Quantity of Digital Media Use and Problem Behavior

Some recent studies have suggested that more frequent social media use is tied to more concurrent conduct problems and delinquency among both younger (Ohannessian & Vannucci, Reference Ohannessian and Vannucci2020) and older (Galica et al., Reference Galica, Vannucci, Flannery and Ohannessian2017) adolescents. However, these cross-sectional associations have not entirely held up in longitudinal research, as seen in a recent study where time online was linked to later internalizing symptoms and to comorbid internalizing and externalizing symptoms, but not externalizing symptoms in the absence of internalizing (where externalizing was measured as a combination of inattention, impulsivity, and antisocial behavior; Riehm et al., Reference Riehm, Feder and Tormohlen2020). Similarly, our own research suggests that social media use and phone ownership in early adolescence are not associated with later conduct problems (once baseline conduct problems are accounted for) and that days on which young adolescents use more technology for a variety of purposes do not tend to be days when they report a greater likelihood of conduct problems (Jensen et al., Reference Jensen, George, Russell and Odgers2019). However, some longitudinal associations have been found: Research with Korean adolescents suggests that technology use for entertainment is related with later online and offline delinquency, and internet use for communication is related to later offline delinquency (though internet use for information seeking seems to protect against offline delinquency; Lim et al., Reference Lim, Kim and You2019). Other studies have investigated the opposite direction of effects (that earlier conduct problems might increase later social media engagement), which has been supported from adolescence (delinquency) into young adulthood (social media use; Galica et al., Reference Galica, Vannucci, Flannery and Ohannessian2017) but not from childhood (behavior problems) into adolescence (screen time; Männikkö et al. Reference Männikkö, Ruotsalainen, Miettunen and Kääriäinen2020). Taken together, the displacement hypothesis is not strongly supported by the literature (i.e., there is little evidence that those youth who are online most are getting into less trouble) and there is considerable inconsistency in findings around whether digital media engagement might be linked with higher problem behaviors over time. More experimental, longitudinal, and ecologically valid research is needed in this domain.

Overlap between Online and Offline Delinquency

Online delinquent and problem behavior can take many forms. A commonly used typology classifies cybercrime and cyberdeviance into four types: cybertresspass (e.g., malware), cyberpornography, cyberviolence (e.g., cyberbullying, trolling, flaming), and cyberdeception and theft (e.g., digital piracy; Graham & Smith, Reference Graham and Smith2019; Wall, Reference Wall2001). For instance, some youth trespass into off-limits online spaces in ways that could have severe criminal penalties (e.g., cracking into bank accounts) whereas others trespass in ways that are less likely to be prosecuted but nonetheless problematic (e.g., hacking into a peer’s social media account). The prevalence of these (usually covert) behaviors among teenagers is understudied and hard to ascertain, but surveys from the security industry suggest that up to 40% of youth have hacked into a social media account, email, or bank account (primarily “for fun” and “out of curiosity;” Richet, Reference Richet2013).

In reality, the line between online and offline spaces in delinquency is a blurry one. Indeed, emerging evidence suggests that long-standing types of offline delinquency now also manifest online, and the two contexts are not entirely separable. For example, qualitative interviews with ex-gang members and violence-prevention workers have revealed the existence of so-called digitalist gangs (Whittaker et al., Reference Whittaker, Densley and Moser2020) who use social media as a tool for attention for themselves and their gang. These gangs are more likely to be newer and less established (compared to less digitally connected “traditionalist” gangs), and to engage in activities like boasting, taunting, and posting videos of violent confrontations online. These types of online posts can serve to spark very real offline violence, as seen in the so-called Twitter feuds covered by the popular press (Patton et al., Reference Patton, Eschmann and Butler2013). In a recent study of Black youth involved in gangs in Chicago, 11% of posts included a picture of a gun, although not all these pictures were necessarily shared with aggressive intent (Patton et al., Reference Patton, Frey and Gaskell2019). Further, research suggests that gang members are more likely than nongang members to engage online in piracy, harassment, threats, and the facilitation of drug sales, assault, theft, and robbery (Pyrooz et al., Reference Pyrooz, Decker and Moule2015), suggesting considerable overlap between online and offline crime.

Youth who engage in delinquent behavior in both online and offline formats may be at particular risk. A recent study found that those adolescents (ages 12–17) who committed both online and offline delinquency were the most likely to experience increased risk factors and fewer protective factors, whereas the online delinquency only group had fewer risk and more protective factors and the offline delinquency only group fell in between the two (Rokven et al., Reference Rokven, Weijters, Beerthuizen and van der Laan2018). In a rare longitudinal study, Korean youth who engaged in cyber-delinquency were more likely to report more engagement in later offline delinquency (Nam, Reference Nam2020), which may suggest that, at least for some, online delinquency may serve as a gateway to later offline (and potentially higher consequence) crime.

