8
Research
As previous chapters in this report have discussed, social media poses both risks and benefits to young people, but quantifying these risks and benefits is difficult for many reasons. Some of the barriers are logistical. It is difficult to do research on the differential effects of a ubiquitous exposure; in a population made up entirely of smokers, for example, smoking could not possibly explain much of the likelihood of developing lung cancer (Pearce, 2011). Mental health outcomes are also complicated to study. The same clinical presentation may be produced by any number of brain dysfunctions, and the same mental problems can manifest in widely different ways (Maung, 2016; Paulus and Thompson, 2019). Given the challenges in measuring both exposures and outcomes, to say nothing of the variability in psychological responses to stimuli, it is hard to offer an overall summary about the relationship between social media and mental health beyond observing that the effects, both helpful and harmful, accrue differently to different users.
There are potential harms and benefits associated with social media that might come to light through careful analysis of the platforms’ algorithms over which, as Chapter 5 discussed, social media companies retain extremely tight control. This report has advocated for greater openness to algorithmic audit on the part of platforms, but some of the barriers to algorithmic research are more subtle. Facebook whistleblower Frances Haugen suggested that only 300 or 400 experts in the world design social media algorithms (Krass, 2022). In such a rarefied field, the small pool of qualified researchers poses an impediment to better insight into the
workings of these platforms. Scarcity of expertise in turn compels careful stewardship of the resources remaining for operational research.
This chapter discusses bottleneck problems holding back research on the harms and benefits of social media use. The first section presents a research agenda that elevates pressing unanswered questions in the field. Next, we set out a strategy to make such a research agenda possible through facilitating independent research on platform algorithms.
A RESEARCH AGENDA
As Chapters 3 and 4 made clear, despite many years of research assessing the relationship between social media and health and wellbeing, the evidence about specific factors linked to harms and benefits remains limited. Some of this challenge is inherent in the nature of the research itself. The standard of evidence needed to establish a causal relationship between an outcome and exposure is high. Most of the studies on this relationship have only established an association between social media use and different mental and physical health outcomes. As Chapter 4 explained, evidence of such association is useful and can drive hypothesis formation and further inquiry but is not sufficient, on balance, to lead this committee to recommend additional restrictions on young people’s access to social media.
Consider, for example, the Utah legislation putting age and time limits on children and teens’ social media use. The legislators’ intent to protect time for sleep and schoolwork and to prevent at least some compulsive use could just as easily have unintended consequences, perhaps isolating young people from their support systems when they need them. The committee recognizes that policy makers must make decisions under uncertainty, and some policies present interesting natural experiments that could inform our understanding of the field. Nevertheless, a stronger evidence base on certain key question could remove much uncertainty from the work of policy makers.
Recommendation 8-1: The National Institutes of Health, the National Science Foundation, and other research funders should support a research agenda that gives priority to the health consequences of social media use, the epidemiology of problematic use, the mechanisms through which social media use influences health, efforts to remediate harms associated with social media use, the role of parents and other adults in influencing positive use, and algorithmic audits. Across topics, the agencies should emphasize the need for validated tools to measure exposure to social media affordances, data sharing, and the establishment of long-term cohort
studies. Special emphasis should be given to study designs that attempt to understand causal directions.
As previous chapters in this report have made clear, social media as a category includes a wide range of applications and websites: the affordances that drive the social media experience of a video game differ from those of a video sharing site; the affordances central to the social networking sites differ from those involved in group chats. For research to be more useful to both industry and policy makers, better effort might be made to unpack how specific affordances influence health. This is, in theory, something that could be manipulated experimentally if the research team designed a platform offering or withholding the affordances being studied. The committee commends the Templeton Foundation’s recent effort to develop an experimental social media platform for algorithmic research and to recruit 5,000 participants to use this platform (Bail et al., 2022). A wider use of this strategy, especially if pursued over the long term, would give valuable insight into how social media algorithms could be harnessed for users’ benefit.
Industry collaboration might also be a venue for these types of questions, depending on industry’s willingness to collaborate with academic researchers. Users who express interest in joining a study could have certain affordances (auto scroll, for example) deactivated. Comparison with a participating control group could yield useful insights into the extent to which the affordance in question influences health.
Much of the committee’s assessment of the research agenda in this area underscores the need for more experimental and controlled study designs. Problems of reverse causation are particularly vexing when studying mental health outcomes; it is difficult to say if a problem such as depression is the cause or the effect of spending excessive amounts of time on social media. For this reason, the committee encourages rigorous study designs and statistical methods that allow for causal statistical inference on an interconnected network (Ogburn et al., 2022).
The committee commends research using randomized designs to study social networks, such as paying an experimental group of participants to stop using social media and following them over time (Allcott et al., 2019; Baym et al., 2020). Such research is difficult and expensive to conduct and would therefore benefit from being an explicit priority of the major government funders. There are also less onerous ways to answer key questions through capitalizing on natural experiments. Analysis of the staggered rollout of Facebook across college campuses and the simultaneous deterioration in student mental health provided compelling evidence of the platform’s potential to harm (Braghieri et al., 2022). As new waves of public policy attempt to limit media use in young people, it will
be critical to study the consequences of such restrictions. Adolescents in Utah, for example, should be followed for changes in measures of physical and mental health attributable to the restrictions that will go into effect in 2024 (Metz and Ortutay, 2023).
An Approach to Research Questions
Through their position as large, government funders of academic research, the National Institutes of Health (NIH) and the National Science Foundation (NSF) could encourage a more standardized approach to social media research. The development of validated measures of social media use would be an important first step.
Measuring Exposure and Outcomes
One barrier to better research is a problem with measuring a person’s media use. Researchers have developed apps, for example, that passively monitor screen time with relativly good accuracy (Wade et al., 2021). Perhaps for this reason, there is an overreliance on screen time as a measure of exposure. It is difficult to get beyond this relatively crude measure of online engagement, which is a barrier to a more nuanced understanding of the field.
