Both nationally and internationally, the question of the effect of social media use on Generation Z (born circa 1997-2011) (Dougherty et al., 2020), and especially on girlsʼ mental health, is being debated. The debate is characterized by two main factions – one claims that the fear of screen time and social media is another example of moral panic, while the other, led by social psychologist Jonathan Haidt (Haidt, 2024), expresses serious concern about whether the “digitalization of childhood” may have had major, unintended health consequences. This position paper addresses this topic from a research methodology and precautionary perspective.

Much suggests that we need to rewind a bit to understand how we have arrived at where we are today regarding Norway. As early as 2004-2005, the contours of online communities and social mediaʼs influence on young people in their formative years were visible. At that time, I addressed whether “Our Socially Constructed Risk Culture” (Krumsvik, 2001) underestimates virtual risk and that young peopleʼs digital Bildung (Krumsvik, 2007) needs to be more prominently on the agenda. More recently, the problematic aspects of social media were addressed (Krumsvik, 2018), whether young people should be shielded from screens (Krumsvik, 2023a), why the knowledge base is divided (Krumsvik, 2023b), how AI can contribute methodologically (Krumsvik, 2024a; 2024b), and my new book provides a comprehensive analysis of the area (Krumsvik, 2023c).

Central to todayʼs debate is Jonathan Haidtʼs thesis that the launch of Instagram and the iPhone 4, the first smartphone with a front-facing/selfie camera in 2010/2011, put social media use into “overdrive,” and that this use may be linked to the increase in young peopleʼs mental illness both nationally and internationally around 2010-2012. However, there are divided opinions about the causal relationships, and Haidt faces opposition from other researchers (e.g. Odgers, 2024), partly because the knowledge base (both today and) from that time is insufficient. One reason for this is that this research topic was not highly prioritized nationally or internationally around 2010. Therefore, there is a lack of adequate population studies, experimental studies, lack of control groups in cohort studies, and a lack of Mixed Method studies and longitudinal cohort studies on social media use and mental (ill)health from around 2010.

A consistent weakness with population studies is that, although they are representative and important in many areas, they only study correlation (and not causation), and they do not fully capture social media usage trends (and occasional paradigm shifts) until several years after they occur. In addition, research is inherently retrospective and therefore (also for this reason) cannot fully keep up with the expansive development of social media. A central problem is that such population studies did not use thoroughly questions about young peopleʼs digital lifestyles, screen time, mobile phone use, and social media use both inside and outside of school before, during, and just after the period 2010-2012. In Norway, such questions were not included in some population studies until around 2015.

Looking at newer studies, the links between social media and mental (ill)health are present in some reviews (Keles et al. 2019), while they are more insignificant in others (Orben, 2020; Orben & Przybylski, 2019). Most reviews in this field clearly state that there is some uncertainty surrounding the findings, as the primary studies are of varying quality, have conceptual flaws, and are characterized by few longitudinal studies (Krumsvik, 2023).

This means that both factions have a somewhat shaky knowledge base to rely on. As such, there is no reason for certainty either way if one only looks at population studies from around 2010 onwards.

In other words, research has “not kept up with the times” in this area. It is therefore an important lesson that we must not fall into the same research trap now that a new technological paradigm shift is on the horizon with the advent of artificial intelligence, and below I will point out some factors that research so far on social media and screen timeʼs relationship to young peopleʼs mental (ill)health should have taken into account.

The expansive digitalization of formative years, education, and society at large necessitates a shift in research focus, high domain knowledge, domain-specific research expertise, and AI-generated, methodological innovation (e.g. Eichstaedt et al., 2018) to keep pace with technological development as a continuous “moving target.” This is important so that the research foci in population studies not only focuses on well-known public health problems, such as obesity, smoking, or alcohol use, but also incorporates well-founded questions about virtual health risks.

The lack of this has led to the current situation where we do not have solid baseline data on social media use from around 2010 in representative population studies, longitudinal studies, cohort studies, or experimental studies, making it challenging to identify clear causal relationships between social media use and changes in mental (ill)health over time.

