The First Phase: Research on Internet Use and Early Online Communication Contexts

The Internet became available to the public in 1991, and we see survey reports and journal articles on youths’ internet use starting around the late 1990s and early 2000s (Finkelhor et al., Reference Finkelhor, Mitchell and Wolak2000; Kraut et al., Reference Kraut, Patterson, Lundmark, Kiesler, Mukopadhyay and Scherlis1998; Roberts et al., Reference Roberts, Foehr, Rideout and Brodie1999; Sanders et al., Reference Sanders, Field, Diego and Kaplan2000; Stahl & Fritz, Reference Stahl and Fritz2002; Subrahmanyam et al., Reference Subrahmanyam, Greenfield, Kraut and Gross2001; Turow, Reference Turow1999). Given the novelty of the Internet and that youth were among the early adopters of it, studies generally focused on two questions – what youth did online (Roberts et al., Reference Roberts, Foehr, Rideout and Brodie1999; Turow, Reference Turow1999) and how their internet use related to their safety and psychological well-being (Kraut et al., Reference Kraut, Patterson, Lundmark, Kiesler, Mukopadhyay and Scherlis1998; Sanders et al., Reference Sanders, Field, Diego and Kaplan2000; Stahl & Fritz, Reference Stahl and Fritz2002; Subrahmanyam et al., Reference Subrahmanyam, Greenfield, Kraut and Gross2001). The majority of studies used self-report survey designs and showed that from the very beginning, youth who had access to the Internet used it for communication, with popular applications including email and chat rooms (Turow, Reference Turow1999).

Each new media technology such as radio, film, and television has been greeted by concerns about its negative effects on youth (Wartella & Jennings, Reference Wartella and Jennings2000; Wartella & Robb, Reference Wartella, Robb, Calvert and Wilson2009), and the Internet was no different. Early concerns centered on risky behaviors related to visiting problematic content (e.g., pornography), having contact with strangers, inappropriate/unsafe interactions (e.g., sexual solicitation, threatening or harassing contact) (e.g., Finkelhor et al., Reference Finkelhor, Mitchell and Wolak2000; Stahl & Fritz, Reference Stahl and Fritz2002), and psychological well-being (Kraut et al., Reference Kraut, Patterson, Lundmark, Kiesler, Mukopadhyay and Scherlis1998; Sanders et al., Reference Sanders, Field, Diego and Kaplan2000). The latter concern stemmed from two related elements of youths’ internet use at that time – first, computer-mediated interactions were text-based, and users were disembodied, so they did not have access to face-to-face cues such as gaze, gestures, emotional tone, and body language. Thus, online interactions were perceived to be lower in quality. Additionally, internet use was not very diffuse and so youths’ online interactions mostly occurred with strangers and others from outside their offline social networks. Given these elements of youths’ online interactions, the concern was that lower quality online interactions with strangers were displacing/replacing higher quality face-to-face interactions with friends and acquaintances; thus, early scholarship examined the implications of adolescents’ internet use for social isolation, loneliness, and depression. A detailed description of this research is beyond the scope of this chapter, and the interested reader is referred to the papers above and a monograph on digital youth coauthored by the first author (Subrahmanyam & Šmahel, Reference Subrahmanyam, Reich, Waechter and Espinoza2011).

This body of work suggested that the relation between internet use and psychological well-being was complex with contradictory results. For instance, the HomeNet study, a longitudinal field study conducted in Pittsburgh between 1995 and 1998 (Kraut et al., Reference Kraut, Scherlis, Mukhopadhyay, Manning and Kiesler1996, Reference Kraut, Patterson, Lundmark, Kiesler, Mukopadhyay and Scherlis1998) found that during the first two years of the study, increased time spent online was associated with declines in well-being (social involvement, loneliness, and depression). However, in the third year of the study, internet use was associated with smaller declines and even reversals (i.e., improvements) in well-being (Subrahmanyam et al., Reference Subrahmanyam, Greenfield, Kraut and Gross2001). The HomeNet study was unique in that a diverse sample of 93 families (208 adults and 110 children and adolescents) were given computers and internet access in 1995. They were then surveyed on several measures and their online activities were automatically recorded whenever they went online. The seminal study was conducted at a time when people had little exposure to technology, and thus the researchers were able to get a detailed picture of youths’ online activities and well-being from their first exposure to this technology and for a short period of time thereafter. Despite the study’s contradictory findings, the appeal of online settings for youth was clear – a 16-year-old HomeNet participant declared, “I really want to move to Antarctica – I’d want my cat and Internet access and I’d be happy.” The HomeNet study was conducted in 1995 – not much seems to have changed in that regard since that time!

Other studies in the first phase of research were more qualitative and focused on obtaining a rich picture of what adolescents and emerging adults (college students) were doing online (Greenfield & Subrahmanyam, Reference Greenfield and Subrahmanyam2003; Turkle, Reference Turkle1995a). Recall that youth were among the early adopters of the most common communication venues of that era, including internet relay chat rooms, multiuser dungeons (MUDs), and the commercially available chat rooms hosted by AOL, Yahoo, and instant messaging (also hosted by AOL). These spaces were significantly different from the social media apps were available when this chapter was written. They were text-based, accessed via computers and low-speed internet, and users were disembodied and largely anonymous (Subrahmanyam & Šmahel, Reference Subrahmanyam, Reich, Waechter and Espinoza2011).

Adult researchers were not using these venues, and thus were unfamiliar with the text-based language and code that digital media users were constructing and coconstructing within them; using qualitative methods from a variety of disciplines including ethnography, participant observation, and discourse analysis, these researchers provided a rich picture of the structure, content, modes of online communication, and youth subculture that was emerging within new online venues such as MUDS, chat rooms, blogs, bulletin boards, and webpages (Greenfield & Subrahmanyam, Reference Greenfield and Subrahmanyam2003; Huffaker & Calvert, Reference Huffaker and Calvert2005; Šmahel & Subrahmanyam, Reference Šmahel and Subrahmanyam2007; Subrahmanyam et al., Reference Subrahmanyam, Greenfield and Tynes2004; Suzuki & Calzo, Reference Suzuki and Calzo2004). These studies provided a window into adolescents’ emerging online lives and showed that youth used these spaces in the service of core developmental issues, including identity exploration, intimacy, health, and sexuality. Whereas youths’ online lives were psychologically connected to their offline counterparts, they were not mirror images of each other. Given the features of new online environments such as anonymity, disembodiedness, and lack of face-to-face cues, youths’ communication within them was often exaggerated and were observed with new intensities (Šmahel & Subrahmanyam, Reference Šmahel and Subrahmanyam2007; Subrahmanyam, Reference Subrahmanyam2007). For instance, within online chat rooms, there was one sexual comment per minute, and one obscene comment every two minutes (Subrahmanyam et al., Reference Subrahmanyam, Šmahel and Greenfield2006).