In the absence of validated measurement tools, the field has developed an overreliance on self-reported frequency and duration of media use (Hancock et al., 2022). While this work has been useful in generating hypotheses, its use is limited as a way of guiding policy more precisely. It is difficult for users to provide reliable estimates of how much time they spend using various apps and websites, both because it is common to use several media at the same time and because all of these uses are relatively mundane, not occasions that would stand out in a person’s memory with sufficient clarity to be reported later to a researcher (Hancock et al., 2022). These measures do not capture the affordances used. More importantly, measures of time spent on a social networking site does not capture if the time was spent happily, enviously, or in a neutral state.
The same attention to measurement would also bring more clarity to the concept of well-being, an oft discussed but poorly operationalized outcome. There are many scales to measure well-being, although far fewer have been validated for adolescents (Rose et al., 2017). Research funders could help introduce more comparability into research through attention to the toolboxes researchers use to measure both social media exposures and well-being.
NIH has previously engaged in the development of standardized data collection tools for uniform measurement of complex variables. Over
several years in the 2010s for example, the agency worked with scientists around the country to develop the NIH Toolbox for Assessment of Neurological and Behavioral Function (Hodes et al., 2013). The value of this toolbox comes from both introducing standardized assessments and in capturing comparable information across studies and over time (Hodes et al., 2013). Social media research would benefit from a similar attention to standardization of measurement. Given that the field is still relatively new, prompt investment in a measurement tool kit for social media exposure could forestall continued disagreement about the validity of comparison among different studies.
Larger Samples, Longer Study Time
The committee recognizes that the appropriate measures of how users interact with various social media affordances will take time to develop and validate. The differential harms and benefits of social media may accrue differently in different demographic groups. Therefore, it will be particularly important to validate measures on large, nationally representative samples and to undertake additional validation among small but important populations. Transgender and gender nonconforming youth, for example, may be a population of particular concern owing to a combination of factors including vulnerability to bullying, heightened reliance on online support, and a seemingly increased risk of excessively using social media (Cingel et al., 2022).
Data repositories and data sharing are useful tools to make good use of scarce data on small populations. The committee applauds NIH’s new data sharing policy and similar efforts at NSF (NIH, 2020; NSF, 2018). Yet it is difficult for researchers studying social media to cooperate with these policies if they are using real-world, platform data, as the platforms generally disallow such sharing. It is possible that the same end goal of the data repository might be served through growing the field of algorithmic researchers and removing some of the legal barriers to their work, a topic discussed later in this chapter. Investigators may benefit from guidance as the most suitable repository for their data, one that would be well known to colleagues in the field and reliably funded more than 5 years out (BU Data Services, n.d.).
For understandable logistical reasons, cross-sectional studies have dominated the research on social media to date (Hancock et al., 2022). A longer time horizon on research will be particularly valuable in studying the effects of social media on young people. Ongoing longitudinal studies, such as the Adolescent Brain Cognitive Development (ABCD) Study, have already uncovered some associations between the length of time teens spend using screen media and poorer academic performance, sleep,
and behavioral problems (Paulich et al., 2021). An 8-year, nationally representative cohort study has provided some of the strongest evidence to date on both the risks and protective factors for problematic social media use (Coyne et al., 2020). Ongoing cohort studies by the same team have identified psychological and demographic characteristics associated with using social media in problematic ways, such as being bullied, being a member of a racial or ethnic group other than white, and having limited ability to self-regulate emotions (Coyne, 2023).
More long-term cohort studies are needed to understand the risks and benefits of social media and to determine how social media exposures influence development over the life course. Such studies are expensive and logistically complicated and may therefore require more explicit attention from funders (Caruana et al., 2015). There is also room to embed some attention to media use in existing ongoing national studies, such as the National Institute on Drug Abuse’s Monitoring the Future study and the Centers for Disease Control and Prevention’s Youth Risk Behavior Surveillance System. Across projects, there should be a greater emphasis on understanding both the risk and protective factors that are associated with various physical and mental health indicators.
Incorporating a Diversity of Experiences
This report has emphasized that social media affordances are tools and that these tools are used differently by widely diverse cross-sections of young people. A better understanding of these tools depends on more clarity as to both how they are used differently and the different influences they have on different users.
The sociological concept of intersectionality, referring to the way various social forces and identities manifest themselves in varying levels of power, is important to frame the research approach needed (Crenshaw, 2017). Consider, for example, a relatively accepted finding that social media use can help mitigate feelings of loneliness and isolation for LGBTQ+ teens and give them a space to explore their identities (Karim et al., 2022; Talbot et al., 2022). Given that we see a clear and different dynamic at play among this group than in the population at large, it is reasonable to consider that similar patterns may occur in other historically marginalized groups. The same attention is warranted to consider how young people of color or living in rural areas use social media in different and unique ways. One of the most important sources of heterogeneity among teens is developmental stage, for which age is a rough proxy. A more precise attention to developmental stage accounting for pubertal timing and other accepted indicators of development would bring more clarity to our understanding of various risks and benefits across the sec-
ond decade of life. Another crucial way to account for the diverse experience of adolescents is to include young people in a participatory research process. Such inclusion can provide a validity check for researchers and could also improve the relevance of proposed research questions. The perspective of people who are living the experience of social media in adolescence also adds nuance to an academic work and allows young people to contribute to research that will affect them and their futures.
An Emphasis on Mechanisms
Both researchers and research funders could advance our understanding of social media and well-being by designing proposals with an emphasis on the mechanisms through which social media use affects health. The displacement of time spent on sleep, exercise, studying, and interacting with people in person is, for example, one pathway through which gaming and social media can influence young people (Hall and Liu, 2022). Displacement may be one of the better-studied mechanisms in Table 8-1, possibly because some displaced behaviors are relatively straightforward to measure. The displacement of negative behaviors, as in using social media to control rumination, is less well studied. A body of research that investigates if and how the social and psychological mechanisms listed in Table 8-1 influence health would improve the field. The committee notes that Table 8-1 does not offer a comprehensive list; there are several other processes, such as self-efficacy and curiosity, that will likely, in the fullness of time, emerge as important mechanisms that explain the link between social media and health.
As Chapter 4 indicates, there is relatively little work in the field linking subjective measures of stressors to physiological endpoints, such as inflammatory markers. Assessment of the physical effects of social media use is mostly limited to either the most extreme (e.g., suicide) or the most obvious (e.g., sleep deprivation) outcomes. Attention to intermediary physiological outcomes could clarify if there is a relationship between social media use and other serious health problems, such as cardiovascular disease or other slowly-developing chronic conditions.