Additionally, such population studies have sometimes only included sporadic questions about general internet use, which presents some conceptual challenges. These conceptual challenges make it difficult to fully understand how early exposure to social media affected, for example, girlsʼ mental (ill)health around 2010. Even in recent years, research on social media and mental (ill)health has sometimes failed to conceptually capture the subtle and complex ways (network and generation effects) social media can affect an individualʼs mental (ill)health. This makes it challenging at the individual level to know whether underlying factors are the cause, whether social media leads to mental illness, or whether pre-existing mental illness increases the use of social media.

Because of this, the research field is now faced with the research challenge of trying to recapitulate 2010 scenarios. This is done by conducting retrospective analyses of other relevant studies, as there is no access to direct social media usage data integrated into representative population studies from around 2010. Thus, the increase in mental illness from 2010 is seen in relation to the expansive increase in the spread of mobile telephony and social media from other studies, as well as changes in behavior among young people (e.g., increase in registered cases of self-harm at emergency clinics/hospitals) and other demographic variables. This is both methodologically sound, important, and interesting, but naturally fraught with much uncertainty about “what influences what.” The knowledge base could have been significantly improved if the technology giants had shared their vast amounts of user data with the research community, but they have not been willing to do so thus far. In addition, the research field must also rely on other retrospective studies where participants are asked to recall their use of social media and their mental (ill)health. Such recollections can be inaccurate and affected by forgetfulness or bias in memory.

The knowledge base – then as now – is also characterized by conceptually “zooming out” on digital platforms that are generic and have high complexity, at the expense of “zooming in” specifically on the algorithm-generated nature of social media, which has greater potential to affect young peopleʼs mental health due to its underlying psychological nature and addictive “tentacles” (often referred to as nomophobia, FOMO, multitasking, technoference, and nudging).

Another research challenge in getting the “ship back on course” is that even though it is possible to conduct cross-sectional and longitudinal studies today, it can be challenging to generalize the findings back to the period around 2010, as social, diagnostic, and technological variables and contexts have changed significantly in 15-16 years.

The knowledge base also shows that it is becoming increasingly important to distinguish between research that investigates whether it is good for young people to have a lot of screen time and spend a lot of time on social media since all their friends are there, and research that looks at whether high screen time and social media are harmful to young peopleʼs mental health. A good deal of research actually focuses on the former but is often interpreted as also answering the latter perspective. This is probably also an important seed of societal confusion about “what influences what,” and it is therefore important that especially researchers who investigate the former clearly explain the limitations this also imposes on healthcare workers, school leaders, teachers, parents, students, and the media.

Another challenge is that over a fifteen-year period, there may have been changes in how mental illness is diagnosed and reported. Increased awareness of mental disorders and reduced stigma may lead more young people to report their mental health problems. This can manifest as an increase in registered cases without necessarily reflecting an actual increase in the prevalence of the disorders. This also demonstrates a research challenge with population studies since they rely heavily on self-reported data, which is subjective and influenced by the individualʼs ability to conceptually understand what they are being asked about social media, as well as their willingness to report personal and stigmatizing information.

Another aspect of this is young peopleʼs self-developed conceptualization of everyday phenomena. This can sometimes result in a “psychologization” of everyday language and pathologization of common everyday challenges, which can also create problems in interpreting “what influences what” in such population studies. Much suggests that longitudinal research designs and research designs based on Mixed Method Research and methodological pluralism should be used to a greater extent to address such research phenomena in the future.

Where can we go from here in Norway? When there are no large, representative population studies with this thematic focus from this time, we must use the knowledge base we have. In Norway, two studies from 2009-2012 with over 20,000 students and 3,000 teachers from upper secondary education are particularly important and can provide some of the baseline data on screen time and social media use that population studies lack from this time. These two studies also take into account the uniquely Norwegian phenomenon of all upper secondary school students receiving their own PC in 2008 (Utdanningsforbundet, 2008), as the only country in the world, while nine out of ten students had mobile phones (the majority being smartphones). (Krumsvik et al. 2011; Krumsvik & Jones, 2015; Krumsvik et al., 2013; Krumsvik et al. 2020). This can be seen in light of some of the main findings showing high screen time among students and many digital distractions (Krumsvik, 2021) in classrooms at an earlier time than international studies in other countries can show. An addictive relationship to social media among some girls in the period 2009-2013 is also found, providing insight into several important aspects; the data indicate that social media already had a significant influence on young peopleʼs daily lives and their social interactions early in the history of social media. This suggests that any long-term effects of social media on mental health may also have begun to develop earlier than many may have thought. The fact that girls reported an addictive relationship to social media sixteen years ago points to gender differences in how young people engage with and are influenced by social media use. By identifying high screen time and addictive relationships to social media as prominent among youth, these data help to some extent isolate potential risk indicators for mental illness.