Online Depictions of Offline Delinquency

In addition to delinquent acts performed online, social media can be used to portray delinquent acts performed offline. A study of undergraduate students revealed that exposure to online depictions of delinquency (including abusing an intimate partner, illegally carrying a weapon, physical fighting, selling drugs, driving while under the influence, setting fire to property, stealing, and vandalism) was frequent, with 81% of students being exposed to at least one offending behavior online (McCuddy & Vogel, Reference McCuddy and Vogel2015). Furthermore, those students who viewed more delinquency in their online social networks were more likely to engage in delinquent behaviors themselves (though this was a much stronger association in smaller social networks). Unfortunately, the cross-sectional nature of this study does not allow us to ascertain the direction of effects (i.e., whether youth who engage in delinquent behaviors are more likely to affiliate with other youth who do so and post about it online, or whether exposure to online depictions of delinquency may shift youth norms and behaviors).

In an innovative program of research, the Blackberry project (Underwood et al., Reference Underwood, Rosen, More, Ehrenreich and Gentsch2012) has followed a sample of students (and their text messages) over the course of high school. Qualitative coding of real, naturalistic text message data has revealed that most of these teens engaged in at least some antisocial text messaging, and that this text messaging about antisocial activities was associated with increases in multiple reporters’ accounts of rule-breaking behavior (Ehrenreich et al., Reference Ehrenreich, Underwood and Ackerman2014). Furthermore, findings suggest that the reason for associations between peer network delinquent texting topics and youth externalizing problems might be better characterized as selection (externalizing adolescents choosing deviant peer groups) rather than socialization (deviant peer groups driving externalizing behavior; Ehrenreich et al., Reference Ehrenreich, Meter, Jouriles and Underwood2019).

Quantity of Digital Media Use and Online and Offline Aggression

Some have asked whether level of engagement with digital media (e.g., time spent online) presents a risk factor for cyber and traditional aggression. In a recent meta-analysis, links between general social media use and offline violence-related behaviors could not be formally synthesized because only three studies were available; however, the available studies each show that youth who are using social media more frequently tend to report more concurrent violence-related behaviors (Vannucci et al., Reference Vannucci, Simpson, Gagnon and Ohannessian2020). Some cross-sectional research has also suggested that adolescents who spent more time online were more likely to be cyberbullying perpetrators (Hinduja & Patchin, Reference Hinduja and Patchin2008), with those who spend particularly high and problematic levels of time online being at the most risk (Kircaburun et al., Reference Kircaburun, Demetrovics, Király and Griffiths2020) and those with particularly low levels of time being (understandably) at very low risk of cyber-perpetration (Zych et al., Reference Zych, Farrington and Ttofi2019). It may be that in the average range of technology use, time online and time on social media are not closely related to cyberbullying perpetration.

Overlap between Online and Offline Aggression

Youth who perpetrate bullying online appear to mostly be the same youth who perpetrate bullying offline (Fanti et al., Reference Fanti, Demetriou and Hawa2012; Hinduja & Patchin, Reference Hinduja and Patchin2008; Olweus, Reference Olweus2012; Sourander et al., Reference Sourander, Klomek and Ikonen2010) as confirmed by a meta-analysis that concluded that traditional bullying perpetration is among the strongest predictors of online bullying perpetration (Kowalski et al., Reference Kowalski, Giumetti, Schroeder and Lattanner2014). It is common for cyberbullying perpetrators and victims to know one another in person – for example in 57% of the cyberbullying cases at a high school the victim reported that the perpetrator was a schoolmate (P. K. Smith et al., Reference Smith, Mahdavi, Carvalho, Fisher, Russell and Tippett2008). In a profile analysis, youth who engaged in cyberbullying tended to engage in all other types of bullying as well (relational, verbal, and physical offline bullying) and were at elevated risk for other externalizing behaviors (e.g., using substances and carrying weapons; Wang et al., Reference Wang, Iannotti and Luk2012). A longitudinal analysis of the transactional associations between face-to-face bullying perpetration and cyberbullying perpetration found that higher levels of earlier offline bullying perpetration predicted increases in cyberbullying perpetration (controlling for previous cyberbullying perpetration), but cyberbullying perpetration did not predict increases in offline bullying perpetration (Espelage et al., Reference Espelage, Rao, Craven, Bauman, Cross and Walker2012); this suggests that cyberbullying does not appear to be a first foray that grows into later offline bullying perpetration, but rather that offline bullying perpetration may come to extend to online environments.

