Exploring Text Messaging as a Platform for Peer Socialization of Social Aggression

Abstract

This study examined whether adolescents’ social aggression is socialized through exposure to peers’ socially aggressive text messaging. Using data on the socially aggressive content of text messages that 221 participants (M_age = 15.02; 46.7% female) sent to and received from peers, and teacher ratings of participants’ in-person social aggression, this study found that exposure to peers’ socially aggressive texting about out-dyad peers predicted positive changes in adolescents’ text-based and in-person social aggression. Gender differences were examined, and results were mixed. In ninth grade, girls sent more socially aggressive text messages than boys; however, by 10^th grade these differences disappeared. Gender differences in adolescents’ in-person social aggression and their exposure to peers’ socially aggressive texting were nonsignificant at both time points. There was no evidence of gender differences in the links between exposure to peers’ socially aggressive texting and adolescents’ socially aggressive texting. However, marginal differences were found in the associations between exposure to peers’ socially aggressive texting and adolescents’ in-person social aggression. Results suggest that texting provides an additional platform for peer socialization of adolescents’ social aggression.

Contemporary adolescents engage in social aggression with peers in both digital and in-person contexts (Coyne, Ehrenreich, Holmgren, & Underwood, 2019; Underwood, 2003; Underwood & Ehrenreich, 2016), and the extent to which they are exposed to peers’ socially aggressive digital communication may foster the development of their own social aggression. This study examined exposure to peers’ social aggression by capturing the content of every text message that adolescents sent to and received from peers in two 2-day periods. Studying text messaging allowed for detailed measurement of exposure to peers’ social aggression across interaction partners in a manner that was naturalistic and ecologically valid, and in a way that would be extremely difficult for offline social aggression because social aggression is subtle and adolescents are sophisticated in hiding negative behaviors from adults.

Social aggression includes interpersonal behaviors such as social exclusion, friendship manipulation, and malicious gossip that are enacted directly or indirectly with the intent to cause harm to an individual’s social status or friendships (Cairns, Cairns, Neckerman, Gest, & Gariepy, 1988; Dodge, Coie, & Lynam, 2006; Underwood, 2003). Social aggression is similar to indirect aggression (Lagerspetz, Björkqvist, & Peltonen, 1988) and relational aggression (Crick & Grotpeter, 1995), in that these constructs also include social exclusion, malicious gossip, and friendship manipulation (Archer & Coyne, 2005). However, in addition to behaviors included in both indirect and relational aggression, social aggression also includes nonverbal forms of social exclusion and direct forms of relationship harm (Underwood, 2003). Nevertheless, because of similarities between these constructs, our literature review includes findings from all three bodies of work and carefully notes the construct used by the original investigators.

Engaging in social and relational aggression can have pernicious effects on adolescents’ well-being that are both immediate and lasting. In the short term, although youth who engage in social and relational aggression are typically more popular and central in peer networks, they also tend to be liked less by their peers (i.e., lower social preference; Cillessen & Mayeux, 2004; Neal, 2010; Xie, Cairns, & Cairns, 2002). Over time, youth who engage in social and relational aggression are more prone than non-aggressive youth to develop externalizing problems, such as rule breaking and substance use (Card et al., 2008; Ehrenreich, Beron, & Underwood, 2016), and are more likely to experience internalizing problems, such as anxiety and depression (Crick & Grotpeter, 1995; Crick, Ostrov, & Werner, 2006; Ehrenreich et al., 2016).

As potential costs of engaging in social aggression have become clearer, research has increasingly focused on understanding how youth become socially aggressive. There may be some genetic basis for social and relational aggression (Brendgen et al., 2008; Lundwall, Sgro, & Wade, 2017; Tackett, Kushner, Herzhoff, Smack, & Reardon, 2014), but adolescents’ relationships and interactions with others—notably, their peers—likely also play a central role. Research on peers and social aggression has primarily focused on peer relationship qualities (e.g., friendship exclusivity and intimacy [Grotpeter & Crick, 1996]; friendship jealousy [Kraft & Mayeux, 2018; Parker, Low, Walker, & Gamm, 2005]). A smaller body of work has examined how social aggression is shaped through social interactions with peers, providing new insights into whether and how peers socialize social and relational aggression (Brendgen et al., 2008; Ellis & Zarbatany, 2007; Shi & Xie, 2012, 2014; Werner & Crick, 2004; Werner & Hill, 2010). However, important questions remain.