There are also open questions about how psychological and social mechanisms can co-occur and interact. A better understanding of the content to which teenagers are exposed on social media may be crucial to understanding this relationship. Problems related to social comparison or social sharing, for example, depend on some insight into the content that was shared or the pages viewed (Hancock et al., 2022). The experimental social media platforms described earlier in this chapter could be helpful in understanding how content influences the psychological experience of social media and how various mechanisms interact to influence physical
TABLE 8-1 Conceptual Mechanisms Linking Social Media Use and Well-Being
| Conceptual Mechanisms | Description |
|---|---|
| Social Structure | |
| Network size | Number of online friends, number of likes received, perceived network size, etc. |
| Social support and social capital | Perceptions of tangible and intangible assistance from one’s social network |
| Psychological Processes | |
| Social comparison Connectedness Fear of missing out | Evaluating oneself through comparison with others Perceptions of feeling socially connected with others Apprehension that others are having experiences without one |
| Overload Social compensation | Perceptions of too many social demands Use of social media to compensate for challenges encountered offline |
| Behavioral Dynamics | |
| Displacement and enhancement | Social media use displaces or reinforces meaningful interpersonal communication and social activities |
SOURCE: Hancock et al. (2022, p. 206). Reproduced with permission from American Psychological Association. No further reproduction or distribution is permitted.
and psychological outcomes. Such understanding might allow for a better harnessing of contagion in order to constrain excessive or unhealthy use.
Some of the most meaningful dynamics in a social network may be the hardest to measure. People who are connected through social networks have a relationship (of unspecified quality or mutuality) that connects them, meaning that statistical models that assume observations in a dataset to be independent are fundamentally flawed (Ogburn and VanderWeele, 2017). The use of tools, such as linear regression, to estimate causal relationships in a social network is therefore not valid and can result in inflated estimates of cause and effect (Ogburn and VanderWeele, 2017). Pressing questions of emotional and behavioral contagion through social networks depend on better understanding of how to analyze these data where participants are connected to each other and how these relationships influence the outcomes they experience.
Consider, for example, the important outcome of nonuse of social media, either through quitting a platform, taking a break, or uninstalling the application used to access it. It is possible that nonuse could be one of the more influential contagions in a social network but one that is, by definition, silent. Some evidence suggests that teenagers who do not use the major social networking platforms are perceived as less popular but
are better students (Schwartz et al., 2021). Qualitative research among people who quit social media suggests that despite initial, heightened anxiety about being left out, those feelings dissipate after a few months (Pennington, 2021). A better understanding of how nonuse works over time could inform policy actions such as prompts to users to pause their activity on a platform.
In a similar way, understanding the mechanisms through which social media advertising influences adolescents could help inform decisions as to how ads are targeted to them. Advertising is so central to the experience of social media, in the feeds, the banners, and the promoted content, that it transcends any one platform. As Chapter 2 explained, advertising can be subtle, including the endorsements of influencers or unboxing videos. A better understanding of how advertising interacts with adolescent development is necessary to make the experience of the platforms more positive for young people.
Patterns of Use
One obstacle to making broader policy recommendations regarding social media use and teenagers is our limited understanding of what drives their use and how adolescents use social media differently from young adults or the general population. A recent comprehensive meta-analysis on the influence of social media on well-being found that a modest majority of studies (54 percent) drew participants from a population made up entirely of undergraduates (Hancock et al., 2022). These young adults, as well as the older adults in the general population, are at a different life stage to younger teens and children and would therefore be expected to interact with social media differently. Understanding why children join social media and how they use the platforms will be an important precursor to any policy intervention aimed to change their use patterns.
Measuring how adolescents use social media includes measuring the extent to which they frequent different communities and interact with different affordances. Fanfiction communities, for example, may be supportive online groups, apps that encourage anonymous gossip would be expected to be far less so. Information about the relative prominence of such platforms (similar to what Pew surveys have estimated for the major platforms) would also facilitate a more systematic discussion of the variation in young people’s social media use.
At its heart, descriptive research should answer the questions, “how [young people] use social media, in which context, and for which purposes” (Hancock et al., 2022, p. 226). Such research will help explain how social media is used for good or neutral purposes in what evidence
indicates is the majority of young people (Coyne, 2023). It could also shed light into the epidemiology of problematic use. Despite widespread public concern about the addictive potential of social media, scientific research on the topic is more guarded (Doucleff, 2023; Panova and Carbonell, 2018; Shah, 2023; Waters, 2021). If social media or gaming addiction is similar to physiological addictions (e.g., alcoholism) or other behavioral addictions (e.g., gambling addiction), then the onset of symptoms may be sporadic and could be best studied through long-term prospective cohort designs (Hancock et al., 2022). A better understanding of patterns of overuse would be a necessary precursor to any efforts to include discussion of problematic use at the meetings to update diagnostic guides, such as the Diagnostic and Statistical Manual of Mental Disorders (APA, 2013). If the overuse of social media, even among a minority of users, is a serious addiction, then such diagnostic attention will be necessary.
Society would benefit from a more precise understanding of the etiological pathway through which ordinary social media use becomes uncontrolled social media use. Internet gaming disorder and problematic social media use are unambiguously harmful endpoints, yet it is not clear who is at greatest risk to develop them. Identifying teens at risk for these disorders could go a long way to mitigating the harms associated with social media.
Strategies to Mitigate Harm and Maximize Benefits
It is also possible, however, that for many people the framing of social media use or overuse as an addiction may create the perception among users that the technology controls them, driving a mindset that influences users to more negative experiences of mental state and relationships online (Lee and Hancock, 2019). In contrast, users who view the technology as a tool over which they have control have a more positive experience on platforms (Lee and Hancock, 2019). Given the potential power of mindsets to modify the experience of social media, explicit research attention to the role of user mindset would be helpful.