Exposure to Online Violent Content and Offline Aggression

The impact of exposure to violent content in video games has been much talked of and controversial. Scholars have proposed that violent video games normalize aggression and can elicit and reward aggressive cognitions (e.g., hostile attributions), quick violent reactions, and aggressive fantasies (Gentile et al., Reference Gentile, Li, Khoo, Prot and Anderson2014), though others have noted that selection effects are also likely at play (Breuer et al., Reference Breuer, Vogelgesang, Quandt and Festl2015; Heiden et al., Reference Heiden, Braun, Müller and Egloff2019). Early in the field’s history, a meta-analysis of early video game research concluded that evidence strongly supports exposure to violence in video games as a causal risk factor for increased aggressive behavior (Anderson et al., Reference Anderson, Shibuya and Ihori2010), but this finding has not entirely held up over time, with more recent registered reports (e.g., Przybylski & Weinstein, Reference Przybylski and Weinstein2019) and meta-analyses of high-quality longitudinal studies finding zero to tiny associations between violent video gaming and later violent behavior (Drummond et al., Reference Drummond, Sauer and Ferguson2020). One domain that has not yet been extensively researched is that of the potential intersections between social aspects of online gaming and in-game aggression, which has gained growing attention with the advent of online multiplayer gaming (with live video, audio, and or/chat streams; Freeman, Reference Freeman2018). More information is needed on whether the synchronous and semi-anonymous online multiplayer gaming context may socialize and/or reinforce youth verbal (e.g., hate speech, insults) or even serious physical aggression (e.g., the phenomena of SWATting; Lamb, Reference Lamb, Kelly, Lynes and Hoffin2020) in ways not yet captured in the literature to date.

Quantity of Digital Media Use and Sexual Risk Taking

In a recent meta-analysis, the average association (across 14 cross-sectional studies) between social media use and sexual risk taking was r = 0.21 (95% CI 0.15, 0.28), representing a small to medium significant association, with stronger associations for younger adolescents and very small associations for later adolescents (Vannucci et al., Reference Vannucci, Simpson, Gagnon and Ohannessian2020). Three of these studies included in the meta-analysis captured online sexual acts, including frequency of sexy online presentation (Vandenbosch et al., Reference Vandenbosch, Beyens, Vangeel and Eggermont2016), frequency of risky sexual online self-presentation (Koutamanis et al., Reference Koutamanis, Vossen and Valkenburg2015), and frequency of sending sexts (Gregg et al., Reference Gregg, Somers, Pernice, Hillman and Kernsmith2018) whereas the remaining 11 studies captured more traditional indicators of adolescent risky sexual behavior. It does, then, appear that social media use and sexual risk taking tend to co-occur, though the cross-sectional nature of all studies makes it impossible to parse the direction of effects.

Exposure to Online Sexual Content and Offline Sexual Risk Taking

Exposure to sexual content online (e.g., internet pornography) has been linked to offline sexual risk taking, though, as with much research reviewed in this chapter, a lack of longitudinal or experimental designs limits ability for causal inference. For instance, a meta-analysis of six cross-sectional studies revealed that exposure to sexually explicit websites was linked to higher odds of intercourse without a condom in two studies and was perhaps related to having ever had sexual intercourse and having had multiple partners, though significant statistical heterogeneity made meta-analysis difficult, and most studies were weakened by their limited accounting for important potential confounding variables (L. W. Smith et al., Reference Smith, Liu and Degenhardt2016). In a relevant experiment on social norms, young adults who were assigned to and viewed sexual content posted by “peers” in a lab-generated Facebook feed tended to estimate that more of their peers engaged in sex without a condom, and in turn expressed higher willingness to engage in this risky behavior themselves (relative to young adults assigned to view nonsexual content on the Facebook feed; S. D. Young & Jordan, Reference Young and Jordan2013). This highlights the important role of descriptive norms in intentions around risky behaviors and is consistent with longitudinal research that shows that adolescents’ self-report of exposure to online sexual content is related to normative beliefs and, in turn, increased likelihood of intentions to engage in and actual sexual behavior (Bleakley et al., Reference Bleakley, Hennessy, Fishbein and Jordan2011).

Sexting, Cybersex and Offline Sexual Risk

Sexting refers to the exchange of sexually explicit text or images, usually via private messaging, in a way that need not be synchronous or reciprocal (Daneback et al., Reference Daneback, Cooper and Månsson2005). Cybersex is a related concept that can occur via computer (rather than just by text or private message) and encompasses synchronous sexual talk and/or behaviors with a partner over video, voice, or text chat and that often includes an element of sexual gratification through masturbation (Daneback et al., Reference Daneback, Cooper and Månsson2005; Judge & Saleh, Reference Judge, Saleh and Rosner2013). Although sexting and cybersex share some features with other types of exposure online to sexual content (e.g., pornography), they are also distinct, as they are usually characterized as more interactive as opposed to one-sided consumption.