The Second and Third Phase: Research on Digital Communication Tools from Social Networking Sites to Social Media Apps

The next wave of online communication tools included the now defunct MySpace and Friendster, as well as Facebook. They were the first generation of social networking sites and were introduced in the early to mid-2000s. Social network sites were defined by boyd and Ellison (Reference boyd and Ellison2007) as “web-based services that allow individuals to (1) construct a public or semi-public profile within a bounded system, (2) articulate a list of other users with whom they share a connection, and (3) view and traverse their list of connections and those made by others within the system” (p. 211). For a detailed history of social networking sites, we refer the reader to boyd and Ellison’s Reference boyd and Ellison2007 article published in the special volume of the Journal of Computer-Mediated Communication, which was the first collection of research on social networking sites. Since then, there have been dramatic advances in hardware (e.g., smartphones and tablets), software, and internet access (e.g., high speed Wi-Fi), and the term “social media” has come to replace the term “social networking sites.” It is difficult to pinpoint who coined the term social media and when it began to be used in popular culture (Bercovici, Reference Bercovici2010). As noted earlier, this handbook uses the term social media to refer to digital tools that can be used for social interaction and selective self-presentation.

Compared to the early text-based online communication tools, social networking sites and social media apps are multimodal and allow users to interact and communicate via text, images, audio, and video. They make it possible to have both private and public interactions, and they vary in the extent to which users are disembodied and anonymous (Subrahmanyam & Greenfield, Reference Subrahmanyam and Greenfield2008). As access to the Internet and mobile devices became widespread, users were also more likely to interact with people they knew from their offline lives. In contrast to the earlier generation of online venues, where youth primarily interacted with people they met online, research revealed that youth used social networking sites and then social media platforms/applications to interact and make plans with friends from their offline lives and to keep in touch with peers they were not able to meet in person (Pempek et al., Reference Pempek, Yermolayeva and Calvert2009; Subrahmanyam et al., Reference Subrahmanyam and Greenfield2008). Social media platforms also allowed users to easily engage in self-expression and self-presentation via text and audiovisual content such as status updates, emojis, pictures and videos (Manago et al., Reference Manago, Graham, Greenfield and Salimkhan2008; Michikyan & Subrahmanyam, Reference Michikyan, Subrahmanyam and Yan2012).

As found in the first studies of online communication venues, youths’ offline and online social media lives were psychologically connected; online behaviors were again exaggerated, with youth reporting wider online networks (Manago et al., Reference Manago, Taylor and Greenfield2012). Given many adolescents’ 24/7 access to social media and, by extension, their immersion in elevated levels of peer interaction, self-disclosure, and self-presentation, research has examined the implications of social media use for the intrapersonal need for identity and the interpersonal need for intimacy as well as for psychological well-being. The chapters of this handbook will give the reader a detailed picture of the theoretical and empirical scholarship related to adolescents’ use of social media and their mental health. In the remaining part of this chapter, we draw on lessons learned from the extant research on youths’ use of social media to identify methodological challenges and conceptual issues pertaining to adolescent digital media use research.

Fluid Digital Media Landscape

The fluid nature of technology and rapid pace of change has always been an intrinsic element of the digital media landscape. During the first phase of digital media research, researchers not only had to contend with changes in hardware, but also in internet speed, software, and communication applications, and widening of users to include both strangers as well as friends and acquaintances. Additionally, there were constant shifts in the communication applications that were popular among adolescents at any given time, and changes in the features and elements within applications. MySpace, which was at one time a favored social networking site, was eventually supplanted by Facebook. There were also constant changes in elements of social media apps such as the top 8 list in MySpace or the Like button on Facebook. Rapid change in technology and rates of adoption also meant that there were changes in who was online. As noted earlier, initially youth mostly interacted online with strangers, including adults and peers who they did not know from their offline lives. Subsequently, as technology became more diffuse and widespread, more of their peers were online. Simultaneously there was an explosion in the popularity of more private and closed systems via social networking sites, within which people created profiles and chose who they interacted with and who could see the information they shared on their profiles. They were thus more likely to interact online with peers from their offline lives.

Perhaps the most challenging issue was how frequently changes occurred – there was often a lag between when a digital media platform emerged and gained popularity and when researchers began to investigate its use in earnest. In some cases, researchers found themselves investigating a platform that was no longer in vogue, as youth had moved on to the next new context that had appeared on the digital scene. This fluidity of digital platforms is particularly challenging for longitudinal studies as it complicates comparisons between different waves of data. Logistically, this meant that researchers had to focus broadly on a category of applications (e.g., chat rooms or social networking sites) and not target specific applications (e.g., AOL chat rooms, MySpace, or Facebook). This is also the approach adopted in this handbook. Given the fundamentally transient nature of digital platforms, even focusing on application categories does not ensure continued relevance after a platform’s eventual demise; so, we used a developmental lens for our early studies and focused on developmental tasks including identity, sexuality, and intimacy. Such a developmental approach ensures that study results are relevant long after the shelf life of a particular digital media platform or category of platforms/applications. An additional approach to ensure the continued relevance of research on an application is to focus on elements or features of digital platforms and the activities that they support. This issue also relates to conceptual considerations and is discussed in further detail in the latter part of this chapter.

A related methodological challenge that arose in early digital media studies was that each new platform had different communication features or capabilities. From the earliest studies of online communication, communication scholars interested in computer-mediated communication investigated how communicative cues in online settings shape interaction within them (Culnan & Markus, Reference Culnan, Markus, Jablin, Putnam, Roberts and Porter1987; Walther, Reference Walther1992). Subsequently, drawing from this body of work and Gibson’s notion of affordances in the context of object perception (Gibson, Reference Gibson1979), the term media affordances (Hutchby, Reference Hutchby2001) has been used to refer to the qualities of different digital platforms, including mobile phones and social media (boyd, Reference boyd and Papacharissi2011; Ellison & Vitak, Reference Ellison, Vitak and Sundar2015; Reid & Reid, Reference Reid and Reid2007; Subrahmanyam & Šmahel, Reference Subrahmanyam, Reich, Waechter and Espinoza2011; Treem & Leonardi, Reference Treem and Leonardi2013). Because different platforms have different affordances, it is important for researchers to be flexible and use different approaches when studying youths’ use of these technologies.

In our own work at the Children’s Digital Media Center @ Los Angeles, techniques from discourse analysis and participant observation were adapted to investigate how adolescent digital media users utilize the communication cues available in online chat rooms to construct and co-construct conversational coherence (Greenfield & Subrahmanyam, Reference Greenfield and Subrahmanyam2003; Subrahmanyam & Manago, Reference Subrahmanyam and Manago2012) in the service of key developmental tasks such as identity and sexuality (Subrahmanyam et al., Reference Subrahmanyam, Greenfield and Tynes2004). These studies were qualitative in design and used a single chat transcript to analyze the online culture that adolescent digital media users were co-constructing. Subsequent studies utilized a combination of qualitative and quantitative analysis to examine a larger number of utterances in chat rooms as well posts on online blogs (Subrahmanyam et al., Reference Subrahmanyam, Šmahel and Greenfield2006, Reference Subrahmanyam, Garcia, Harsono, Li and Lipana2009). When social networking sites and text messaging became popular and youths’ digital communication occurred in private spaces, we shifted to self-report measures and adapted techniques from social network analysis (Reich et al., Reference Reich, Subrahmanyam and Espinoza2012; Subrahmanyam et al., Reference Subrahmanyam and Greenfield2008), mixed-method (Michikyan, Reference Michikyan2019; Michikyan et al., Reference Michikyan, Subrahmanyam and Dennis2015), and daily diary (Subrahmanyam et al., Reference Subrahmanyam, Frison and Michikyan2020) designs to better investigate the developmental implications of youths’ digital media use for their well-being. Thus, it is important for researchers investigating youths’ social media to be flexible and adapt their methodological and analytical approaches based on an analysis of the digital media platform’s affordances, how youth use it, and its potential mental health implications.

