New Study Finds Most Experts Share Concerns About the Effects of Smartphones and Social Media on Adolescent Mental Health

The largest-ever expert survey finds consensus about rising mental health problems, along with general agreement about some causes and potential policy responses

Are smartphones and social media platforms harming children’s mental health? Is the evidence of harm strong enough to justify taking protective action, such as keeping children and adolescents from opening social media accounts before they turn 16? The answers to these questions depend on who you are as a decision maker and what standard of proof you choose as the basis for your decision.

The highest standard of proof in the legal system is “beyond a reasonable doubt.” This is the standard used by juries in criminal trials in part because we agree as a society that a false positive (sending an innocent person to prison) is much worse than a false negative (letting a guilty person go free). Applied to questions about social media, this standard would require strong evidence of harm, and/or a high degree of consensus among experts that social media was harmful. The question asked here is: “can we be certain, or nearly certain?”

Should legislators require proof of harm beyond a reasonable doubt when they consider regulating social media platforms? No, because the job of a legislator is to make decisions and policies under conditions of uncertainty. It often takes the scientific community decades to reach consensus or near-certainty, especially on matters related to media effects. If legislators had to wait for scientific consensus before they could pass any laws to protect the public, they would be largely unable to adapt to changing environments, including novel public health threats and dangerous new products (e.g., food safety).

Legislators, like jurors in civil trials, often need to rely on a lower standard of proof: “the preponderance of the evidence.” This is the standard used in civil trials because the cost of being wrong, whether in favor of the plaintiff or the defendant, are seen as roughly equal. The jury must consider the evidence and then offer its best judgment about which way the facts of the case point. The question asked here is not about certainty, it is about probability: which side is probably right?

Applied to questions about social media, this standard of proof would simply require that overall, on balance, the evidence (as judged by experts) points more strongly toward harm than toward harmlessness. If the preponderance of the evidence indicates harm, then legislators should strongly consider doing something about it.¹

Should parents use the preponderance standard as they consider whether to let their children open accounts on Instagram, Snapchat, or TikTok at the age of 10? No, because the job of a parent includes protecting their children from harm while also fostering their development, so they don’t ask “will TikTok probably harm my child, or probably not?” The question parents are asking is “Is this product safe enough for my child to use?”

The standard of proof most appropriate for health-related parenting decisions is often “the precautionary principle,” which says that if there are reasons to suspect that some new product is harmful or dangerous, then precautions should be taken even if the experts are far from consensus, and even if the odds that the product is harmful are less than 50%. Better safe than sorry.

This is also the standard of proof used in many public health deliberations. As former Surgeon General Vivek Murthy wrote, when explaining why he was calling for warning labels to be placed on social media platforms:

“One of the most important lessons I learned in medical school was that in an emergency, you don’t have the luxury to wait for perfect information. You assess the available facts, you use your best judgment, and you act quickly.”

He wrote that in June of 2024. But a major new study was just posted online that offers additional information supporting his concerns and his call for action—even in the absence of “perfect information.”

The Consensus Study

A group of researchers set out to discover just how much consensus there actually is within the academic community on the potential harms of social media for adolescents, and on the state of the evidence about those harms. The study was led by a core team composed of Valerio Capraro, Laura Globig, Steve Rathje, and Jay Van Bavel. They recruited a large group of researchers (including the two of us). They submitted the manuscript and have posted it online in a public archive, in order to facilitate public discussion about the work. It has not yet been peer reviewed.

To structure the study, the researchers selected 26 claims from the text of The Anxious Generation, primarily from Chapters 1, 5, and 6. (We worked with them on this phase, to refine the claims they proposed, in order to be confident that the claims to be tested were the claims we made. We were not otherwise involved in the genesis of the study, nor in its design).