Sexting is prevalent in adolescence, with between a quarter to a half of teens reporting engaging in sexting to some extent (Baiden et al., Reference Baiden, Amankwah and Owusu2020; Frankel et al., Reference Frankel, Bass, Patterson, Dai and Brown2018; Maheux et al., Reference Maheux, Evans, Widman, Nesi, Prinstein and Choukas-Bradley2020). Sexting can take many forms, with qualitative research with emerging adults revealing that sexting occurs in various relational contexts including casual sexual, dating and intimate relationships, and nonsexual peer contexts (Burkett, Reference Burkett2015). A study conducted in Belgium found high rates of textual and visual online sexual behavior (with consistently higher rates among boys than girls); about half of teens (55% of boys, 40.6% of girls) had attempted to sexually arouse their romantic partner via online communication, 20% of teens reported sending sexy pictures to a dating partner, and 7.6% of adolescents reported undressing in front of a webcam for a romantic partner (Beyens & Eggermont, Reference Beyens and Eggermont2014). A profile analysis of adolescent women revealed that they tended to follow one of four patterns with relation to online sexual behavior: abstinent, participating in multiple behaviors including risky behaviors, mostly seeking sexual content, and mostly receiving sexual contacts (Maas et al., Reference Maas, Bray and Noll2018). Motivations for sexting include sexual arousal, humor, flirtation, and seeking reassurance about appearance. Sexting and cybersex are in some ways normative (and present little risk for negative outcomes like STI and unintended pregnancy) but can also carry their own risks, including receiving unwanted and unsolicited sexts, privacy violations, and feeling pressured to engage in sexting (Burkett, Reference Burkett2015).

Cross-sectional research seems to suggest that those youth who are more sexually active and (to a somewhat lesser extent) who engage in certain types of sexual risk behaviors are also more likely to be engaged in sexting (Frankel et al., Reference Frankel, Bass, Patterson, Dai and Brown2018; Romo et al., Reference Romo, Garnett and Younger2017), with photo-based sexting being more strongly tied to offline sexual activity than text-based sexting (Houck et al., Reference Houck, Barker, Rizzo, Hancock, Norton and Brown2014). A meta-analysis of 8 studies that examined sexting risk for sexual and risky sexual behaviors concluded that those youth who sexted were significantly more likely to be sexually active, to have had multiple past year partners, and to have used alcohol or drugs before sex (L. W. Smith et al., Reference Smith, Liu and Degenhardt2016). A separate meta-analysis of 15 studies (14 cross-sectional) with a wider age span (including adolescents and young adults) found that youth who engage in sexting are moderately more likely to have lifetime and recent sexual experience, and slightly more likely to engage in unprotected sex and have more sexual partners (Kosenko et al., Reference Kosenko, Luurs and Binder2017). Rare longitudinal studies on this topic suggest that sexting may serve to increase risk for later offline sexual activity and risk taking. For instance, one study concluded that sexting is associated with later sexual activity but not with later risky sexual activity (sex without a condom, substance use before sex, and multiple sexual partners; Temple & Choi, Reference Temple and Choi2014). Similarly, degree of engagement with chat rooms, dating websites, and erotic contact websites has been associated with later sexual activity in both sexually experienced and nonsexually experienced Belgian adolescents (Vandenbosch et al., Reference Vandenbosch, Beyens, Vangeel and Eggermont2016). Finally, a study of objectively coded text message content suggests that evidence of sexting at age 16 was associated with reporting an early sexual debut, having sexual intercourse, having multiple sex partners, and engaging in drug use in combination with sexual activity two years later (Brinkley et al., Reference Brinkley, Ackerman, Ehrenreich and Underwood2017). This is consistent with a profile analysis that suggested that youth who engaged in the riskiest behavior over time engaged in both online sexual risk behaviors (e.g., sexting or arranging a sexual encounter with someone met only online) and offline sexual risk behaviors (e.g., hooking up and unprotected sex; Baumgartner et al., Reference Baumgartner, Sumter, Peter and Valkenburg2012).

As with the other outcomes reviewed here, more longitudinal and experimental research is needed to ascertain what drives these associations: Are sexually active youth more likely to also express that sexuality in sexting? Does sexting serve as a gateway to later in-person sexual behaviors and risk taking? Are sexting, sexual activity, and sexual risk taking driven by other risk factors (e.g., disinhibition; Dir & Cyders, Reference Dir and Cyders2015)? Only well-designed empirical studies will tell.