Measuring Digital Media Use

Researchers investigating youths’ digital media use must make decisions about how they measure usage and the research designs they adopt. These decisions have methodological as well as conceptual implications, and in this section, we address them from a methodological perspective. Using the “historical lens” that we have adopted heretofore in this chapter, we see that from the earliest studies of youths’ internet use and continuing into extant social media apps, amount of time spent online has been researchers’ favored measure of operationalizing social media usage. This was influenced by prior research on mass media such as television, and research on the first generation of electronic media, including computers and games, when computers were often in common spaces and shared among members of the family. Thus, it was reasonable that digital media users would be able to estimate the time they spent on average during a given period (day or week). With youths’ widespread access to mobile technologies and high-speed internet, the issue of time use has become considerably complicated. As with all retrospective self-report measures, internet time use measures are susceptible to inaccurate/distorted/biased estimates (Parry et al., Reference Parry, Davidson, Sewall, Fisher, Mieczkowski and Quintana2021; Scharkow, Reference Scharkow2016). An alternative way of obtaining an estimate is to use software to automatically record internet use, as in the HomeNet study; however, given that multitasking with multiple windows on a screen or with multiple devices is ubiquitous, it is important to distinguish between open/active windows and applications to which a user may or may not actually be paying attention.

Another approach to studying digital media use is by analyzing the actual content of digital communication. In fact, this was the method by which researchers analyzed conversation in the first generation of digital media platforms such as online teen chat rooms, blogs, and bulletin boards (Subrahmanyam et al., Reference Subrahmanyam, Šmahel and Greenfield2006, Reference Subrahmanyam, Garcia, Harsono, Li and Lipana2009; Suzuki & Calzo, Reference Suzuki and Calzo2004). These applications were publicly available and so accessing the content was relatively easy for researchers. As the digital landscape moved toward closed networks with private (e.g., private messaging on Facebook, direct messaging on Twitter, private messaging on smartphones) and public communication (e.g., Facebook wall, publicly available tweets), researchers deployed automatic means to capture the content of youths’ digital communication (Negriff, Reference Negriff2019; Underwood et al., Reference Underwood, Rosen, More, Ehrenreich and Gentsch2012). Underwood et al. (Reference Underwood, Rosen, More, Ehrenreich and Gentsch2012) pioneered this technique by providing adolescent participants in a longitudinal study with BlackBerry devices and automatically recording text messages and other contents of their private communication. While this approach provides an unfiltered window into adolescents’ digital worlds, it is logistically challenging, as it provides a vast amount of data that then has to be analyzed by researchers, machine learning models, or a combination of the two (Dinakar et al., Reference Dinakar, Weinstein, Lieberman and Selman2014). The biggest concern of this approach, of course, is that the analytical technique – whether human or machine – may impute intentions, emotions, biases, and motives that were not intended by the social media user.

While there is no easy remedy for the measurement challenges outlined above, some possible solutions are briefly described next. This is not intended to be an exhaustive list, but to provide a selective sampling to illustrate how to approach measurement of youths’ social media use. First, researchers should consider a mixed-methods design to capture users’ intentions; in one study on online self-presentation, we asked participants to describe a picture they posted, and their identity-related meaning making was coded and then quantitatively analyzed. Note that codes were based on participants’ own descriptions of the picture and their reasons for posting instead of the researchers trying to deconstruct the image and post (Michikyan et al., Reference Michikyan, Subrahmanyam and Dennis2015). Second, we encourage researchers to consider daily diary designs and ecological momentary assessment techniques to get more accurate estimates of users’ social media use and activities over several days at a time. Daily diary studies have been used extensively in social psychology to study frequent everyday interactions (Bolger et al., Reference Bolger, Davis and Rafaeli2003); at the end of each day, participants are asked to report on their interactions that day and about other variables such as well-being, conflict, etc. In the experience sampling method, participants are asked to self-report what they are doing, feeling, and thinking at random points during times they are awake (Larson & Csikszentmihalyi, Reference Larson and Csikszentmihalyi2014). In both methods, participants report about their social media use when it is fresh in their mind, helping to limit memory distortions that are more likely when asked to estimate or recall use and activities on average. Both designs have the added advantage of yielding multiple data points over time, which are essential to address key questions regarding the longer-term implications of social media use and for the examination of within-person effects (Gonzales, Reference Gonzales2014; Jelenchick et al., Reference Jelenchick, Eickhoff and Moreno2013; Kross et al., Reference Kross, Verduyn and Demiralp2013; Pouwels et al., Reference Pouwels, Valkenburg, Beyens, van Driel and Keijsers2021; Subrahmanyam et al., Reference Subrahmanyam, Frison and Michikyan2020).

Finding Equivalent Comparison Groups

A final methodological challenge stems from the widespread use of social media among adolescents. Typically, when studying the influence or impact of a variable, psychologists compare groups of people with varying levels of the variable in question. This is true whether the comparison group is naturally occurring in a correlational or descriptive study (e.g., coffee drinkers and nondrinkers, alcohol drinkers and nondrinkers, and video-game players and nonvideo-game players) or created by the experimenter’s manipulation in an experimental design. Because digital media have become ubiquitous in adolescents’ lives, it is virtually impossible to find a group of youth who do not use social media and are truly equivalent to a group of youth who use them, at least in the Global North. The lack of a naturally occurring control or comparison group is an intractable design challenge facing social media researchers, and correlational designs have dominated the literature to date. A few researchers have conducted clever experiments to test the effects of digital communication (Gross, Reference Gross2009; Sherman et al., Reference Sherman, Michikyan and Greenfield2013, Reference Sherman, Payton, Hernandez, Greenfield and Dapretto2016; Vogel et al., Reference Verduyn, Lee and Park2015; Weinstein, Reference Weinstein2017); given the dearth of such studies, there is an urgent need for more experimental designs to help unearth the mechanisms by which social media use shapes well-being.