Therapeutic efforts to alter mindset might prove helpful to those young people suffering from an unpleasant or compulsive relationship with social media. There may also be room to use online platforms to reach young people who are suffering from any number of mental health difficulties. Ultimately a concern with this population of users is at the heart of most of the public discussion about teens and social media. For this reason, research funders would do well to consider treatment and remediation strategies to help adolescents who are harmed online. Some of these may be behavioral treatments or coping strategies for young people who have a problem with overuse or unhealthy use of social
media. There are also harms associated with social media use that come from victimization, with relatively little research on the best strategies to counter it (Foody et al., 2015). It would be valuable to compare different therapeutic strategies’ ability to alleviate the distress of being victimized and to cultivate resilience (Foody et al., 2015). The same types of comparison could give insight into the best strategies to remediate less common online harms, such as the rehabilitation of young people who were radicalized online. While some evidence suggests that the more heavy-handed actions relying on law enforcement are less successful, the best cognitive and emotional strategies for deradicalization are not clear (Brown et al., 2021).
The move to longer-term prospective studies described earlier in this chapter would shed light on the developmental trajectory for young people who are harmed online through victimization, radicalization, problematic use, or other means. Such research would also help identify potential cognitive and intellectual effects associated with use of social media starting in childhood. There is wide concern, for example, that social media and technology use poses a distraction to young people that could take a toll on their capacity for concentration and memory but only modest evidence that it is so (Firth et al., 2019; Siebers et al., 2022). A recent review of strategies to improve attention and concentration among children and adolescents found preliminary suggestions that mindfulness training and physical activity could improve capacity for sustained attention but cautioned that these results “need to be replicated with greater methodological rigor” (Slattery et al., 2022, p. 17).
The same need for rigor and attention applies to many of the important questions of how to maximize the benefits of social media. As Chapter 3 explained, publication bias may partly account for a relatively small body of evidence on the benefits of social media. Harmful outcomes such as depression and anxiety are investigated frequently, but beneficial ones are not. A greater research emphasis on positive outcomes, such as feelings of connectedness and support, could help reframe this discussion.
The Role of Parents
Concrete evidence as to how young people are helped or harmed online is important partly because in the absence of evidence, advice reverts to strategies based on accepted wisdom and common sense. The recent Surgeon General’s report Social Media and Youth Mental Health, for example, included advice that parents keep regular family meal times free of devices and to work with other parents to set consistent boundaries on social media (HHS, 2023). Most people would agree that these are good things. At the same time, the causal pathway through which these
behaviors mitigate the harms of adolescents’ social media use is not clear or direct. There are many overburdened and well-meaning parents for whom regular family mealtimes and relationships with other adults in their children’s friend group remain aspirational. With these parents in mind, the committee encourages more research exploring precisely how parents and other adult authority figures can best influence adolescents’ use of social media and their safety online.
There is evidence that poor relationships with parents are associated with internet gaming disorder, excessive social media use, and overuse of technology in general, but the direction of this relationship is not clear (Schneider et al., 2017; Trumello et al., 2021; White-Gosselin and Poulin, 2022). Good quality longitudinal studies have identified the distracting interruption of technology in parent–child interactions as an influence on adolescents’ anxiety, depression, civic engagement, and prosocial behavior (Stockdale et al., 2018). Both adolescents and their parents are concerned about the distractions of phone use, especially during family time, but there is little evidence as to how this problem is best overcome, and what studies there are come from a relatively homogenous population of middle-class, White Americans (McDaniel, 2019).
When young people are suffering, looking for a cause in their home environments is an understandable reaction. Yet many parents already blame themselves for their children’s mental health problems (Moses, 2010). Targeted research could inform a more productive discussion of how parents encourage healthy media habits. Researchers may do well to consider sensitive windows when parents are open to instruction and use these to promote warm and supportive parenting (Zurcher et al., 2018). A better understanding of how parents mediate young people’s healthy or unhealthy use of social media could help inform more specific, targeted advice that both parents and policy makers would welcome.
ENCOURAGING THE USE OF REAL-WORLD DATA
Attention to the research questions laid out in the previous section would give powerful insight into the influence social media has on adolescents. At the same time, there are only so many questions that can be answered without the platforms’ explicit cooperation. In Chapter 5, the committee recommended that researchers and platforms formally collaborate, taking steps to standardize a format for data at the application program interface (API), for example. The committee recognizes, however, that this recommendation might take considerable time to implement. In the meantime, there are researchers and journalists with pressing questions about the advertising, recommendation, and content moderation actions of social media companies. They pursue these questions to the benefit of society and at potential personal liability.
By way of illustration, some research designs aim to give insight into the platforms’ advertising and recommendation algorithms that lie at the heart of social media operations. A process called web scraping, which uses automated software to efficiently pull data from websites at large scale, sometimes provides the data for such research. For example, New York University researchers used a web scraping tool, a browser plug-in called Ad Observer, to collect information about the advertisements Facebook shows its users (with the users’ explicit authorization) (Edelson and McCoy, 2021b; NYU Cybersecurity for Democracy, 2023). Their research suggested inconsistencies in the way Facebook enforces rules on political advertising (Abdo et al., 2022; Edelson and McCoy, 2021a).
Web scraping is in a legal gray area, although the scraping of publicly available information is usually permissible. However, if data scraping is not welcome by the platform, it can be characterized as unauthorized access akin to hacking (Roberts, 2018). Under the Computer Fraud and Abuse Act (CFAA) subjects who exceed their authorized access to a protected computer risk civil and criminal liability.1
Yet what constitutes authorized access is often unclear. The Supreme Court decision in Van Buren v. United States provided some additional guidance on this situation.2 That decision clarified that publicly available data can be scraped from websites without violating the CFAA regardless of the platforms’ attitude toward their research provided the web scraping did not access an off-limits area of a computer.
The precise implications of the decision for web scraping that violates a platform’s terms of service remain unclear. On the one hand, the Supreme Court reserved the question whether exceeding authorized access applies only to areas of computers that are protected by “technological (or ‘code-based’) limitations on access” or whether it also applies “to limits contained in contracts or policies.”3 This reservation leaves open the possibility that web scraping could lead to CFAA liability (Abdo et al., 2022).