Quantity of Digital Media Use and Substance Use

On the whole, research does seem to suggest that those youth who are most engaged with digital media are at least somewhat more likely to misuse alcohol and other substances. This is captured in a recent meta-analysis that identified 14 cross-sectional studies of amount social media use and adolescent substance misuse, with an average pooled effect size of r = 0.19, in the small to moderate range (Vannucci et al., Reference Vannucci, Simpson, Gagnon and Ohannessian2020). Individual study findings suggested that adolescents who are more engaged with social media are also more likely to report regular alcohol use and binge drinking, tobacco use, and marijuana use compared to those who are less digitally connected (Gommans et al., Reference Gommans, Stevens, Finne, Cillessen, Boniel-Nissim and ter Bogt2014; Kaufman et al., Reference Kaufman, Braunschweig and Feeney2014; Ohannessian et al., Reference Ohannessian, Vannucci, Flannery and Khan2017; Sampasa-Kanyinga & Chaput, Reference Sampasa-Kanyinga and Chaput2016; Spilková et al., Reference Spilková, Chomynová and Csémy2017). These associations also seem to persist in adolescents even once potential confounds of impulsivity, sensation seeking, peer relationships, and symptoms of depression are controlled for (Brunborg et al., Reference Brunborg, Andreas and Kvaavik2017). One recent longitudinal study suggested that frequency of social media posting and “checking in” on social media was associated with greater likelihood of subsequent initiation of tobacco and cannabis use, though other types of digital media use (e.g., “chatting and shopping” and “reading news/articles and browsing photos) were less consistently linked to risk of subsequent tobacco and cannabis initiation (Kelleghan et al., Reference Kelleghan, Leventhal and Cruz2020). Of note, some research has suggested that much of these observed associations may be due to exposure to alcohol-related content on social media, and that once this mediator is partialed out there is no unique association between digital media engagement and alcohol use (Erevik et al., Reference Erevik, Torsheim, Andreassen, Vedaa and Pallesen2017). We thus turn our attention next to the types of alcohol-related content posted and viewed on social media.

Alcohol- and Drug-Related Posting and Substance Use Behaviors

Adolescents post about substance use on social media in a myriad of ways and for various purposes. These can include text-based posts describing alcohol attitudes, intentions, and behaviors (that make up over half of youth alcohol-related posts) as well as image-based alcohol depictions (Moreno et al., Reference Moreno, Cox, Young and Haaland2015). For the most part, when images featuring alcohol or other substances are shared on social media, they tend to be posted by someone in the picture rather than others (Morgan et al., Reference Morgan, Snelson and Elison-Bowers2010). and alcohol depictions tend to be incidental images (e.g., a person holding a drink while a photo is taken) rather than the primary focus of the image (e.g., a picture of drinking games or a person visibly drunk; Hendriks et al., Reference Hendriks, Gebhardt and Van Den Putte2017). Among this sample of Dutch young people aged 12–30, alcohol posting among adolescents under age 18 (legal drinking age) was rare, but young adults endorsed mostly posting images that include alcohol for “entertainment” and choosing not to post alcohol-related images because they thought it was “stupid,” because they drank little, to reduce risk of a future employer seeing it, and because it was not consistent with their identities (Hendriks et al., Reference Hendriks, Gebhardt and Van Den Putte2017). A distinction between legality or illegality of behavior is also relevant for marijuana depictions on social media, which an even larger majority of youth see as inappropriate to post (Lauckner et al., Reference Lauckner, Desrosiers, Muilenburg, Killanin, Genter and Kershaw2019). Nonetheless, when adolescents post about substance use on social media, posts are usually positive in nature, pro-alcohol posts outnumber anti-alcohol posts by a factor of more than 10, and negative consequences of use (e.g., hangovers or embarrassment) are rarely depicted (Cavazos-Rehg et al., Reference Cavazos-Rehg, Krauss, Sowles and Bierut2015; Moreno et al., Reference Moreno, Briner, Williams, Brockman, Walker and Christakis2010, Reference Moreno, Kota, Schoohs and Whitehill2013).

It is quite clear from the literature that adolescents who post more alcohol-related content on social media tend to drink more (Roberson et al., Reference Roberson, McKinney, Walker and Coleman2018; Stoddard et al., Reference Stoddard, Bauermeister, Gordon-Messer, Johns and Zimmerman2012; Westgate & Holliday, Reference Westgate and Holliday2016). In a meta-analysis of 19 studies on alcohol-related social media use (that included posting, viewing, and liking others’ alcohol-related posts), alcohol-related social media use was moderately and significantly related to alcohol consumption and alcohol-related problems, with stronger associations emerging in cross-sectional and self-report (of alcohol-related social media use) studies compared to longitudinal and observational research (Curtis et al., Reference Curtis, Lookatch and Ramo2018). Indeed, posting about alcohol is associated with self-reported drinking frequency, heavy drinking, drinking quantity, and likelihood of alcohol use disorder (Glassman, Reference Glassman2012; Marczinski et al., Reference Marczinski, Hertzenberg, Goddard, Maloney, Stamates and O’Connor2016; Moreno & Whitehill, Reference Moreno and Whitehill2014).