The study leaders assembled a large panel of experts through targeted outreach to prominent researchers representing a range of perspectives on the issue. They then used snowball sampling, asking initial invitees to recommend additional experts. This process yielded a group of 229 researchers who were formally invited to participate in the study. Additional researchers gained access to the survey via posts in relevant academic forums, bringing the total number of individuals who received the survey at some point to 288 (See footnote 2 for details on the expert sample).²

In Phase 1, the study leaders distributed Survey #1 to the recruited experts and also posted it on relevant academic forums. The survey asked participants to evaluate each of the 26 claims. A total of 137 experts evaluated at least one claim. For each claim, they were asked whether they believed it was “probably true,” “don’t know,” or “probably false.” They were also asked whether empirical evidence supported or contradicted the claim, and what types of evidence (if any) were available.

In Phase 2, the responses from Phase 1 were synthesized into 26 draft consensus statements—each three to five paragraphs long—which were then sent to the same set of researchers for evaluation in Survey #2. The researchers were asked to rate the accuracy of each statement. For example, a short summary of the statement for Claim #13 is given in the paper as “While there is evidence for a negative association between behavioral addiction and mental health, the direction of the causal relationship remains unclear.” Note that these statements generally focus on what experts believe about the state of the evidence, rather than on what experts believe about any particular harms. The experts were also asked what they would want changed in order to say that the statement was accurate. Based on this feedback, the study leaders revised the statements and conducted additional survey rounds. This iterative process continued until each statement either achieved unanimous “combined accuracy” (defined as being rated accurate or somewhat accurate by all respondents) or had gone through three rounds.

So what did the Consensus Study find?

Beyond a Reasonable Doubt

The goal of Phase 2 was to reach unanimous consensus among a diverse group of researchers, many of whom hold differing views on this topic. As a result, most of the consensus statements had to be carefully crafted to take account of everyone’s concerns. For most claims, this led to consensus statements that were more hedged or cautious than the original claim. For example, in Phase 1 (Survey #1), two thirds of the experts had agreed with Claim #14 which stated that “Heavy daily use of smartphones and social media can cause social deprivation.” Only 17% said that claim was “probably false.” Yet when it came to writing a consensus statement in Phase 2 that could attain near-unanimity, the statement became much weaker than the original claim: “The strength and even the direction of the potential causal relationship between social media use and social deprivation likely depends on various individual and social factors.” Of course, most interesting outcomes in the social sciences depend on individual and social factors, so nearly all of us were able to agree with that milder statement.

This is not a problem with the consensus project. If the goal is to attain a near-unanimity in order to reach “beyond a reasonable doubt,” and if the panel of experts contains even a few people on each side of a controversial issue, then most of the 26 statements are going to have multiple qualifications and limitations written in.

Despite these caveats, a few claims achieved overwhelming agreement, with the corresponding consensus statements concluding that the evidence was substantial. Here is how those areas of agreement are stated in the paper:

At the end of the consensus process, over 97% of experts agreed (i.e., rated it as accurate or somewhat accurate; 4 or 5 on 1-5 scale) that there is evidence that adolescent mental health has declined over the past two decades in several countries, including the USA (99.2%), several nations in the Anglosphere (99.0%), Nordic countries (97.9%), and Western European countries (97.1%), albeit with some heterogeneity across countries and measures. Moreover, 97.6% of experts agreed that there is evidence that heavy smartphone and social media use can cause some sleep problems. (p. 12)

Evaluating the sufficiency of evidence is inherently complex. Experts may disagree not only on what the evidence shows but also on how much evidence is required before making causal claims. Scholars may even disagree on the definition of basic terms like ‘addiction’ — so they may be interpreting the same question differently. Near-unanimity among researchers on any complex issue is rare, especially in the social sciences. (You can read all 26 consensus statements in the report, including the levels of agreement that each one elicited.)