At the same time, the Supreme Court’s opinion in Van Buren contains multiple indications that it would not regard those who breached a policy, rather than a technical barrier, as having violated the CFAA. As an initial matter, the court’s ruling that a police officer’s accessing a law enforcement database in violation of department policy did not constitute exceeding authorized access under the statute clearly establishes that the mere breach of a document prohibiting certain uses of a computer is not necessarily sufficient to give rise to a CFAA violation. The court also noted
___________________
1 18 U.S. Code. § 1030.
2 Van Buren v. United States, 141 S. Ct. 1659 (2021).
3 Van Buren v. United States, 141 S. Ct. 1659 (2021).
that the CFAA limits civil liability to access to computers that inflict damage or loss, which the court equated to “technological harms—such as the corruption of files,” a standard that likely does not apply to web scraping.4 Furthermore, the Supreme Court disparaged constructions of what constitutes exceeding authorized access that would “criminalize[] every violation of a computer-use policy.”5 In so doing, it criticized the possibility of treating failure to follow specified terms of service as a CFAA violation. Doing so would “criminalize everything from embellishing an online-dating profile to using a pseudonym on Facebook.”6 In short, the decision warned that treating every violation of a computer-use policy as exceeding authorized access would have the nonsensical result of turning “millions of otherwise law-abiding citizens [into] criminals.”7
In addition, commentators have found it unlikely that “obscure provisions buried in unread terms of service” would constitute the type of “gates-up” protection necessary to give rise to CFAA liability (Goldman, 2021; Ohm, 2023). Indeed, when reaffirming that the CFAA did not permit LinkedIn to prevent a data analytics firm from scraping data off its website, the U.S. Court of Appeals for the Ninth Circuit interpreted Van Buren as holding the CFAA inapplicable to public websites that lack any limitations on access.8 The U.S. Department of Justice further reduced the likelihood of criminal liability by announcing that it would not prosecute “good-faith security research” under CFAA (DOJ, 2022).
The Van Buren decision thus greatly reduces but does not completely eliminate the risk that journalists and researchers who violate platforms’ terms of service in the course of digital investigations may be subject to liability under the CFAA, particularly if a platform employs technical barriers such as rate limits and challenge-response authentication tests that are meant to deter accessing a website too frequently or with the assistance of an automated script.9 Researchers might circumvent these barriers, with unclear implications for their liability under CFAA.
In short, many of the digital tools necessary for independent research into a platform’s policies and practices may present a degree of risk under the CFAA. This lack of clarity has a chilling effect on research, which in turn limits our understanding of the platforms that play a crucial role in society. Those conducting extensive scraping of social media sites to col-
___________________
4 Van Buren vs. United States, 593 U. S. ____ 15; 141 S. Ct. 1659–1660 (2021).
5 Van Buren vs. United States, 593 U. S. ____ 17; 141 S. Ct. (2021).
6 Van Buren vs. United States, 593 U. S. ____ 18; 141 S. Ct. (2021).
7 Van Buren vs. United States, 593 U. S. ____ 17; 141 S. Ct. (2021).
8 hiQ Labs, Inc. v. LinkedIn Corp., 31 F.4th 1180 (9th Circuit 2022)
9 Automated scripts are short pieces of code that can be triggered to work automatically.
lect data for research may also face potential liability under the Federal Trade Commission (FTC) Act, contract law, and tort law.
Recommendation 8-2: Social media companies should make a good faith effort to ensure access to data that would make research on the effects of social media on child and adolescent health possible, including the omission from their terms of service any prohibitions on researchers’ use of publicly available data.
A good faith effort to allow researchers to access social media data could be valuable to companies facing a great deal of public distrust and scrutiny. Companies can also help enable stronger research designs by partnering in the design of experimental studies. Much of the uncertainty regarding the harms and benefits of social media use stems from the opacity of the process of data gathering.
There is also evidence that the industry is moving away from an inclination to share data. In early 2023, for example, Twitter, once known for allowing relatively open use of its data, put restrictions on this policy including charging for access to the platforms’ data and its API at a relatively unclear rate (Jingnan, 2023). In an open letter, various civil society groups and researchers criticized this policy as a step backward for data access, transparency, and accountability (CITR, 2023a).
A few months after the Twitter decision, Reddit followed suit, restricting access to its API, Pushshift, and requiring payment for data (Isaac, 2023; Wiggers, 2023). In an open letter, the Coalition for Independent Technology Research pointed out that this abrupt withdrawal of this interface, cited in more than 1,700 scholarly papers, put the future of independent research at risk (CITR, 2023b).
Given the current efforts to restrict data access, it is reasonable to consider that social media companies may not voluntarily cooperate with Recommendation 8-2. And, people conducting extensive scraping of social media sites to collect data for research may also face potential liability under the FTC Act, contract law, and tort law. Indeed, in the current climate companies are becoming progressively more opposed to data sharing and cooperation with researchers. In addition, the committee recognizes that companies may prefer not to share with researchers data they would otherwise sell, as they do to advertisers. For this reason, it is prudent to consider that more compelling steps may be needed.
Researchers’ lack of access to social media data is a problem and one that has attracted congressional attention in recent years (Vogus, 2022). Some proposed legislation requires companies to share data with researchers; others are concerned with a strategy to protect user privacy in the data sharing process. Table 8-2 shows key similarities and differences
among prominent legislative proposals. The committee does not endorse or oppose any piece of legislation shown in Table 8-2. Rather, it commends the common concern with ensuring research access to social media data. The committee shares this goal, an overriding one for anyone concerned with understanding the effects of social media on adolescents.
Recommendation 8-3: Congress should pass legislation to ensure researchers can access data to examine the effects of social media on child and adolescent health.
In its spirit, making data accessible to researchers is a matter of openness to people with a legitimate public interest need. As Table 8-2 makes clear, the momentum to encourage research on social media platforms has not so far been insensitive to the platforms’ need to protect users’ privacy and their own proprietary information. Proposal vetting from the NSF, for example, is an unambiguously high bar on the quality of the proposed research and qualifications of the researchers. Such vetting could do much to assuage the platforms’ concerns about the credibility of the people accessing their platform.