Although far less studied, there is also some evidence that similar linkages may be at play for other substances as well. For tobacco, adolescents who posted positive tobacco-related content on Twitter were more likely to report past month cigarette and any tobacco use relative to those who did not post about tobacco on Twitter (Unger et al., Reference Unger, Urman and Cruz2018), and although posting about tobacco use is much less common than alcohol use among Dutch emerging adults, cigarette-related social media posts are nonetheless associated with real-life cigarette use (Van Hoof et al., Reference Van Hoof, Bekkers and Van Vuuren2014). For marijuana, research in young adults suggests that they do indeed post cannabis-related images on Instagram (Cavazos-Rehg et al., Reference Cavazos-Rehg, Krauss, Sowles and Bierut2016) and that posting marijuana-related content to social media is associated with more pro-marijuana attitudes and actual marijuana use among racial-ethnic minority college students from low-income areas; however, no such associations emerged for alcohol depictions, alcohol attitudes, and alcohol use, which may suggest that these associations are most relevant when a behavior is illegal or less normative (Lauckner et al., Reference Lauckner, Desrosiers, Muilenburg, Killanin, Genter and Kershaw2019). Recent research suggests that marijuana-related posting is not uncommon even in adolescence, however, which underscores the necessity of more research in this domain. For instance, in Washington (a state where cannabis is legal for recreational use among adults over the age of 21), nearly a third of adolescents reported sharing marijuana-related content on social media, with about 11–13% sharing images or videos of people smoking marijuana and 24% sharing marijuana-related memes (Willoughby et al., Reference Willoughby, Hust, Li, Couto, Kang and Domgaard2020).

Nearly all of the above research has examined the role of alcohol- and drug-related posting to public (e.g., Twitter) or semi-public (e.g., Facebook, Instagram) platforms, but much less research has attended to the role of private communications (e.g., private direct messaging and text messages). However, the research that has examined private messaging suggests it plays a key role. One study found that about a quarter of late adolescents (in the summer after 12th grade) reported discussing substance use on public social media, whereas nearly half report doing so via private digital channels (George et al., Reference George, Ehrenreich, Burnell, Kurup, Vollet and Underwood2019). In our own work (Jensen et al., Reference Jensen, Hussong and Baik2018) college students in the USA and Korea have reported that they prefer private text messages to public-facing social networking sites to facilitate alcohol involvement, and private text messaging was more related than public social media to frequency of alcohol use and heavy episodic drinking. We have also shown that counts of alcohol-related words in sent and received private text messages are associated with higher odds of same-day drinking (Jensen & Hussong, Reference Jensen and Hussong2019). Longitudinal research suggests that these associations may be bidirectional, with those youth who had previously been using substances being more likely to evidence later public and private substance-related discussions, and public and private conversations predicting later increases in marijuana use (but not alcohol or tobacco use; George et al., Reference George, Ehrenreich, Burnell, Kurup, Vollet and Underwood2019). Taken together, these findings highlight the importance of future research that attends to how private digital communication channels may be uniquely indicative of substance use risk.

Exposure to Others’ Alcohol- and Drug-Related Posts and Substance Use Behavior

In addition to adolescents’ own posting behaviors being associated with substance use and misuse, so too is there a sizable body of evidence to suggest that adolescents’ peers’ posts also have the potential to impact their behavior. The majority of studies seem to support the hypothesis that exposure to others’ substance use online is related to pro-substance attitudes and actual substance use behavior (Cabrera-Nguyen et al., Reference Cabrera-Nguyen, Cavazos-Rehg, Krauss, Bierut and Moreno2016; Curtis et al., Reference Curtis, Lookatch and Ramo2018; Pegg et al., Reference Pegg, O’Donnell, Lala and Barber2018). Results from recent longitudinal designs are particularly informative. Even after controlling for developmental risk factors for initiation of alcohol use, exposure to peers’ alcohol-related social media content predicted an adolescent’s likelihood of drinking initiation one year later (Nesi et al., Reference Nesi, Rothenberg, Hussong and Jackson2017). Similarly, adolescent exposure to alcohol-related social media content predicted alcohol consumption six months after exposure after accounting for both the adolescent’s and their peers’ drinking habits (Boyle et al., Reference Boyle, LaBrie, Froidevaux and Witkovic2016). Some studies suggest that different types of exposures may be more influential and long-lasting: Adolescents who had more exposure to pictures (but not text) about friends partying or drinking in their social networks were more likely to increase or maintain their smoking levels over time (Huang, Unger, et al., Reference Huang, Soto, Fujimoto and Valente2014). This is consistent with findings that image-based alcohol-related content posted by college freshmen may be more related to substance use intentions down the road than purely text posts on social media (D’Angelo et al., Reference D’Angelo, Zhang, Eickhoff and Moreno2014). Among young adults in Norway, disclosure of and exposure to alcohol-related content online was tied to later alcohol use, though the strength and consistency of these associations were reduced once relevant covariates were accounted for (Erevik et al., Reference Erevik, Torsheim, Andreassen, Vedaa and Pallesen2017).