Consider Digital Media as a Developmental Context

There is wide agreement in the literature that human behavior across developmental time can only be fully understood in context (Bronfenbrenner & Morris, Reference Bronfenbrenner, Morris, Damon, Lerner and Lerner2006; Vygotsky, Reference Vygotsky1978). With regard to adolescent development, the role of contexts such as families, peer groups, schools, and neighborhoods has been well documented (Petersen, Reference Petersen1993; Steinberg & Morris, Reference Steinberg and Morris2001), and as noted at the start of this chapter, digital media should be viewed as an important context in the lives of adolescents. Digital media broadly, and social media more specifically, are incredibly diverse, and a variety of applications are available, each with its own unique communication contexts and affordances. An important question is the extent to which researchers should focus on particular social media platforms in any study. As an example, consider the case of Twitter and Snapchat, which share similar features such as photo sharing, but also differ in their affordances and how they are used (Alhabash & Ma, Reference Alhabash and Ma2017). From an affordances perspective (Treem & Leonardi, Reference Treem and Leonardi2013), both social media platforms – Twitter and Snapchat – allow for visibility (i.e., the ability to make information about oneself, once fully or partly invisible, visible to others), editability (i.e., the ability to construct, reconstruct, and coconstruct the information intended to convey to others), and association (i.e., the ability to establish a relationship with others and with a specific content), but they differ in terms of persistence (i.e., the ability to access and review the information in its original form after the user has completed the communication or interaction). Empirical evidence also suggests that the specific digital media platform matters to the user in terms of how and why it is used (Alhabash & Ma, Reference Alhabash and Ma2017; Madden et al., Reference Madden, Lenhart and Cortesi2013; Utz et al., Reference Utz, Muscanell and Khalid2015); if it matters to the user, then it seems that it should matter to researchers’ conceptualizations of digital media both when investigating and disseminating their results.

At the same time, social media platforms are adopting similar features –Snapchat Stories vs. Instagram Stories vs. Facebook Stories. What was once unique to Snapchat – the limited time feature – is no longer the case, as Instagram and Facebook now afford the ability to automatically vanish videos and images. The most recent trends – similarities across social media platforms and the emergence of newer platforms like TikTok – push us to consider whether and how we need to distinguish between the various social media platforms that youth use when investigating implications for development and well-being. Given how quickly digital media platforms evolve and the variety of affordances and activities possible on each platform, it is challenging for researchers to remain consistent in their conceptualization as the platforms themselves change. Perhaps one way to reconcile this conundrum – a lack of consistency in conceptualizing and operationalizing specific digital media platforms – is to be consistent in the recognition and articulation of digital media platforms as unique developmental contexts with specific affordances through which users influence and are influenced by the context (Subrahmanyam & Greenfield, Reference Subrahmanyam and Greenfield2008; Subrahmanyam et al., Reference Subrahmanyam, Šmahel and Greenfield2006).

Researchers should also be mindful of both the contextual factors outside of the digital world (e.g., race, socioeconomic status, immigration generation status) as well as within the specific social media context (i.e., implicit and explicit norms and expectations shared among users within a specific digital media context) (De Choudhury et al., Reference De Choudhury, Sharma, Logar, Eekhout and Nielsen2017; Elsaesser et al., Reference Elsaesser, Patton, Weinstein, Santiago, Clarke and Eschmann2021; Michikyan & Suárez-Orozco, Reference Michikyan and Suárez-Orozco2017; Nesi & Prinstein, Reference Nesi and Prinstein2019; Valkenburg & Peter, Reference Valkenburg and Peter2007). For brevity, we only discuss contextual factors within social media, and provide two examples to illustrate the potential role of social media–specific contextual factors that could shape psychological well-being and mental health. One example is the positivity norm found within many social media platforms, where the unspoken expectation is for users to engage in a set of behaviors in an attempt to put their best “face” forward (Qui et al., Reference Qiu, Lin, Leung and Tov2012). Another example is digital status seeking – wherein users engage in specific behaviors to obtain peer status (Nesi & Prinstein, Reference Nesi and Prinstein2019). These examples illustrate that just as researchers take into account key features of particular developmental contexts such as peer groups and schools, they should take into account the contextual characteristics of the specific digital media platform studied – including its affordances and unique cultural elements (e.g., positivity norm), even when such contextual characteristics may not be analytical variables. Taken together, these examples suggest that consistency in the recognition and articulation of digital media can provide a more complete understanding of digital media usage in the service of development and mental health.

Reimagine Digital Media Usage

Another critical issue concerns the conceptualization and operationalization of digital media usage. Digital media usage has been conceptualized and operationalized in myriad ways (see Kross et al., Reference Kross, Verduyn, Sheppes, Costello, Jonides and Ybarra2020; Schønning et al., Reference Schønning, Hjetland, Aarø and Skogen2020, for a review). Extant research has mostly operationalized digital media usage in terms of the extent to which users engage in different activities via digital platforms, for example through self-presentation and self-disclosure using social media and the frequency and time spent on these activities (e.g., Gil-Or, Reference Gil-Or, Levi-Belz and Turel2015; Manago et al., Reference Manago, Graham, Greenfield and Salimkhan2008; Masur & Scharkow, Reference Masur and & Scharkow2016; Michikyan, Reference Michikyan2019; Michikyan, Dennis, & Subrahmanyam, 2014a; Michikyan et al., Reference Michikyan, Subrahmanyam and Dennis2015; Qui et al., Reference Qiu, Lin, Leung and Tov2012; Twomey & O’Reilly, Reference Twomey and O’Reilly2017; Wright et al., Reference Wright, White and Obst2018). In the next subsections, we discuss the different ways that digital media usage has been operationalized and make recommendations for how usage can be conceptualized to study adolescent digital media use more accurately and meaningfully.

Digital Screen Time: As noted earlier in the methodological challenges section, digital screen time is frequently used in investigations of youths’ digital media use, and as a retrospective self-report measure, it is susceptible to both over- and underreporting (Parry et al., Reference Parry, Davidson, Sewall, Fisher, Mieczkowski and Quintana2021; Scharkow, Reference Scharkow2016). Here we examine whether total digital screen time, on its own, is a meaningful and accurate measure of digital media usage as related to psychological well-being (see Meier & Gray, Reference Meier and Gray2014; Orben & Przybylski, Reference Orben and Przybylski2019a, for a similar argument). Scholars (e.g., Orben, Reference Orben2020) have asked whether self-reported measures of digital screen time should be “retired.” The arguments for retiring self-reported measures of total digital screen time seem valid given that the size of the negative effects of digital media use on mental health is either nonexistent (Coyne et al., Reference Coyne, Rogers, Zurcher, Stockdale and Booth2020) or too small to have a practical significance or to warrant a meaningful scientific debate (Orben & Przybylski, Reference Orben and Przybylski2019b). In fact, some have argued that adolescent digital media users would need to spend a physically impossible amount of time using digital media – more than 63 hours per day – to experience noticeable decline in their well-being (see Orben & Przybylski, Reference Orben and Przybylski2019a). While it might be premature to “retire” the concept of self-reported measures of digital screen time without more research, it is important to “reimagine” it. A more meaningful approach to conceptualizing and operationalizing digital screen time would be to combine – the amount of time + activity + specific time frame + motive of use – within a single item or question (e.g., “How much time did you spend today chatting with friends on Instagram to tell them about your problems and troubles?”). Doing so would increase the meaningfulness and accuracy of measuring digital screen time. Another possibility is to combine objective measures of screen time (e.g., via tracking apps) with objective measures of content or activity, and a subjective measure of motivation (see Subrahmanyam et al., Reference Subrahmanyam, Frison and Michikyan2020; Underwood et al., Reference Underwood, Rosen, More, Ehrenreich and Gentsch2012). Future research should aim to tease apart the role of these different components – for instance, examining whether chatting with friends on Instagram about problems improves mental health when it happens for short durations of time, but undermines mental health when it happens for longer periods of time. Taken together, it appears that it might not the amount of digital screen time in and of itself that matters for psychological well-being, but rather, how adolescents use their time engaging with digital media.