The committee recognizes that increasing data access will pose thorny challenges. Additional steps can be taken to ensure researchers treat proprietary or confidential data with professionalism; the use of a code of conduct, for example, can help build trust between researchers and companies (OECD, 2007). In advancing the legislation recommended, it will be important to articulate technical steps that could improve the anonymization of user data, recognizing nonetheless that complete anonymization of social media data is not always possible or entirely effective (Keller, 2022). There is no easy or obvious answer to questions of how to balance society’s interest in greater transparency of social media operations against individual users’ expectations of privacy (Keller, 2022). It is the committee’s hope that by showing commitment to this question Congress can encourage an open discussion among researchers, industry executives, and privacy experts. The FTC guidance recommended in the previous chapter would also facilitate the process of data sharing by being clear about the companies’ competing obligations to cooperate with external researchers and to protect the customers’ privacy.
By providing protection against exorbitant penalties and certifications for confidentiality, legislative action could help tip the risk-to-benefit calculation for researchers. These steps are also necessary in service of an important end goal: advancing society’s understanding of how social media influences young people. In the long run, this research could also benefit the platforms, improving their public image and perceptions of openness around their policies.
TABLE 8-2 Recent Legislative Proposals on Social Media Platform Research
| Access to Whom? | What Types of Data? | Restrictions on the Research Project? | Which Applications and Websites Must Make Data Available? | Data Privacy Safeguards? | Safe Harbor for Independent Access? | |
|---|---|---|---|---|---|---|
| Platform Accountability and Transparency Act | Researcher affiliated with a university ad proposal vetted by NSF | Determined by NSF to be feasible, proportionate, and not an undue burden on platform | Project must be approved by NSF, have IRB approval or be IRB exempt, “aim to study activity on a platform,” and meet other criteria determined by NSF | More than 25 million unique users per month for the previous year, allows for user-generated content and sells advertising | FTC can set requirements for data sharing, such as data encryption, deidentified data delivery, etc. | Yes; no civil or criminal liability for the researchers |
| 117th Congress, 2nd Session (2021–2022) | ||||||
| Social Media Disclosure and Transparency of Advertising Act | Academic researchers and FTC | An ad library with information about advertisers who buy ≥$500 of ads in a year | None | More than 100 million users per month for the majority of the previous 12 months, sells advertising; FTC can update definition | None | No. |
| 117th Congress, 2nd Session (2021–2022) |
| Access to Whom? | What Types of Data? | Restrictions on the Research Project? | Which Applications and Websites Must Make Data Available? | Data Privacy Safeguards? | Safe Harbor for Independent Access? | |
|---|---|---|---|---|---|---|
| Digital Services Oversight and Safety Act | Researchers affiliated with institution of higher education or relevant nonprofit | To be determined by FTC regulations | Related to content moderation, product design, and algorithms and their effect on society, including health effects | More than 10 million active users in a month, platform stores information at request of users, FTC may vary specifications | Tiered access with more sensitive information having more safeguards, made available to fewer researchers | Yes; immunity from laws related to violating terms of service, if researcher is creating a research account or data donated by informed users; platform cannot discriminate against researcher |
| 117th Congress, 2nd Session (2021-2022) | ||||||
| Kids Online Safety Act | Researchers affiliated with institution of higher education or relevant nonprofit and approved by the Assistant Secretary of Commerce for Communications and Information | Data relating to the harms and safety of minors such as promotion of self-harm or eating disorders, patterns of use that suggest addiction, promotion of unlawful products or service | Public interest research on the safety and well-being of minors | A site used or reasonably expected to be used by minors | NTIA to establish standards including confidentiality standard | Yes; no action may be taken for violating terms of service |
| 117th Congress, 2nd Session (2021–2022) |
| American Data Privacy and Protection Act | Any research project | What is “reasonably necessary, proportionate, and limited” for the scientific project | Public interest research | No mandatory data sharing | All relevant research laws and human subjects protections | No |
| 117th Congress, 2nd Session (2021–2022) |
NOTE: NSF = National Science Foundation; IRB = institutional review board; FTC = Federal Trade Commission; NTIA = National Telecommunications and Information Administration.
SOURCE: Adapted with permission from Vogus, 2022.
REFERENCES
Abdo, A., R. Krishnan, S. Krent, E. W. Falcon, and A. K. Woods. 2022. A safe harbor for platform research. https://knightcolumbia.org/content/a-safe-harbor-for-platform-research (accessed September 19, 2023).
Allcott, H., L. Braghieri, S. Eichmeyer, and M. Gentzkow. 2019. The welfare effects of social media. https://www.nber.org/system/files/working_papers/w25514/w25514.pdf (accessed September 19, 2023).
APA (American Psychiatric Association). 2013. Diagnostic and Statistical Manual of Mental Disorders: DSM-5. 5th ed. Arlington, VA/Washington, DC: American Psychiatric Association.
Bail, C., D. S. Hillygus, and A. Volfovsky. 2022. Can social media promote intellectual humility? https://www.templeton.org/grant/can-social-media-promote-intellectual-humility (accessed September 19, 2023).
Baym, N. K., K. B. Wagman, and C. J. Persaud. 2020. Mindfully scrolling: Rethinking Facebook after time deactivated. Social Media + Society 6(2).
Braghieri, L., R. Levy, and A. Makarin. 2022. Social media and mental health. American Economic Review 112(11):3660-3693.
Brown, R. A., T. C. Helmus, R. Ramchand, A. I. Palimaru, S. Weilant, A. L. Rhoades, and L. Hiatt. 2021. What do former extremists and their families say about radicalization and deradicalization in America? Santa Monica, CA: RAND Corporation.
BU Data Services. n.d. Selecting a data repository. https://www.bu.edu/data/share/selecting-a-data-repository (accessed September 19, 2023).
Caruana, E. J., M. Roman, J. Hernández-Sánchez, and P. Solli. 2015. Longitudinal studies. Journal of Thoracic Disease 7(11):E537-E540.
Cingel, D. P., A. R. Lauricella, L. B. Taylor, H. R. Stevens, S. M. Coyne, and E. Wartella. 2022. U.S. adolescents’ attitudes toward school, social connection, media use, and mental health during the COVID-19 pandemic: Differences as a function of gender identity and school context. PLOS ONE 17(10):e0276737.
