think we’re going to have to leave it there. We are out of time, so thank you so much for that, it’s absolutely fascinating, a really good report, a great body of work. And I loved the diagram in it and I hope that other people download it and read it. It’s well worth the time.

Yvonne Kelly

Thank you