Nonetheless, despite the difficulty of finding consensus, the fact that near-perfect consensus was achieved on several claims related to mental health trends has important implications for the debate. When we began collecting studies on mental health trends in 2019, we found frequent claims that “the kids are alright” and that any apparent trends were due to changing diagnostic or other extraneous factors, or that declines were limited to the U.S. Those who said that the trends were real and occurring in multiple nations were sometimes labeled as fostering a moral panic. Even in 2024, we still came across some such claims. These consensus statements can be taken by policymakers and the public as expressing a high level of confidence (among this set of researchers) that the declines are real and are international. The experts also reached consensus on the two claims for one causal pathway: that heavy social media use can cause sleep deprivation (claim #8), and that chronic sleep deprivation can cause declines in mental health (claim #9).

But what should legislators and parents take away from the failure to reach near-perfect consensus on most of the other claims? Should they conclude that the experts are so divided on questions of causality and harm that we just don’t know enough to take action?

No. Legislators and parents are not asking for near-certainty. Legislators want guidance on what to do, so they want to know about the preponderance of the evidence. Parents want to know whether there is enough evidence of risk that they should use an even lower standard, the precautionary principle. What do we find when we take “preponderance of the evidence” as our standard of proof, as assessed in Survey #1? We find that all 26 of the claims evaluated from The Anxious Generation were judged more likely to be true than false.

What Have Some Skeptics Said About Consensus?

In recent years, a few prominent researchers have repeatedly claimed that the experts have already reached a consensus that social media is not harming adolescent mental health, overall, and therefore policymakers should look elsewhere for ways to improve mental health.

Here is one such statement from Johannes, Masur, Vuorre & Przybylski (2024):

Most scholars have reached consensus: Concerns about general social media use don't seem warranted. [Emphasis added]

Johannes et al. believe that the evidence for their view is so strong and the consensus among researchers is so high that the case is essentially closed:

The disconnect between having delivered that evidence and requests for more of it puts social scientists in an unenviable position. After all, is the question about social media’s potential harm not answered? Wouldn’t it be prudent to stop investing resources into researching effects that evidence suggests are not there? [Emphasis added]

Odgers (2024) voiced a similar frustration, and similarly advocated that policymakers look elsewhere for solutions, not at regulating or delaying social media: “Hundreds of researchers, myself included, have searched for the kind of large effects suggested by Haidt. Our efforts have produced a mix of no, small and mixed associations.” She also stated that “we have a generation in crisis and in desperate need of the best of what science and evidence-based solutions can offer. Unfortunately, our time is being spent telling stories that are unsupported by research.” [Emphasis added]

What Do The Experts Believe Is True?

Are these claims true? Is there really a consensus among researchers that social media is not harmful and therefore policymakers should not waste their time considering regulations for which there is no research backing?

Because such claims have been and continue to be widely circulated in top journals and on social media by well-known researchers, it is important that they be evaluated, and Survey #1 from the Consensus Study offers a dataset that allows such evaluation. Here is how the findings from Survey #1 were described in the paper:

Experts generally believed the claims to be “probably true.” Responses to the “beliefs” question from Survey 1 indicate that for all 26 claims, more experts rated them as “probably true” than as “probably false.” For 23 of these claims, it was a majority who said they were “probably true.” None of the 26 claims was judged to be “probably false” by more than 21% of experts.

To explore these findings further, we downloaded the data from Survey 1 (you can find it all on this page) to create our own figures. What we offer below is our own analysis of the data from Survey 1. We do not claim to speak for the other 118 listed authors of the Consensus Study.

Figure 1 shows the percentage of experts who rated each claim as “probably true” (in blue) and “probably false” (in red). We calculated the ratio of True to False for each claim, listed those ratios along the right side of the figure, and sorted the rows by those ratios. The T/F ratio is a straightforward way to quantify which way the experts lean on each claim.

We note, however, that the ratio does not capture the “don’t know” responses, which is why we include that percentage alongside each ratio. High “don’t know” rates indicate greater uncertainty and should be interpreted alongside the T/F ratio.