CITR (Coalition for Independent Technology Research). 2023a. Letter: Imposing fees to access the Twitter API threatens public-interest research. https://independenttechresearch.org/letter-twitter-api-access-threatens-public-interest-research (accessed July 12, 2023).
CITR. 2023b. Restricting Reddit data access threatens online safety & public-interest research. https://independenttechresearch.org/reddit-data-access-letter (accessed July 12, 2023).
Coyne, S. 2023. Who is most at risk and who is most likely to benefit from using social media long-term?: A differential susceptibility approach. Paper presented at Assessment of the Impact of Social Media on the Health and Wellbeing of Children and Adolescents Meeting 1, Washington, DC, January 9. https://www.nationalacademies.org/event/01-05-2023/assessment-of-the-impact-of-social-media-on-the-health-andwellbeing-of-adolescents-and-children-meeting-1 (accessed January 12, 2023).
Coyne, S. M., A. A. Rogers, J. D. Zurcher, L. Stockdale, and M. Booth. 2020. Does time spent using social media impact mental health?: An eight year longitudinal study. Computers in Human Behavior 104.
Crenshaw, K. 2017. On intersectionality: Essential writings. New York: The New Press.
DOJ (U.S. Department of Justice). 2022. Department of Justice announces new policy for charging cases under the Computer Fraud and Abuse Act. https://www.justice.gov/opa/pr/department-justice-announces-new-policy-charging-cases-under-computer-fraud-and-abuse-act (accessed July 12, 2023).
Doucleff, M. 2023. Teens say social media is stressing them out. Here’s how to help them. The Guardian. https://www.theguardian.com/global/2021/aug/22/how-digital-media-turned-us-all-into-dopamine-addicts-and-what-we-can-do-to-break-the-cycle (accessed September 19, 2023).
Edelson, L., and D. McCoy. 2021a. Facebook is obstructing our work on disinformation. Other researchers could be next. The Guardian. https://www.theguardian.com/technology/2021/aug/14/facebook-research-disinformation-politics (accessed September 19, 2023).
Edelson, L., and D. McCoy. 2021b. We research misinformation on Facebook. It just disabled our accounts. The New York Times. https://www.nytimes.com/2021/08/10/opinion/facebook-misinformation.html (accessed September 19, 2023).
Firth, J., J. Torous, B. Stubbs, J. A. Firth, G. Z. Steiner, L. Smith, M. Alvarez-Jimenez, J. Glee-son, D. Vancampfort, C. J. Armitage, and J. Sarris. 2019. The “online brain”: How the internet may be changing our cognition. World Psychiatry 18(2):119-129.
Foody, M., M. Samara, and P. Carlbring. 2015. A review of cyberbullying and suggestions for online psychological therapy. Internet Interventions 2(3):235-242.
Goldman, E. 2021. Do we even need the Computer Fraud & Abuse Act (CFAA)?–Van Buren v. US. https://blog.ericgoldman.org/archives/2021/06/do-we-even-need-the-computerfraud-abuse-act-cfaa-van-buren-v-us.htm (accessed July 12, 2023).
Hall, J. A., and D. Liu. 2022. Social media use, social displacement, and well-being. Current Opinion in Psychology 46:101339.
Hancock, J. T., S. X. Liu, M. Luo, and H. Mieczkowski. 2022. Social media and psychological well-being. In Technology: Social Science Research in the Age of Big Data, edited by S. C. Matz. American Psychological Association. Pp. 195-238.
HHS (U.S. Department of Health and Human Services). 2023. Social media and youth mental health: The U.S. Surgeon General’s advisory. https://www.hhs.gov/surgeongeneral/priorities/youth-mental-health/social-media/index.html (accessed January 2, 2024).
Hodes, R. J., T. R. Insel, S. C. Landis, and NIH Blueprint for Neuroscience Research. 2013. The NIH Toolbox: Setting a standard for biomedical research. Neurology 80(11 Supplement 3):S1.
Isaac, M. 2023. Reddit wants to get paid for helping to teach big A.I. systems. New York Times. https://www.nytimes.com/2023/04/18/technology/reddit-ai-openai-google.html (accessed January 2, 2024).
Jingnan, H. 2023. Twitter’s new data access rules will make social media research harder. NPR. https://www.npr.org/2023/02/09/1155543369/twitters-new-data-access-rules-will-make-social-media-research-harder (accessed January 2, 2024).
Karim, S., S. Choukas-Bradley, A. Radovic, S. R. Roberts, A. J. Maheux, and C. G. Escobar-Viera. 2022. Support over social media among socially isolated sexual and gender minority youth in rural U.S. during the COVID-19 pandemic: Opportunities for intervention research. International Journal of Environmental Research and Public Health 19(23):15611.
Keller, D. 2022. User privacy vs. platform transparency: The conflicts are real and we need to talk about them. https://cyberlaw.stanford.edu/blog/2022/04/user-privacy-vs-platform-transparency-conflicts-are-real-and-we-need-talk-about-them-0 (accessed July 12, 2023).
Krass, P. 2022. Transparency: The first step to fixing social media. https://ide.mit.edu/insights/transparency-the-first-step-to-fixing-social-media (accessed May 31, 2023).
Lee, A. Y., and J. T. Hancock. 2019. Social media mindsets: A new approach to understanding social media use & psychological well-being. Stanford University. https://osf.io/preprints/psyarxiv/f8wny (accessed January 2, 2024).
Maung, H. H. 2016. Diagnosis and causal explanation in psychiatry. Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences 60:15-24.
McDaniel, B. T. 2019. Parent distraction with phones, reasons for use, and impacts on parenting and child outcomes: A review of the emerging research. Human Behavior and Emerging Technologies 1(2):72-80.
Metz, S., and B. Ortutay. 2023. Utah social media law means kids need approval from parents. AP News. https://apnews.com/article/social-media-utah-kids-84bd1f6481071726327bce25cf3e7522 (accessed September 19, 2023).
Moses, T. 2010. Exploring parents’ self-blame in relation to adolescents’ mental disorders. Family Relations 59(2):103-120.
NIH (National Institutes of Health). 2020. Final NIH policy for data management and sharing. https://grants.nih.gov/grants/guide/notice-files/NOT-OD-21-013.html (accessed September 19, 2023).