First, few of these studies have examined gender differences in peer socialization of social or relational forms of aggression despite persistent interest in understanding the extent to which boys and girls differ in social and relational aggression. Research on gender differences in the perpetration of social and relational aggression is abundant, but findings are inconsistent. Early research suggested that girls are more socially, relationally, and indirectly aggressive than boys (Crick, 1997; Lagerspetz et al., 1988), but more recent evidence has called these differences into question (Salmivalli & Kaukiainen, 2004; Underwood, Scott, Galperin, Bjornstad, & Sexton, 2004; see Card et al., 2008, for a meta-analysis). The few studies that have examined gender differences in peer socialization of social and relational aggression have produced similarly conflicting findings. In an early study, Werner and Crick (2004) examined whether having relationally aggressive friends was associated with increases in the relational aggression of 979 elementary school-age children across a single year. Friends’ relational aggression predicted positive changes in children’s own relational aggression, but only among girls. More recently, Shi and Xie (2012) examined peer group influences on the social aggression of 321 adolescents and found that youth who affiliated with socially aggressive peers tended to increase in social aggression across a single year. No differences were found between all-boy and all-girl groups in peer effects on social aggression. Conflicting results from these two studies might be due to differences in the age groups they studied. Regardless, research that continues to explore gender differences in peer socialization of social and relational forms of aggression is needed.

Second, previous work on peer socialization of social aggression has focused on peer interactions that presumably occurred in in-person, face-to-face contexts. Considering well-documented increases in adolescents’ use of mobile technologies for interacting with peers (Anderson & Jiang, 2018; Lenhart, 2012), it has become increasingly important that research on peer socialization expand its consideration of adolescents’ peer interactions to include those facilitated by these technologies. Digital communication technologies expand adolescents’ opportunities to interact with peers, and thus may potentially amplify peer socialization of social aggression (Nesi, Choukas-Bradley, & Prinstein, 2018a, 2018b). By not accounting for the frequent interactions that adolescents have with peers via digital communication technologies, research may be missing a potentially important sphere of peer influence. Ultimately, to test this “amplification” hypothesis, research will need to examine whether exposure to peers’ socially aggressive texting contributes to the development of adolescents’ social aggression over and above the effect of in-person exposure to social aggression. This will require data capturing adolescents’ exposure to peers social aggression in both digital and in-person contexts (the current study only captured data on the former). Observing exposure to social aggression across in-person settings for a sustained period of time would be challenging because social aggression can be subtle, some of the behaviors are non-verbal, and adolescents would likely be reactive to live observers. Nevertheless, it is an important first step to examine whether exposure to peers’ socially aggressive text messaging, by itself, predicts changes in adolescents’ social aggression. Toward this goal, this study examines the extent to which text messaging provides a platform for peer socialization, by examining how adolescents’ exposure to peers’ socially aggressive text messaging predicts changes in their own socially aggressive behaviors.

Peer Socialization of Social Aggression

Adolescents spend much of their time interacting with peers (Hartup, 1993). Notably, unlike other social partners, including teachers, parents, or siblings, adolescents choose the peers with whom they interact. Their preference in selecting peers is typically guided, in part, by existing similarities (Aboud & Mendelson, 1998; Cairns, Neckerman, & Cairns, 1989; Kandel, 1978), which may be based on any number of interests, attributes, and behaviors—including socially or relationally aggressive behavior (Cairns et al., 1988; Werner & Crick, 2004). By choosing similar peers, adolescents construct for themselves concentrated peer contexts that mirror their own behavioral dispositions (Dishion, Spracklen, Andrews, & Patterson, 1996; Kindermann & Gest, 2018). For socially aggressive adolescents, who tend to choose similarly aggressive peers (Cairns et al., 1988; Werner & Crick, 2004), peer selection processes might drive increases in their exposure to peer group members’ social aggression over time, as their group becomes more homogenously aggressive. In turn, ongoing interactions with similarly aggressive peers might provide opportunities for peer socialization that amplify existing behaviors (Brendgen et al., 2008; Ellis & Zarbatany, 2007; Shi & Xie, 2012, 2014; Werner & Hill, 2010). The processes of peer socialization of social aggression might mirror those observed in deviancy training (Dishion et al., 1996) and likely operate through several of the same mechanisms, including group norms, reinforcement, and modeling.

Peer Group Norms.

Social norms perspectives suggest that adolescents engage in behaviors that are common among members of valued or desired peer groups in order to gain or retain acceptance in that group (Cialdini & Trost, 1998). Whereas members of peer groups in which socially aggressive behaviors are common might feel encouraged or compelled to engage in similar behaviors, social aggression might be socially censured in groups where such behaviors are infrequent (Boivin, Dodge, & Coie, 1995). A study of 2,731 children found that students in classrooms in which relational aggression was normative, as indicated by high average classroom levels of relational aggression, showed increased relational aggression over the course of a year (Kuppens, Grietens, Onghena, Michiels, & Subramanian, 2008). In a separate study, peer norms for relational aggression were examined as predictors of changes in social aggression among 245 students in Grades 3 through 8 (Werner & Hill, 2010). Results indicated that peer norms for relational aggression predicted increases in students’ relational aggression over time. Increases were steeper among students nearing the end of middle school than among younger elementary school students.