For example, “don’t know” was the most common response for three claims: Nordic mental health (52% don’t know), the effects of attention fragmentation (50% don’t know), and the percent of college students who would prefer that social media not exist (45% don’t know). In these cases, even though more experts leaned toward “probably true,” the high levels of uncertainty suggests lower confidence in these claims.

Figure 1. Expert opinion about the truth or falsity of 26 claims, sorted by the ratio of experts who said “probably true” to “probably false,” and the percentage of those who reported “Don’t know” (You can find the data for all graphs in this spreadsheet. Note that ratios are rounded to the closest integer.)

As you can see, 21 of the 26 claims achieved a True/False ratio of five to one or more. The other five claims were also judged more likely to be true than false, only with smaller ratios. These ratios should not be taken to reflect the true ratio of belief among all researchers, because, as we’ll discuss below, we can only guess at the ways that this sample of researchers may differ from an impossible-to-obtain sample of all relevant researchers. But the ratios do help us see where these researchers lean strongly toward believing that the claim is true, and where there is more disagreement.

The Preponderance of the Evidence

Are Johannes et al. and Odgers correct that there is little or no evidence of harm? Are those who say social media is harming adolescents just “telling stories that are unsupported by research?” The Consensus Study addresses those questions. In both Survey 1 and the three waves of surveys given in Phase 2, experts were asked to provide citations of relevant research. In the end, 1,400 references were provided, and many of these studies suggest evidence of harm through social media use.

Survey #1 in Phase 1 also contains information relevant to evaluating claims about the absence of evidence. After expressing their judgments about the truth of each claim, the experts were asked to say how well they think the existing empirical evidence aligns with the claim. The exact wording of the alignment question was: “To the best of your knowledge, how does the empirical evidence align with this claim? (supports, mixed, don’t know, contradicts).”

Here is how the results from this question are presented in the report:

Experts generally reported being aware of more evidence in favour of a claim than against. Responses to the “alignment” question show that, for each claim, fewer than 5% of experts reported awareness of evidence contradicting the claim, and for each claim, experts reported being aware of more supporting evidence than opposing evidence. (p.9)

We graphed these responses in Figure 2 using the same approach as in Figure 1.

Figure 2. Expert opinion about the evidence for 26 claims, sorted by the ratio of experts who said that the available evidence “supports” the claim to experts who said it “contradicts” the claim, along with the percentage of those who said they “don’t know.”

All 26 claims achieved a “support/contradict” ratio greater than 5 to 1. The number of experts who said that they did not know of any evidence was often large too because for some of the claims, there is not a great deal of research available. In these cases the S/C ratio fails to convey the fact that many or most of the experts don’t know of specific studies. Nonetheless, to the extent that the experts knew of (and often nominated) specific studies, the preponderance of those studies supported each and every claim.

To be clear, we are not claiming that Survey #1 can be used to establish that there is a consensus. Establishing consensus was the job of Phase 2. But we think that the results from Survey #1 can be used to refute earlier claims that there is a consensus among researchers around the harmlessness of social media. We think the results from Survey #1 contradict the claim that the question of harm has already been “answered” and therefore that it would be “prudent to stop investing resources into researching effects that evidence suggests are not there.”

We now look more closely at the specific claims evaluated, by category, based on the chapter they were drawn from in The Anxious Generation. What are the claims that the majority of experts in this particular study thought were “probably true?”

Claims About Mental Health Trends

In Chapter 1 of The Anxious Generation we presented a series of graphs to document the sudden surge in rates of internalizing disorders (especially anxiety and depression), self-harm, and suicide that began in many countries in the early 2010s. Our claims from Chapter 1 were translated into statements presented to researchers in the following words:

1. Over the last two decades, there has been a decline in mental health among adolescents in the USA.

2. The decline in mental health among girls in the USA began in the early 2010s.

3. The decline in mental health among girls in the USA since the early 2010s is more pronounced than the decline among boys during the same period.