NSF (National Science Foundation). 2018. Data management guidance for CISE proprosals and awards. https://www.nsf.gov/cise/cise_dmp.jsp (accessed September 19, 2023).
NYU Cybersecurity for Democracy. 2023. Ad observer. New York University. https://adobserver.org (accessed September 19, 2023).
OECD (Organisation for the Economic Co-operation and Development). 2007. OECD principles and guidelines for access to research data from public funding. https://www.oecd.org/sti/inno/38500813.pdf (accessed July 12, 2023).
Ogburn, E. L., O. Sofrygin, I. Díaz, and M. J. van der Laan. 2022. Causal inference for social network data. Journal of the American Statistical Association 1-15. https://doi.org/10.1080/01621459.2022.2131557.
Ogburn, E. L., and T. VanderWeele. 2017. Vaccines, contagion, and social networks. The Annals of Applied Statistics 11(2):919-948.
Ohm, P. 2023. The Computer Fraud and Abuse Act after Van Buren. https://www.acslaw.org/analysis/acs-journal/2020-2021-acs-supreme-court-review/the-computer-fraud-andabuse-act-after-van-buren (accessed July 12, 2023).
Panova, T., and X. Carbonell. 2018. Is smartphone addiction really an addiction? Journal of Behavioral Addictions 7(2):252-259.
Paulich, K. N., J. M. Ross, J. M. Lessem, and J. K. Hewitt. 2021. Screen time and early adolescent mental health, academic, and social outcomes in 9- and 10- year old children: Utilizing the Adolescent Brain Cognitive DevelopmentSM (ABCD) Study. PLOS ONE 16(9):e0256591.
Paulus, M. P., and W. K. Thompson. 2019. The challenges and opportunities of small effects: The new normal in academic psychiatry. JAMA 76(April):353-354.
Pearce, N. 2011. Epidemiology in a changing world: Variation, causation and ubiquitous risk factors. International Journal of Epidemiology 40(2):503-512.
Pennington, N. 2021. Quitting social media: A qualitative exploration of communication outcomes. Qualitative Research Reports in Communication 22(1):30-38.
Roberts, E. 2018. Is web scraping illegal? Depends on what the meaning of the word is. https://www.imperva.com/blog/is-web-scraping-illegal (accessed September 19, 2023).
Rose, T., S. Joe, A. Williams, R. Harris, G. Betz, and S. Stewart-Brown. 2017. Measuring mental wellbeing among adolescents: A systematic review of instruments. Journal of Child and Family Studies 26(9):2349-2362.
Schneider, L. A., D. L. King, and P. H. Delfabbro. 2017. Family factors in adolescent problematic internet gaming: A systematic review. Journal of Behavioral Addictions 6(3):321-333.
Schwartz, D., H. L. Fritz, A. R. Kelleghan, Y. Ryjova, A. Omary, L. M. Taylor, and T. Luo. 2021. Adolescents who are nonusers of fashionable social networking platforms. Cyberpsychology: Journal of Psychosocial Research on Cyberspace 15(3):Article 7.
Shah, S. 2023. Let’s free teens from social media addiction. https://www.tampabay.com/opinion/2023/06/03/lets-free-teens-social-media-addiction-column (accessed September 19, 2023).
Siebers, T., I. Beyens, J. L. Pouwels, and P. M. Valkenburg. 2022. Social media and distraction: An experience sampling study among adolescents. Media Psychology 25(3):343-366.
Slattery, E. J., E. O’Callaghan, P. Ryan, D. G. Fortune, and L. P. McAvinue. 2022. Popular interventions to enhance sustained attention in children and adolescents: A critical systematic review. Neuroscience & Biobehavioral Reviews 137:104633.
Stockdale, L. A., S. M. Coyne, and L. M. Padilla-Walker. 2018. Parent and child technoference and socioemotional behavioral outcomes: A nationally representative study of 10- to 20-year-old adolescents. Computers in Human Behavior 88:219-226.
Talbot, C. V., A. Talbot, D. J. Roe, and P. Briggs. 2022. The management of LGBTQ+ identities on social media: A student perspective. New Media & Society 24(8):1729-1750.
Trumello, C., L. Vismara, C. Sechi, P. Ricciardi, V. Marino, and A. Babore. 2021. Internet addiction: The role of parental care and mental health in adolescence. International Journal of Environmental Research and Public Health 18(24).
Vogus, C. 2022. Independent researcher access to social media data: Comparing legislative proposals. Center for Democracy and Technology. https://cdt.org/insights/independent-researcher-access-to-social-media-data-comparing-legislative-proposals (accessed September 19, 2023).
Wade, N. E., J. M. Ortigara, R. M. Sullivan, R. L. Tomko, F. J. Breslin, F. C. Baker, B. F. Fuemmeler, K. Delrahim Howlett, K. M. Lisdahl, A. T. Marshall, M. J. Mason, M. C. Neale, L. M. Squeglia, D. L. Wolff-Hughes, S. F. Tapert, and K. S. Bagot. 2021. Passive sensing of preteens’ smartphone use: An Adolescent Brain Cognitive Development (ABCD) cohort substudy. JMIR Mental Health 8(10):e29426.
Waters, J. 2021. Constant craving: How digital media turned us all into dopamine addicts. The Guardian. https://www.theguardian.com/global/2021/aug/22/how-digital-media-turned-us-all-into-dopamine-addicts-and-what-we-can-do-to-break-the-cycle (accessed September 19, 2023).
White-Gosselin, C.-É., and F. Poulin. 2022. Associations between young adults’ social media addiction, relationship quality with parents, and internalizing problems: A path analysis model. Canadian Journal of Behavioural Science / Revue canadienne des sciences du comportement. https://doi.org/10.1037/cbs0000326.
Wiggers, K. 2023. Reddit will begin charging for access to its API. https://techcrunch.com/2023/04/18/reddit-will-begin-charging-for-access-to-its-ap/ (accessed July 12, 2023).
Zurcher, J. D., H. G. Holmgren, S. M. Coyne, C. P. Barlett, and C. Yang. 2018. Parenting and cyberbullying across adolescence. Cyberpsychology, Behavior, and Social Networking 21(5):294-303.