An innovative experiment confirms this pattern: Litt and Stock (Reference Litt and Stock2011) created two Facebook profiles, one that portrayed alcohol use as normal and a control that displayed no alcohol; after viewing one of the two profiles participants were assessed on willingness to use alcohol and alcohol attitudes. Participants who viewed the alcohol normative profile had higher levels of willingness to use alcohol, more favorable images of alcohol users, more positive attitudes toward alcohol, and lower perceived vulnerability to the consequences of alcohol use, suggesting that exposure affects attitudes concerning alcohol. Results from Roberson and colleagues (Reference Roberson, McKinney, Walker and Coleman2018) build on this idea – higher numbers of people who display drinking in an individual’s online network predict more pro-alcohol attitudes. Taken together, it does appear that exposure to substance use in adolescents’ online peer networks is associated with increased risk for substance use and misuse, and we thus turn next to potential explanatory mechanisms for this association.

Deviancy Training

Socialization takes many forms, and deviancy training is one mechanism of peer socialization (Dishion et al., Reference Dishion, Spracklen, Andrews and Patterson1996). The process often plays out with a youth discussing an antisocial topic, which is reinforced by the peer’s response (e.g., by laughter, encouragement, or more antisocial discussion; Piehler & Dishion, Reference Piehler and Dishion2007). One of the central difficulties of studying deviancy training in youth is the difficulty of capturing their interactions as they play out, and thus a promising direction for future research is the time-linked analysis of deviancy training in naturalistic peer-to-peer interactions via digital media. Digital communication offers an unprecedented window of opportunity to observe and understand how youth communicate and reinforce one another in their real interactions. Evidence gleaned from the content of youth text messages suggests that those youth whose antisocial text messages are reinforced by peers’ positive responses are more likely to see increases in their problem behavior over time. A study of adolescents’ text message exchanges noted that antisocial comments in text are often met with laughter (e.g., “lol” and “haha”) from their conversational partners, which is similar to the deviancy training observed in past face-to-face observational research (Ehrenreich et al., Reference Ehrenreich, Underwood and Ackerman2014). Furthermore, these antisocial conversations were associated with increases in rule-breaking behavior a year later.

Some social networking sites include features that can serve to amplify the ability of peers to positively reinforce youth behavior. The Facebook Influence Model (Moreno et al., Reference Moreno, Kota, Schoohs and Whitehill2013) posits that peer influence is amplified within the online social networking environment, which in turn shapes downstream cognitions and behaviors around risk. Whereas the seminal studies on deviancy training in face-to-face interactions pinpointed communication features like laughing or encouragement as powerful (albeit minimal) reinforcers of deviant talk, Facebook and Instagram allow youth to send the same message with the click of a “like” or a “❤”. In fact, research suggests that the “like” is a powerful reinforcer (Sherman et al., Reference Sherman, Payton, Hernandez, Greenfield and Dapretto2016).

Social Norms

Selection and socialization processes on social media can alter perceptions of peer norms over time (David et al., Reference David, Cappella and Fishbein2006). Descriptive norms capture perceptions of how many of or how often peers engage in the relevant behavior (e.g., substance use, delinquency) and injunctive norms capture perceptions of how much peers approve of the behavior; both are strongly linked to adolescent behavior (Rimal & Real, Reference Rimal and Real2005). Super Peer Theory (Strasburger et al., Reference Strasburger, Wilson and Jordan2013) asserts that media can serve as a “super peer” in that it can expose teens to information that makes risk-taking behaviors seem normative, and that this normative influence will in turn cause youth to take risks themselves.

Research is generally supportive of the thesis that exposure to risky content online operates by reshaping youth perceptions of normativity. Qualitative studies with adolescents (Moreno et al., Reference Moreno, Briner, Williams, Walker and Christakis2009) and college students (Moreno et al., Reference Moreno, Grant, Kacvinsky, Egan and Fleming2012) tend to suggest that peers’ references to alcohol use on social media are indicative of their actual alcohol use behaviors offline, with younger youth perhaps being most susceptible to the impact of online depictions on normative beliefs. Our research suggests that the amount of “alcohol talk” in received (but not sent) text messages from college students’ entire text messaging network over the course of two weeks is associated with greater perceptions of peer descriptive and injunctive substance use norms, in addition to sent and received alcohol talk being tied to frequency of heavy episodic drinking (Jensen & Hussong, Reference Jensen and Hussong2019). A longitudinal study of adolescents showed the exposure to sexual content in media increased youth perceptions of normative pressure (which captured both injunctive and descriptive norms), which in turn increased sexual activity intentions and behavior (Bleakley et al., Reference Bleakley, Hennessy, Fishbein and Jordan2011). This is highly consistent with experimental evidence that exposure to sexually suggestive photos impacts adolescents’ perception that more of their peers engage in sexual risk taking (S. D. Young & Jordan, Reference Young and Jordan2013) and that college students who viewed a social networking site with alcohol-related content estimated that the average college student drinks more frequently than participants who did not view the alcohol-related content (Fournier et al., Reference Fournier, Hall, Ricke and Storey2013).