4. Over the last two decades, there has been a decline in mental health among adolescents in the Anglosphere (Australia, Canada, Ireland, UK, New Zealand).

5. Over the last two decades, there has been a decline in mental health among adolescents in the Nordic countries (e.g., Denmark, Finland, Iceland, Norway, Sweden).

6. Over the last two decades, there has been a decline in mental health among adolescents in Western Europe overall, although with variation across countries.

As you can see in Figure 3, all six of these claims were judged more likely to be true than to be false, although “don’t know” was the dominant response for the claim about the Nordic nations. This finding from Survey #1, combined with the later consensus statements achieved in Phase 2, contradicts the claims we described earlier from those who say that the decline of adolescent mental health is not real or not international.

Figure 3. Expert opinion about the evidence for 6 claims about mental health trends, sorted by the ratio of experts who said that the claim is “probably true” versus “probably false,” along with the percentage of those who said they “don’t know.”

Claims About the “Foundational Harms” Caused by Social Media

In Chapter 5 of The Anxious Generation we described four “foundational harms” that are caused by a phone-based childhood, with special attention to the role of social media (along with online gaming, short form videos, texting, and all the other interruptions and distractions of the phone-based life). The four are: social deprivation, sleep deprivation, attention fragmentation, and behavioral addiction.

These were translated into causal claims presented to researchers as four pairs of claims. Here is the exact wording:

8. Heavy daily use of smartphones and social media can cause sleep deprivation.

9. Chronic sleep deprivation can cause a decline in mental health.

10. Heavy daily use of smartphones and social media can cause attention fragmentation.

11. Attention fragmentation can cause a decline in mental health (possibly through mediating factors such as its negative impact on social relationships).

12. Heavy daily use of smartphones and social media can cause behavioral addiction.

13. Behavioral addiction can cause a decline in mental health.

14. Heavy daily use of smartphones and social media can cause social deprivation, such as isolation and lack of formative social experiences.

15. Chronic social deprivation can cause a decline in mental health.

As you can see in Figure 4, all eight of these claims were judged more likely to be true than to be false, although for the claim #11 about attention fragmentation the dominant response was “don’t know.”

Figure 4. Expert opinion about the evidence for 8 claims about the four “foundational harms” described in Chapter 5 of The Anxious Generation, sorted by the ratio of experts who said that the claim is “probably true” versus “probably false,” along with the percentage of those who said they “don’t know.”

Claims About Gender-Related Effects

In Chapter 6 of The Anxious Generation, we explained why we think the increases in internalizing disorders since the early 2010s were worse for girls. We included a section explaining the difference between correlational studies and experimental studies, and then presented four mechanisms by which social media use was causing particular harm to girls, more than boys.

1. Girls are more affected by visual social comparison and perfectionism

2. Girls’ aggression is more relational

3. Girls more easily share emotions and disorders

4. Girls are more subject to predation and harassment

Our claims in Chapter 6 were translated into causal claims presented to researchers as follows:

17. Social media increases visual social comparisons among adolescent girls.

18. Social media increases perfectionism among adolescent girls.

19. Social media increases relational aggression among adolescent girls, for example by providing tools for cyberbullying and exclusion.

20. Among adolescent girls, social media increases exposure to other people displaying or discussing their mental disorders.

21. Social media increases sexual predation and harassment of adolescent girls, for example by providing predators with access to potential victims.

As you can see in Figure 5, all of these claims were judged more likely to be true than to be false, although for two of the claims (about relational aggression and perfectionism) “don’t know” was a common response, though not the dominant response.

Figure 5. Expert opinion about the evidence for 5 claims about the gendered nature of harms from social media, as described in Chapter 6 of The Anxious Generation, sorted by the ratio of experts who said that the claim is “probably true” versus “probably false,” along with the percentage of those who said they “don’t know.”