Digital Activities: Social media platforms afford users a range of possible digital activities, and there is empirical evidence that different activities may differentially impact psychological well-being (Kross et al., Reference Kross, Verduyn, Sheppes, Costello, Jonides and Ybarra2020). For instance, digital activities that involve interacting with existing friends via text messaging can enhance well-being (Valkenburg & Peter, Reference Valkenburg and Peter2007), because certain features of text-based content may serve as “digital affiliative cues” that can facilitate emotional bonding (Sherman et al., Reference Sherman, Michikyan and Greenfield2013). Similarly, digital activities that involve visual cues or image-based content (e.g., photos) can also enhance well-being by decreasing loneliness and by increasing happiness and life satisfaction (Pittman & Reich, Reference Pittman and Reich2016), likely because photos can foster increased social connectedness (Bakhshi et al., Reference Bakhshi, Shamma and Gilbert2014; Goh et al., Reference Goh, Ang, Chua and Lee2009); it is worth noting that image-based content can also negatively impact well-being by increasing social comparisons, particularly those focused on appearance and body image (e.g., Lewallen & Behm-Morawitz, Reference Lewallen and Behm-Morawitz2016).

Other aspects of digital media activities that are relevant in adolescent social media use research are the type of activity and level of interactivity – or exchanges between users and between users and specific digital media features (see Stromer-Galley, Reference Stromer-Galley2004, for a detailed argument). Research examining youths’ digital media usage and psychological well-being has focused on the distinction between two types of social activities: “active” use vs. “passive” use (Kross et al., Reference Kross, Verduyn, Sheppes, Costello, Jonides and Ybarra2020; Schønning et al., Reference Schønning, Hjetland, Aarø and Skogen2020); “active” use (e.g., commenting on or responding to someone else’s content) has been found to enhance well-being (Escobar-Viera et al., Reference Escobar-Viera, Shensa and Bowman2018; Liu et al., Reference Liu, Baumeister, Yang and Hu2019), whereas “passive” use (e.g., reading comments or newsfeeds of hundreds of friends and followers without any participation or lurking) has been found to undermine psychological well-being (Escobar-Viera et al., Reference Escobar-Viera, Shensa and Bowman2018; Tandoc et al., Reference Tandoc, Ferrucci and Duffy2015; Underwood & Ehrenreich, Reference Underwood and Ehrenreich2017).

Although the terms “active” and “passive” capture some aspects of social media use that are relevant to mental health and well-being, other aspects of use are not captured by these terms. Consider one example of self-presentation via posting photos, which is considered an “active” use of digital media (Wang et al., Reference Wang, Wang, Gaskin and Hawk2017). Missing from the current conceptualization of “active” digital media usage, however, is whether and how interactive “active” usage is. For instance, when a user posts a picture, does the self-presentation occur during an interaction with other users, or does the user engage in self-presentation outside of an interaction, as a result of an exchange between the user and specific digital media features, or some combination of both? The current conceptualization of “passive” digital media usage places significant emphasis on content consumption with minimum interactivity (e.g., lurking) (Underwood & Ehrenreich, Reference Underwood and Ehrenreich2017). Even with content consumption, “passive” activities such as viewing humorous or inspiring social media content might improve mood and well-being. An important question is how passive “passive” digital media usage is. Is reading someone else’s comments really passive?

This begs the question of what “active” use and “passive” use in the current paradigm really capture (Valkenburg et al., Reference Valkenburg, van Driel and Beyens2021). It appears that the active- vs. passive-use paradigm emphasizes the behavioral component of digital media usage, while mostly ignoring its cognitive and affective components. A focus on the behavioral component of digital media usage assumes that “passive” use (e.g., lurking) is psychologically passive. Drawing from Bandura that “a theory that denies that thoughts can regulate actions does not lend itself readily to the explanation of complex human behavior” (Bandura, Reference Bandura1986, p. 15), we reject the premise that digital media usage is passive. Even “passive” digital media usage wherein the user chooses to observe other users’ activities online without any participation (e.g., reading someone else’s comments) involves some level of psychological activity (e.g., encoding, interpretation, and processing) (see Sherman et al., Reference Sherman, Greenfield, Hernandez and Dapretto2018; see also Bandura, Reference Bandura1997). Instead, we propose that digital media usage be viewed on an interactivity continuum with exchanges between users and between users and specific digital media features (Stromer-Galley, Reference Stromer-Galley2004). In this conceptualization of digital media usage, digital activities are assumed to be inherently active, although the level of interactivity might vary. Drawing from the developmental literature on peer interactions (Rubin et al., Reference Rubin, Bukowski, Parker, Eisenberg, Damon and Lerner2006), the terms reciprocal activities and parallel activities may better capture the distinction between different kinds of social media activities that are relevant to mental health and well-being.

Reciprocal activities involve exchanges via digital media platforms wherein users can take turns to respond and react to one another (e.g., texting, chatting, commenting on another user’s post, liking another user’s photo, etc.). Parallel activities involve actions via digital media platforms that do not engage other users in a particular exchange, or actions that do not involve reciprocity or response from other users (posting a status, a comment, or a photo to express one’s thoughts and feelings, watching a video/film, etc.). Reciprocal and parallel activities vary across the interactivity continuum, with reciprocal activities at the high end of the interactivity continuum and parallel activities at the low end. Given the complexity of digital media usage, reciprocal activities and parallel activities may also be dynamic – influencing one another and at varying levels of intensity. For instance, a user can engage in both high or low levels of reciprocal and parallel activities simultaneously (e.g., texting while reading someone else’s comment, or posting more than one photo and chatting with one or more persons at the same time). This distinction is one possibility and not intended to comprehensively capture all aspects of digital/social media activities. Another distinction proposed in the research is between content creation, production, and consumption (see Schønning et al., Reference Schønning, Hjetland, Aarø and Skogen2020); content curation and/or content distribution are increasingly important aspects of social media use. The point here is that when considering social media usage, researchers must capture the nuances when conceptualizing digital activities. The continuum of parallel to reciprocal activities can help as it can capture both the level and the characteristic of digital media usage – whether and how actively the user is interacting with others online and/or whether and how actively the user is “interacting” with a digital media platform and with themselves.

Consider a Variety of Mechanisms and User Variables as Mediators and Moderators

As noted elsewhere in the literature (Beyens et al., Reference Beyens, Pouwels, van Driel, Keijsers and Valkenburg2020; Subrahmanyam et al., Reference Subrahmanyam, Frison and Michikyan2020) and in this handbook, extant research on the relation between adolescent digital media use and psychological well-being has revealed no clear or consistent patterns, suggesting that the relation is complex. Thus, it is important for researchers to also consider underlying mechanisms as well as individual factors that may shape the pathways between digital media usage and mental health.