In conclusion, despite some shortcomings in population studies, it is important to consider that a high, collective exposure to technology in and out of school likely occurred earlier in Norway than in other countries. Could it also have had positive aspects? Absolutely, and there are also many positive aspects of the digitalization that has taken place in Norway in recent decades (e.g., for dyslexics), as highlighted in other contexts (e.g., Krumsvik, 2021). Nevertheless, an increasingly expansive digital lifestyle among young people from around 2009 may also have had its downsides for vulnerable groups.

The comprehensive HUNT study from 1995-2022 (Krokstad et al., 2022) shows that there has been a coincidence in time between the introduction of social media and smartphones, and an increase in mental illness among young people, especially girls, from around 2010-2022. The Public Health Report from 2024 (Folkehelseinstituttet, 2024) shows that for girls aged 13-24, the proportion reporting mental health problems such as worry and stress has increased from 2010-2022, and girls score significantly higher on most indicators related to mental illness. Regarding self-harm reported to the Norwegian Patient Register from 2008-2013, one of the findings is that there are “(…) particularly high rates in teenage girls of 15-19 years old” in this period (Qin and Mehlum, 2020, p. 9). However, the area is characterized by underreporting at emergency departments in the Norwegian specialist healthcare system, and the researchers therefore estimate that the figures could be three times higher than what they reveal (NSSF, 2020). But the study does not make any connection to social media use, nor does it cover the years from 2014-2022. A recent Norwegian study does, showing an increased risk of self-harm with high use of social media (Tørmoen et al., 2023). Another Norwegian study clearly shows a link between depression and anxiety and negative events on social media (Skogen et al., 2023). A Norwegian study among 10,460 Generation Z-students (aged 18-35 years) from higher education institutions found that more than half of the students consider themselves addicted to social media (mostly females), that there is a significant increase in the prevalence of self-reported mental disorders, with one in five students now reporting having a mental disorder (four of ten among females). 86 percent of students use electronic media (71 percent mention that this is social media) after bedtime, and one in three students meets the formal criteria for an insomnia diagnosis (Sivertsen et al., 2023). And a recently published Norwegian doctoral dissertation (Abrahamsson, 2023) has revealed a causal relationship between a three-year mobile phone ban in 119 Norwegian schools and a reduction in bullying among girls.

As you can see, even though we cannot be certain, there are some worrying coincidences both nationally and internationally that require societal vigilance. Both UNESCOʼs “Precautionary Principle” (UNESCO, 2005), the EUʼs “on the precautionary principle” (EU, 2000), and the Norwegian Public Health Act (Helse- og Omsorgsdepartementet, 2011) mention that “The precautionary principle applies when: there is great scientific uncertainty, scenarios or models based on scientific reasoning show that harmful effects are possible” (p. 50). Although we must respect both sides and be cautious about certainty in either direction, authorities, society, researchers, and the research community have a duty to be vigilant and transparent about the possibility that social media and high screen use may affect both learning in school and especially young girlsʼ mental health. On the one hand, there may be some health risks for vulnerable groups if society as a whole does nothing, and on the other hand, the 'precautionary principle' can be a low-cost measure with great health benefits. Based on this, the aforementioned policy documents, the Education Actʼs obligation to ensure a safe and good psychosocial school environment (Section 12-4) (Kunnskapsdepartementet, 2024), the Norwegian Directorate for Education and Trainingʼs “precautionary principle” (Utdanningsdirektoratet, 2024), and collective, societal altruism, are a good basis for protecting vulnerable groups until we understand these relationships more substantially based on a more solid knowledge base.