Claims About Effective Policies

In chapters 9-12 of The Anxious Generation we proposed four norms that would each address a different collective action problem and that would, collectively, roll back the phone-based childhood. Here they are:

1. No smartphone before high school (or age 14)

2. No social media before 16

3. Phone free schools

4. Far more independence, free play, and responsibility in the real world

The first three of those norms were translated into claims for phase 1. (The leaders of the study did not consider the fourth norm relevant to the consensus project on the negative effects of smartphones and social media). Here is the exact text of the statements presented to researchers:

24. If most parents waited until their children were in high school to give them their first smartphones, it would benefit the mental health of adolescents overall. (Parents would give only basic phones or flip phones before high school).

25. Imposing (and enforcing) a legal minimum age of 16 for opening social media accounts would benefit the mental health of adolescents overall.

26. Phone-free schools would benefit the mental health of adolescents overall.

As shown in Figure 6, the experts thought that all three policies were more likely to provide benefits than not provide benefits.

Figure 6. Expert opinion about the evidence for 3 proposed reforms, drawn from Chapters 9 through 12 of The Anxious Generation.

In other words, this set of experts leaned toward the belief that each of these 3 reforms would probably improve adolescent mental health, although there was dissent as well. It is important to note that in Phase 2 of the study, the experts agreed that the evidence of benefits for these reforms is, in general, still preliminary. So, we cannot say that it is “beyond a reasonable doubt” that these reforms would be beneficial. Much would depend on the details of implementation in any case.

Caveats

As with any survey-based research, there are important limitations to keep in mind when interpreting the results.

First, it is possible that researchers who are more skeptical of claims about social media’s harms were less likely to participate in the study. While we know that a wide range of views were included, including some prominent critics, we cannot be certain that perspectives were represented in proportion to their prevalence in the broader research community.

On the other hand, we note that researchers were given the option of participating anonymously so that they could add their views to the study while not adding their names to the list of authors. In addition, all researchers who were invited to join were also invited to nominate other researchers, so the final group was not chosen just by the leaders of the project. (Please see Sections 3 and 4 of the Supplement for details on how the study leaders aimed to achieve a representative sample.)

Also, we know that some researchers who are skeptical about harm provided their ratings on Survey #1 and dropped out or removed their names later in the process. (Chris Ferguson said on X that he did this.) Conversely, Jon Haidt missed the deadline for responding to Survey #1, so his views are not included in any of the figures in this post.

It is also relevant that those who are more skeptical had disproportionate influence in Phase 2 of the study, because it only took objections from a few researchers to force a revision in a consensus statement in a more cautious or skeptical direction. In fact, we know of several well-known experts who dropped out of the study because they were concerned that the consensus process was leading to statements that were too weak and too hedged. Since several researchers on both sides removed their names from the final report, it is clear that neither side was fully happy with the report, which required compromise among dissenting experts.

Second, some critics may argue that the study was designed to validate The Anxious Generation rather than to neutrally assess the state of the science. We understand the rationale for that concern, which is why the full methods, datasets, and author list are publicly available for scrutiny. The study leaders were transparent in describing their collaboration with us to select claims and in outlining the iterative process used to refine the consensus statements. Importantly, we were not involved at all in the genesis of the project, the initial draft of the claims, the data analysis, or in drafting the initial manuscript. Neither of us had any prior communication with the study leader, Valerio Capraro, before receiving his invitation in August 2024, to refine a list of claims that he provided to us from the book. Of course, there are many other claims—and other ways of framing them—worth investigating, such as those related to the potential benefits of social media. We hope future studies will fill in those gaps.

Conclusion: What Should Parents and Policymakers Do?

So, is there enough evidence for parents and policymakers to act? For those waiting for proof “beyond a reasonable doubt,” the Consensus Study does not provide much guidance, although it does provide a roadmap for getting the kind of evidence that is most needed. But for those using “the preponderance of the evidence” as their standard, the Consensus Study adds yet more evidence that should be considered.