Peer Reinforcement.

Peer socialization of social aggression might also operate through reinforcement (Cillessen & Mayeux, 2004; Eder & Enke, 1991; Werner & Crick, 2004), wherein peers reward adolescents for their expressions of social aggression. Peers’ reinforcement of socially aggressive behavior can be immediate and involve amusement, agreement, or praise. For example, in an observational study of lunch table conversations among middle school students, Eder and Enke (1991) found that when peers immediately reinforced initial negative evaluation comments, by expressing agreement or support, the entire group joined in and a group session of malicious gossip ensued. However, when just one peer immediately followed the initial negative remark with a comment that was not reinforcing, no further gossip occurred.

Peer Modeling.

In addition to social norms and reinforcement, peer socialization may occur through social modeling (Bandura, 1971). Through exposure, adolescents may begin to mimic peers’ socially aggressive behavior, particularly when they witness their peers being socially rewarded for such behaviors. Social and relational aggression confer social benefits (Banny, Heilbron, Ames, & Prinstein, 2011; Cillessen & Mayeux, 2004; Dumas, Davis, & Ellis, 2019) that are highly salient to members of the larger peer group. For example, Cillessen and Mayeux (2004) found that relationally aggressive children typically held higher status (i.e., perceived popularity) among their peers. These links were found to strengthen with age. Through exposure to peers’ social aggression, youth may come to associate these individuals’ high social standing with their socially aggressive behavior and, via vicarious reinforcement, they may begin to model such behavior, implicitly or explicitly, motivated by the goal of improving their own status among peers.

This study builds on the premises that (a) peers are selected by adolescents based on similarities and, in turn, contribute to shaping adolescents’ social aggression through a variety of mechanisms of socialization (i.e., group norms, reinforcement, and modeling) that occur via frequent interaction, and (b) adolescents’ peer interactions have become increasingly facilitated by digital communications technologies. Guided by these perspectives, this study examined whether frequency of exposure to peers’ socially aggressive text messages about others was a positive predictor of changes in adolescents’ own social aggression, both in their text messaging and face-to-face interactions with peers.

Digital communication and peer socialization of social aggression

In 2009, B. Bradford Brown and James Larson insightfully noted that “the common assumption is that, for the most part, adolescent peer relations are carried out through face-to-face interactions in various physical contexts in the community. Although this is still likely to be true, another context is emerging as a major locus of peer interaction, namely, the world of electronic media. (Brown & Larson, 2009, p. 98)” A decade later, research on peer socialization has still not yet fully considered how adolescents’ peer interactions have been impacted by digital communication technologies, such as text messaging, despite significant growth over the past two decades in their use for communicating with peers. Current representative surveys of U.S. youth show that nearly all adolescents (95%; ages 13–17) own or have access to a mobile phone (Anderson & Jiang, 2018), and text messaging remains one of the principle ways that they interact with one another using these devices (Bailey, Schroeder, Whitmer, & Sims, 2016; Lenhart, 2015; Rideout & Robb, 2018; Underwood, Ehrenreich, More, Solis, & Brinkley, 2015). Although alternative platforms for digital communication have gained popularity among youth over the past decade (since data for this study were collected; e.g., Facebook Messenger and WhatsApp), recent evidence suggests that adolescents still prefer text messaging over social media and video chat as a method of communicating with peers (Rideout & Robb, 2018). Furthermore, there is evidence that youth use the messaging features on these alternative platforms in ways that are similar to SMS texting (Bailey et al., 2016). These findings suggest that the substance of youths’ digital communications with peers has not likely changed as their use of these alternative communications apps has increased.

Given the ubiquity of their use, digital communications technologies may now represent an important, additional conduit for socialization that may enhance the power youth have to influence each other’s socially aggressive behavior (Nesi et al., 2018a, 2018b). Text messaging and related mobile communications technologies allow for greater connection between youth and their peers by circumventing barriers (e.g., geographical distance) that may otherwise impede face-to-face peer interactions. For example, although much of adolescents’ interactions with peers occur in person while at school, they are often separated physically from their peers when away from school (e.g., if they live in different neighborhoods), thus hampering their ability to interact. Text messaging extends youths’ ability to stay connected with their broader group of peers from virtually anywhere and at any time, allowing them to be “together” even when they are physically apart. The level of connectedness made possible by these technologies offers youth the ability to interact with peers constantly, with a level of immediacy comparable with in person interactions. Constancy and immediacy are two features that, when combined, may serve to amplify the effects of peer influence (Nesi et al., 2018a, 2018b) through a variety of mechanisms. By using these technologies to maintain constant and immediate contact with peers, adolescents may be exposed to more of their peers’ socially aggressive behavior, and might more frequently receive immediate praise of their own expressions of social aggression, potentially boosting the impact of peer modeling, group norms, and reinforcement. Although there are functional differences between in person and text-based peer interactions, theory suggests that the effect that exposure to peers’ socially aggressive texting has on adolescents’ own socially aggressive behavior may parallel the effects of their exposure to peers’ social aggression in face-to-face contexts.