Possible Mechanisms: Digital media usage has been linked with different mechanisms that can either enhance or undermine mental health and well-being (Escobar-Viera et al., Reference Escobar-Viera, Shensa and Bowman2018; Yoon et al., Reference Yoon, Kleinman, Mertz and Brannick2019). For instance, researchers have demonstrated that specific digital activities (e.g., reading someone else’s comments without any participation), on the one hand, can undermine well-being through upward social comparison (i.e., comparing oneself with someone better off than oneself), negative self-evaluation (Wang et al., Reference Wang, Wang, Gaskin and Hawk2017), and rumination (Feinstein et al., Reference Feinstein, Hershenberg, Bhatia, Latack, Meuwly and Davila2013), as well as through feelings of envy (Appel et al., Reference Appel, Crusius and Gerlach2015) and fear of missing out (Oberst et al., Reference Oberst, Wegmann, Stodt, Brand and Chamarro2017; Przybylski et al., Reference Przybylski, Murayama, DeHaan and & Gladwell2013). On the other hand, digital activities wherein users peruse their own photos on social media might also improve psychological well-being through self-affirmation (see Toma & Hancock, Reference Toma and Hancock2013) and perhaps through downward comparison (i.e., comparing oneself with someone worse off than oneself).

Depending on the type of self-comparison, social media users may experience decreases or increases in their mental health and well-being. Not only do people compare themselves with others, but they also compare their “current self” with their “past self” and generally view themselves as improving over the years, despite how illusory this view may be (M. Ross & Wilson, Reference Ross and Wilson2003). Applying this to digital media usage, it is to be expected that users who engage in downward comparison (i.e., viewing the current self as better than the past self when comparing recent online photos with the earlier photos) may experience increases in their mental health. Although it remains to be seen, the effects of downward comparison may be of even greater significance for adolescent digital media users who are undergoing the task of developing a personal fable (Elkind, Reference Elkind1967; Erikson, Reference Erikson1959; Granic et al., Reference Granic, Morita and Scholten2020a, Reference Granic, Morita and Scholten2020b).

Possible User Variables: Examining user variables that may moderate the pathway between social media usage and well-being can also yield more nuanced insights about the ways that digital media use can enhance or undermine mental health. As an example, we focus on one variable, personality, to illustrate why researchers should consider individual factors as moderators and mediators when studying the relation between digital media usage and psychological well-being and mental health (Ehrenberg et al., Reference Ehrenberg, Juckes, White and Walsh2008; Kircaburun et al., Reference Kircaburun, Alhabash, Tosuntaş and Griffiths2020; Michikyan et al., Reference Michikyan, Subrahmanyam and Dennis2015; C. Ross et al., Reference Ross, Orr, Sisic, Arseneault, Simmering and Orr2009). Other potential moderators identified in prior research include age and gender (Booker et al., Reference Booker, Kelly and Sacker2018; Correa et al., Reference Correa, Hinsley and De Zuniga2010; Simoncic et al., Reference Simoncic, Kuhlman, Vargas, Houchins and Lopez-Duran2014), offline support (Hatchel et al., Reference Hatchel, Subrahmanyam and Negriff2019), and social anxiety (Hatchel et al., Reference Hatchel, Negriff and Subrahmanyam2018; Subrahmanyam et al., Reference Subrahmanyam, Frison and Michikyan2020).

Personality can be defined as a collection of generally stable characteristics that define the self across time and context (Zuckerman, Reference Zuckerman1991) – including traits such as extroversion, introversion, neuroticism, and openness to new experiences (Costa & McCrae, Reference Costa, McCrae, Boyle, Matthews and Saklofske2008). Social media users who are extroverted (e.g., outgoing, talkative) and who are open to new experiences (reflecting curiosity and novelty-seeking) appear to engage in self-enhancing digital activities (e.g., via posting selfies) (Sorokowska et al., Reference Sorokowska, Oleszkiewicz, Frackowiak, Pisanski, Chmiel and Sorokowski2016, Zywica & Danowski, Reference Zywica and Danowski2008), which can further enhance their interpersonal skills and psychological well-being. However, even among extroverts, users who are also experiencing psychological well-being concerns (e.g., lower life satisfaction) may be prone to problematic social media use such as addiction (Nikbin et al., Reference Nikbin, Iranmanesh and Foroughi2020).

Like their extroverted peers, introverted users who are shy or less outgoing and users who are moody (indicative of high neuroticism) also benefit from using social media (Simoncic et al., Reference Simoncic, Kuhlman, Vargas, Houchins and Lopez-Duran2014); however, these groups of users appear to utilize social media to compensate for a lack of offline social networks and a lack of confidence in their interpersonal skills (Ehrenberg et al., Reference Ehrenberg, Juckes, White and Walsh2008; C. Ross et al., Reference Ross, Orr, Sisic, Arseneault, Simmering and Orr2009). Since neuroticism can be manifested as loneliness and anxiety (Cattell & Mead, Reference Cattell, Mead, Boyle, Matthews and Saklofske2008), it is also not uncommon for social media users with high trait neuroticism to engage in frequent parallel activities involving self-presentation (e.g., posting comments and photos) (C. Ross et al., Reference Ross, Orr, Sisic, Arseneault, Simmering and Orr2009), which tend to be more elaborate (Bai et al., Reference Bai, Gao and Zhu2012), more negative (Kern et al., Reference Kern, Eichstaedt and Schwartz2014), and more socially desirable and less truthful (Michikyan et al., Reference Michikyan, Dennis and Subrahmanyam2014). It appears that the reluctance to engage other users via digital media might reflect social anxiety or the fear of being negatively evaluated by other users that is typically experienced by social media users with high trait neuroticism (Bowden-Green et al., Reference Bowden-Green, Hinds and Joinson2021). A major complication in the search for user variables is that different individual factors may interact both with one another as well as with contextual factors, often in a nonlinear way. Thus, the various ways in which different individual factors and contextual factors as well as different mechanisms interact with one another should be considered when conceptualizing the multiple ways digital media usage impacts adolescents’ psychological well-being and mental health.

Conclusions

As digital media are now entrenched in the lives of adolescents, they have become an important contextual influence along the lines of families, peer groups, and schools. Considerable research demonstrates the importance of friends and families in adolescent health and well-being, and it is similarly important to investigate the impact of digital media on adolescent well-being and mental health. This introductory chapter presented an overview of the terms and history of research on this topic and described some of the pressing methodological and conceptual issues confronting researchers investigating this topic. Our discussion highlighted two main themes: (1) Changes in technology are inevitable, and thus researchers will need to be flexible in the methodological approaches they adopt to investigate the short- and long-term implications of youths’ social media use; (2) Researchers must clearly articulate how they conceptualize and operationalize digital media, its role, usage, and pathways of influence. We present a few ways that researchers can adapt to the methodological challenges and clarify how they should innovate when conceptualizing and measuring adolescents’ digital media use. These are but a few suggestions, and we encourage researchers to build and expand on them as they investigate the growing presence of social media in adolescents’ lives.