We note that on the three policy questions, the majority of experts said on Survey #1 that each one would probably help, yet the 3 consensus statements all said that the evidence was not strong enough to draw firm conclusions. For example, the statement for Claim #25, about the likely effects of raising the age for opening social media accounts to 16, put it this way: “Given the preliminary nature of these results, expert responses about the veracity of this claim should be interpreted as opinions rather than definitive, evidence-based conclusions.”

On the three policy questions, we assented to these consensus statements because there are hardly any experiments that truly test the policies using random assignment. To really test them, researchers would have to randomly assign whole schools or school districts to go phone-free (or not). Or they’d randomly assign whole communities to wait until 16 (or not) for their kids to open social media accounts. Such studies are very hard to do and are almost never done. In the absence of a solid body of high quality experiments on these policies, the experts agreed that we do not have “definitive” answers.

But in fact, enacting policies that experts think might work is among the best ways to obtain better information about what really works. Amy Orben, a leading researcher at the University of Cambridge who has at times expressed skepticism about the evidence of harm from social media, recently wrote a thoughtful essay in Nature on the problem that technology moves so much faster than science:

In a nutshell, we propose that if there is inadequate safety information available about a digital product — and processes don’t exist to produce that information at speed — researchers and policymakers should lower the evidence thresholds needed to start testing interventions to address potential harms. This should, in turn, accelerate both our understanding of harms and the provision of regulations. [Emphasis added]

We agree. Policymakers should work with scientists to test the three policy interventions that were evaluated in the Consensus Study: phone-free schools,³ delaying the age at which children get their first smartphone, and raising the age to 16 for opening social media accounts. These three policies are being implemented around the world, but if more could be done using random assignment (of whole schools or towns) and careful evaluation (using validated measures), we would get closer to that “perfect information” that the Surgeon General mentioned.

In conclusion, the Consensus Study is a major step forward in mapping out where there is a high degree of consensus, and where levels of agreement are lower. Parents and policymakers should understand the various levels of proof that are appropriate for different decision makers, and then they should make their decisions accordingly.

1. Of course the costs and risks of action matter too. For the four main reforms proposed in The Anxious Generation, such as delaying the first smartphone and giving more unsupervised play, the costs are minimal, as are the risks.

2. According to the authors, “the term ‘expert’ was not explicitly defined. Instead, we left it to the interpretation of the nominators to decide how to operationalize it. For example, some nominators suggested well-known researchers in one of the relevant fields, while others searched Google Scholar for authors of papers related to the topic. From the context of the nomination process, it seems clear that “experts” were understood to be individuals with expertise in social media and/or mental health.”

You can find a full explanation of the recruitment process in the paper’s methods section, but here are a few key details about the sample: Among those who completed the first and final survey, 85.9% of the sample self-reported holding a PhD, and 57% were parents. The experts surveyed came from 11 academic disciplines, with the largest share from psychology (30.1%), followed by communication and media studies (11.3%), health sciences (11.3%), business and management (11.3%), economics (7.5%), social sciences (6.8%), and psychiatry (3.8%). Participants represented institutions in 20 countries, with the majority based in the United States (54.1%), followed by Canada (9.0%), Australia (7.5%), and the United Kingdom (6.8%). Experts affiliated with EU countries made up 17.3% of the sample.

At the beginning of the survey, participants were asked to rate the overall effects of smartphones and social media on adolescents. The responses were diverse: 10.7% viewed the impact as strongly negative,At the beginning of the survey, participants were asked to rate the overall effects of smartphones and social media on adolescents. The responses were diverse: 10.7% viewed the impact as strongly negative, 15.3% as moderately negative, 2.7% as neutral, 3.3% as moderately positive, and 0.7% as strongly positive. Another 0.7% said they were unsure. The majority—66.7%—said the impact is context-dependent and influenced by a range of moderating factors.

3. If you know of any school or are working at a school that wants to be involved, contact Steve Rathje (Assistant Professor Carnegie Mellon University), a co-author on the Consensus Study.