Status

Adolescents have been known to engage in certain types of problem behaviors (e.g., carrying a weapon, substance use, physical aggression) in service of gaining the status that these behaviors confer (Dijkstra et al., Reference Dijkstra, Lindenberg and Veenstra2010; Osgood et al., Reference Osgood, Ragan, Wallace, Gest, Feinberg and Moody2013; Rulison et al., Reference Rulison, Gest and Loken2013). Nesi and colleagues (Reference Nesi, Choukas-Bradley and Prinstein2018b) assert that some features of social media (e.g., its publicness and widespread availability) may amplify youths’ quest for status through online spaces through selective self-presentation. Although there have been relatively few studies to date that explicitly test the role of status striving as a driver of youth externalizing and risk-taking behavior, some new research suggests that some adolescents are (and are known by peers for) engaging in “digital status seeking” behaviors (behaviors intended to increase “likes” and approval) online, and that these digital status seeking behaviors are longitudinally tied to later increases in substance use and sexual risk behavior (Nesi & Prinstein, Reference Nesi and Prinstein2019). Indeed, the Internet’s culture of “micro-celebrity” may facilitate the extent to which high-status “peers” can impact norms and exert influence (Marwick & boyd, Reference Marwick and boyd2011).

We are beginning to see the role of status in peer influence across the externalizing and risk-taking spectrum. For instance, partying is considered by many teens as a high-status activity, and attendance (and subsequent publishing online) of images and text about parties may boost status by association (Marwick & boyd, Reference Marwick and boyd2011; Nesi et al., Reference Nesi, Choukas-Bradley and Prinstein2018b). Students in a rural high school in the United States tended to drastically overestimate how many of their popular peers were sexting (and those who believed that popular peers had sexted were more likely to have sexted themselves than those who did not hold that perceived norm; Maheux et al., Reference Maheux, Evans, Widman, Nesi, Prinstein and Choukas-Bradley2020). As reviewed earlier, digitalist gangs are also capitalizing on the attention and status that social media can afford (Whittaker et al., Reference Whittaker, Densley and Moser2020). There is even some evidence that being a perpetrator of cyberbullying is predictive of increased peer status over time (Wegge et al., Reference Wegge, Vandebosch, Eggermont and Pabian2016).

Interestingly, youths’ search for status and desire to be perceived positively could also exert a “chilling effect” wherein adolescents may self-censor their real-life behaviors to avoid unfavorable exposure on social media (Marder et al., Reference Marder, Joinson, Shankar and Houghton2016). A mixed-methods study of the chilling effect revealed that teens do engage in impression management around depictions of substance use (e.g., hiding their drink/cigarette when they know a photo will be taken and likely end up online, presumably to avoid potential consequences if it is seen by a parent) but that they rarely alter their actual substance use behaviors (e.g., choosing not to drink or smoke at the party in the first place; Marder et al., Reference Marder, Joinson, Shankar and Houghton2016). Further research on impression management, status seeking, and behavior change will certainly better elucidate the nature of these associations in the years to come.

Unique Online Influences?

As reviewed here, online peer influence does seem to be a predictor of youth externalizing and health risk behaviors. An important question, though, is whether online peers exert unique influence, over and above that which would be expected (or is seen) from real-life, face-to-face peers (i.e., from school or neighborhood). Recent studies have tested this hypothesis, and overall, it seems that, although peers (in general) are still highly influential, there is significant overlap between online and offline networks, and online-only peer relationships seem to exert none to small effects. For instance, McCuddy (Reference McCuddy2021) sought to parse influence by adolescents’ peers who are known in person (and also sometimes online) from those peers who are uniquely known online (and not in person). They uncovered little evidence that online peers expose adolescents to new/unique support for delinquency (e.g., only 7% of those exposed to any general delinquency in a peer network saw this influence from online-only peers, whereas 64% were exposed to both online and offline peer delinquency). Rates were similar for violence (8% exposed only via online peers) and slightly higher for theft (17%) and substance use (21%). Exposure to online peer support for general delinquency and violence were not associated with adolescent problem behaviors in these domains, though online peers appeared slightly more influential for theft and substance use behaviors. In all cases, online peer influence was of lesser magnitude than traditional (face-to-face) peer influence. Another study has similarly failed to find support for unique influence by online-only friends on marijuana use (Negriff, Reference Negriff2019).