The Selectivity Paradigm

The selectivity paradigm of media effects theories states that: (a) individuals can only attend to a limited number of media messages out of the wealth of media messages that can potentially attract their attention, (b) they select these media messages in response to dispositions, needs, and desires that differ from person to person, and (c) only those media messages they select have the potential to influence them. The selectivity paradigm is represented by two different communication theories: uses and gratifications theory (Katz et al., Reference Katz, Blumler and Gurevitch1973) and selective exposure theory (Zillmann & Bryant, Reference Zillmann, Bryant, Zillmann and Bryant1985). Both theories argue that a variety of cognitive and psychosocial factors guide and filter one’s selective media use. An important difference between the theories is that uses and gratifications theory conceives of media users as rational and conscious of their selective media use, whereas selective exposure theory argues that media users are often not aware, or at least not fully aware, of their selection motives.

The Transactionality Paradigm

The transactionality paradigm is an extension of the selectivity paradigm. Early studies into the selectivity paradigm have predominantly focused on the extent to which the dispositions of media users (e.g., needs, moods, attitudes) predict their tendency to select media. In other words, these studies conceptualized selective media use as the outcome, whereas the effects of this media use received less attention. In more recent transactional media effects theories (e.g., Slater, Reference Slater2007; Valkenburg & Peter, Reference Valkenburg and Peter2013a), the selectivity paradigm has become an integrated part of the media effects process. Transactional media effect theories argue that (a) the media user, rather than the media, is the starting point of a process that leads to selective media use, (b) this selective media use may bring about a transaction (i.e., change) in the media user, which is the media effect, and (c) this media effect may, in turn, reciprocally influence media use and the antecedents of media use. For example, it has been shown that adolescents high in trait aggressiveness are more likely to selectively expose themselves to violent websites, which may further enhance their trait aggressiveness (Slater, Reference Slater2003).

The propositions in transactional media effects theories have important implications for theories and research on the effects of social media. First, in comparison with mass media, social media have more filters and algorithms to cater to the preferences of adolescent users, which may stimulate their selective exposure to messages that match these preferences. Second, social media platforms typically allow adolescents to make their posts more personal, vivid, and emotional, which may enhance the likelihood of effects. Third, since 2017, adolescents can not only search for messages related to a specific hashtag but can also follow one or more hashtags, after which posts under these hashtags start to show up more prominently in the users’ timelines or feeds (Scherr et al., Reference Scherr, Arendt, Frissen and Oramas2020). In comparison with mass media content, such posts may be more effective both in attracting the selective attention of recipients of these posts, and in influencing their cognitions, attitudes, and behavior (e.g., Parmelee & Roman, Reference Parmelee and Roman2020).

Following transactional theories, social media use may thus result in selective exposure to messages that match with individuals’ preexisting dispositions (e.g., needs, moods, attitudes), more so than mass media use. These theories thus imply that social media users may also more than mass media users be able to shape their own media effects via this targeted selective social media use. Hence, if we want to understand the effects of social media use on adolescents, we may need to study the antecedents that shape their selective social media use. Selective exposure theories have mostly focused on dispositional antecedents, such as mood and preexisting attitudes. But according to Valkenburg & Peter’s (Reference Valkenburg and Peter2013a) differential susceptibility to media effects model (DSMM), three types of antecedents may predict adolescents’ selective (social) media use and, thus, the effects of this use: dispositional, developmental, and social-context factors.

The Conditionality Paradigm

The conditionality paradigm is closely linked with the selectivity and transactionality paradigms. After all, in both paradigms it is argued that only the messages that individuals select in response to person-specific antecedents have the potential to influence them. Theories that propose conditional media effects share the notion that media effects (a) do not equally hold for all media users, and (b) can be enhanced or reduced by dispositional, developmental, and social-context factors (Valkenburg & Peter, Reference Valkenburg and Peter2013a). In line with earlier media effects theories (e.g., Bandura, Reference Bandura, Bryant and Oliver2009), Valkenburg and Peter’s DSMM postulates that dispositional, developmental, and social-context factors may have a double role in the media effects process: They not only predict media use, but they also influence the way in which media messages are processed and subsequent distal media outcomes. This twofold influence results in three types of differential susceptibility to media effects: dispositional, developmental, and social-context susceptibility.

Social Media Users Shape Their Own Effects

The first proposition is that (1) social media users can individually (or collectively) shape their social media context, and (2) their experiences within this social media context can shape the effects of this context. The first part of this proposition is in line with transactional theories of development and Gibson’s (Reference Gibson1979) affordance theory. Transactional theories of development agree that children can shape and be shaped by their experienced social contexts (Bronfenbrenner, Reference Bronfenbrenner and Bronfenbrenner2005; Sameroff, Reference Sameroff and Sameroff2009). Likewise, Gibson argued that individuals tend to alter their environment by adjusting its affordances to better suit their needs and desires. In other words, an individual’s perceptions of the affordances of a context may lead to specific uses of this context, which in turn shape the experienced effects of this context. A similar proposition has been raised in self-effects theory (Valkenburg, Reference Valkenburg2017), which proposes that social media users carefully craft their messages (e.g., social media posts), which may influence the recipients of these messages (i.e., the social environment) but also the message senders themselves, directly via internalization of overt behavior (Bem, Reference Bem and Berkowitz1972), or indirectly, via the feedback that their messages elicit.

The first part of this proposition, that social media users can individually (or collectively) shape their social media context, has received support. Adolescents can (co)create both the affordances and norms of the social media contexts in which they participate. It has been found, for example, that the sharing of intimate, self-related information is more accepted in the social media context than in equivalent offline contexts (Christofides et al., Reference Christofides, Muise and Desmarais2009). Another (co)created norm is that the sharing of negative emotions is more accepted in private (e.g., WhatsApp) than public social media contexts (e.g., Instagram; Waterloo et al., Reference Waterloo, Baumgartner, Peter and Valkenburg2017). And if adolescents do want to share intimate, self-related information on a public social medium like Instagram, they sometimes turn to a Finsta (a Fake Instagram account where they can be honest and show their true self) in addition to a Rinsta (a Real Instagram account used to post their positive experiences). Finally, overly emotional expressions on in public social media are considered norm violations (Waterloo et al., Reference Waterloo, Baumgartner, Peter and Valkenburg2017).

The second part of this proposition, that adolescents’ unique experiences within their (co) created social media context can shape the effects of this context, has also received support. For example, message recipients can selectively and autonomously expose themselves to uplifting or depressing social media messages, which may subsequently affect their well-being in unique ways. In a qualitative study of Rideout and Fox (Reference Rideout and Fox2018), one adolescent reported: “If I’m feeling depressed, getting on Twitter and seeing funny tweets or watching funny videos on YouTube can really brighten my mood” (p. 20). In this example, a transient dispositional variable (low mood) shaped this adolescent’s selective exposure, which in turn positively shaped their experienced effect (i.e., a brightened mood). In the same study, another adolescent’s preexisting low mood resulted in an opposite effect of social media browsing (i.e., a worsened low mood): “Social media makes me feel worse when I’m scrolling through feeds and seeing news headlines and posts about how terrible something is” (Rideout & Fox, Reference Rideout and Fox2018, p. 19). And yet another adolescent with a preexisting low mood reacted with selective avoidance: “Usually friends post happy things – getting together with others, accomplishments, bragging. I don’t always want to see it when I’m feeling down about myself so I stay off social media” (p. 20).

These qualitative finding illustrate the complex nature of the associations between preexisting disposition (i.e., low mood), selective exposure to social media messages, and postexposure mood. Mood-induced selective exposure to social media messages can enhance mood (adolescent 1), worsen mood (adolescent 2), and it can lead to selective avoidance (adolescent 3). Such unique differences have also been reported in two recent experience sampling studies by Beyens et al. (Reference Beyens, Pouwels, van Driel, Keijsers and Valkenburg2020, Reference Beyens, Pouwels, van Driel, Keijsers and Valkenburg2021), who found considerable differences in experienced effects of social media use. In one study, they found that 46% of the participating adolescents felt better after social media browsing in the past hour, while 44% did not feel better or worse, and 10% felt worse after such use (Beyens et al., Reference Beyens, Pouwels, van Driel, Keijsers and Valkenburg2020).

Such uniquely experienced social media effects also seem to hold for message senders. Several studies have shown that message sending (e.g., posting) can improve the well-being of message senders (Verduyn et al., Reference Verduyn, Ybarra, Résibois, Jonides and Kross2017), a result that has often been explained by the positive feedback that message senders receive (Verduyn et al., Reference Verduyn, Ybarra, Résibois, Jonides and Kross2017). However, social media–induced improvements in well-being can also occur without any involvement of fellow users (Pingree, Reference Pingree2007; Valkenburg, Reference Valkenburg2017). Self-expressions on social media, especially when their intended audience is sizeable, may lead to internalization of these self-expressions, for example, via self-perception. Self-perception theory (Bem, Reference Bem and Berkowitz1972) argues that individuals infer their internal self-concept from retrospectively observing their own overt behavior. If these individuals share positive self-expressions induced by the positivity norm in public social media, these individuals may, due to a desire for a consistency between their overt behavior and their self-concept, adjust their self-concept to match their behavior. For a discussion of self-effects in social media, and the mechanisms that may explain such effects, such as cognitive reframing, biased scanning, and public commitment, see Valkenburg (Reference Valkenburg2017).

Social Media as a Micro- and Mesosystem

A second proposition of a transactional affordance theory of social media uses is that the social media context is a micro-level context, in which effects on participants are more likely than in the mass media context. Bronfenbrenner was one of the first to conceptualize the relationship between individuals and their social contexts. He distinguished between four types of contexts: the micro-, meso-, macro-, and exosystem (Bronfenbrenner, Reference Bronfenbrenner1979, Reference Bronfenbrenner and Bronfenbrenner2005). The microsystem involves direct interactions of the child with their most proximal circle, such as the family, peer group, or neighborhood. The mesosystem represents the possible interactions among these microsystems (e.g., between the family and peer group), whereas the macrosystem refers to the overarching culture or subculture of children. Bronfenbrenner’s fourth context, the exosystem, refers to social contexts that do not allow the child as an active participant but that have the potential to affect the child. An example of an exosystem is the work context of one of the parents of the child. A child cannot actively participate in this context but can in many ways be influenced by it.

At the time of the development of his theory, Bronfenbrenner identified the mass media as an exosystem because it did not allow for active involvement of adolescents, even though it could shape their experiences. Although valid at the time, Bronfenbrenner (1917–2005) could not have foreseen the rapid developments within the media landscape. If he could have, he would probably have categorized the social media context as a microsystem rather than an exosystem. After all, unlike before, the media landscape now does allow for, and even stimulates, direct interactions among participants. For example, idols, an important source of identity formation in adolescence, have been transferred from the exosystem to the microsystem: Whereas movie stars or pop singers used to be celebrities that adolescents could admire from an unsurmountable distance, social media now provide them with ample opportunity for direct communication with their idols. In fact, many of their contemporary idols are YouTubers or Instagram influencers with whom they can directly interact.

If Bronfenbrenner could, he may now also have identified the social media context as part of the mesosystem because it allows for, or even stimulates, interactions with other microsystems (e.g., the family or the peer contexts). Although every traditional microsystem is in part “permeable” to the influences from other microsystems (e.g., family to peers and vice versa; family to school and vice versa), the social media context might be much more permeable to such influences. Conversely, the social media context seems to have penetrated all other microsystems in which adolescents participate, ranging from the family and peer context to the school.

However, if we accept the social media context as a microsystem, we must acknowledge that this context may, due to its proximity, dynamic, and ubiquitous nature, enhance the likelihood of effects on its participants, certainly when compared to the traditional mass media context. And if we accept the social media context as a part of the mesosystem (interactions among microsystems), we need to acknowledge that it may interact with the norms and affordances of other microsystems, such as parents or the school. And such interactions do occur. For example, preventing or counteracting possible negative consequences of social media interactions, and explaining to adolescents that the social media context may not be as perfect as it often appears, are important ingredients of today’s media-specific parenting and school-based prevention and intervention programs (Valkenburg & Piotrowski, Reference Valkenburg and Piotrowski2017).

It Is the Subjective Experience That Counts

A third and final proposition of a transactional affordance theory of social media uses is that the unique way in which individuals experience the norms and affordances of the social media context is the driving force of transactional effects between individuals and this context. This proposition is consistent with both transactional theories of development (Bronfenbrenner, Reference Bronfenbrenner and Bronfenbrenner2005; Sameroff, Reference Sameroff and Sameroff2009) and Gibson’s affordance theory (Gibson, Reference Gibson1979). Affordances, according to Gibson, are the unique ways in which individuals experience the utility of objects. For example, distinct individuals may all perceive another utility of a bottle (e.g., as a water container, a vase, a candle holder, or a weapon). However, to understand such individual differences in experiences of the affordances of social media, I first specify some of these affordances and argue how and why these affordances differ from other micro-level social contexts, such as the family or peer contexts.

A growing number of social media scholars have ventured to identify specific affordances of social media (boyd, Reference boyd and Papacharissi2011; McFarland & Ployhart, Reference McFarland and Ployhart2015; Nesi et al., Reference Nesi, Choukas-Bradley and Prinstein2018a, Reference Nesi, Choukas-Bradley and Prinstein2018b; Sundar et al., Reference Sundar, Jia, Waddell, Huang and Sundar2015; Treem & Leonardi, Reference Treem and Leonardi2013; Valkenburg & Peter, Reference Valkenburg and Peter2011; Valkenburg & Piotrowski, Reference Valkenburg and Piotrowski2017). Some of these scholars have identified four affordances (Treem & Leonardi, Reference Treem and Leonardi2013), others have focused on seven (Nesi et al., Reference Nesi, Choukas-Bradley and Prinstein2018a; Valkenburg & Piotrowski, Reference Valkenburg and Piotrowski2017) or even eight affordances (McFarland & Ployhart, Reference McFarland and Ployhart2015). Many comparable affordances appear in different studies but sometimes under different names (e.g., identifiability vs. cue absence; scalability vs. publicness). In this chapter, the focus is on three affordances that have been mostly identified in earlier literature. For each affordance, I discuss the scarce evidence of individual differences in the perceptions of its utility, as well as its potential consequences for both senders and recipients of social media messages. A more elaborate discussion of these consequences can be found in Nesi et al. (Reference Nesi, Choukas-Bradley and Prinstein2018a, Reference Nesi, Choukas-Bradley and Prinstein2018b)