Co-construction theory.

Both theory and empirical evidence suggest that adolescents’ digital and non-digital environments are intricately intertwined (Mazur & Kozarian, 2010; Subrahmanyam, Smahel, & Greenfield, 2006). According to co-construction theory (Subrahmanyam et al., 2006), youth are not passive to the effects of interacting in digital environments. Rather, they actively shape, or co-construct, the nature of these environments through continued interactions with others in these environments, most notably their peers. Central to the present study, co-construction theory suggests further that adolescents’ digital worlds are tightly linked to their offline worlds (Subrahmanyam & Greenfield, 2008; Subrahmanyam, Reich, Waechter, & Espinoza, 2008; Subrahmanyam et al., 2006). For example, teenagers may refer to events that occurred in their face-to-face peer interactions while texting with a peer to exclude, gossip about, or express their general disdain for another peer. At the same time, youth may also draw from or reference occurrences from their digital peer interactions while engaging in social aggression in face-to-face environments. Given how intermingled adolescents’ digital and non-digital worlds likely are, the impact of digital and face-to-face peer interactions likely play out in similar ways, involve similar mechanisms (e.g., modeling), and their effects likely cut across digital and non-digital contexts.

The current study

The central aim of the present study was to investigate whether text messaging provides a facilitative platform for peer socialization of adolescents’ social aggression, both in person and in digital environments. In line with prior research that found evidence of peer socialization of social aggression through face-to-face interaction (Brendgen et al., 2008; Ellis & Zarbatany, 2007; Shi & Xie, 2012, 2014; Werner & Hill, 2010), we expected to find evidence that adolescents’ social aggression is shaped similarly through exposure to peers’ socially aggressive text messaging about others. First, we expected to find evidence of peer socialization of social aggression such that adolescents’ exposure to peers’ socially aggressive texting about others would predict increases in their own socially aggressive texting from ninth to tenth grade. Second, following co-construction theory (Subrahmanyam et al., 2006), we also expected that adolescents’ exposure to peers’ socially aggressive texting would predict positive changes in their in-person social aggression, as observed and rated by teachers, suggesting that the socializing effect of peers’ socially aggressive text messaging behavior extends to adolescents’ social aggression in face-to-face contexts as well.

This study also aimed to examine the role of peer selection based on social aggression by (a) exploring the concurrent links between adolescents’ social aggression and their exposure to peers’ social aggression, and (b) examining whether adolescents’ initial levels of social aggression predicted changes in their exposure to peers’ socially aggressive texting. Based on adolescents’ well-documented tendency to select similar peers (Aboud & Mendelson, 1998; Cairns et al., 1989; Kandel, 1978), we expected to find positive concurrent links between adolescents’ social aggression and their exposure to peers’ social aggression. Furthermore, we anticipated that adolescents’ social aggression in Grade 9 (both in person and via text messaging) would predict increases in their exposure to peers’ socially aggressive text messaging from ninth to 10^th grade.

Finally, this study sought to explore potential gender differences in peer socialization of social aggression, as facilitated by text messaging. Given the mixed results from prior work on gender differences in peer socialization of social aggression in face-to-face settings (Shi & Xie, 2012; Werner & Crick, 2004), we had no a priori expectations regarding gender differences in the role that exposure to peers’ socially aggressive text messaging plays in shaping adolescents’ in-person and text-based social aggression. Thus, these analyses were exploratory in nature.

Methods

Participants

Participants (n = 297) were initially recruited from a suburban county in the southwestern United States at the end of the third grade for a longitudinal study investigating social aggression, physical aggression, and peer relations. Before the beginning of their ninth-grade year, participants were invited to take part in a follow-up study, the Blackberry Project, which examined digital communication during high school. At this point, they were provided smartphones that were programmed to archive the content of all incoming and outgoing text messages.

The current study examined data collected during the 2008 to 2009 and 2009 to 2010 academic years, when participants were in their ninth- and 10^th-grade years. This span of time was selected for the present study for two reasons. First, prior self-reported rates of adolescent texting have indicated that text messaging frequency begins to peak during this period (Coyne, Padilla-Walker, & Holmgren, 2018). Second, this period coincided with the transition to high school, a time during which adolescents’ social aggression peaks (Karriker-Jaffe, Foshee, Ennett, & Suchindran, 2008), perhaps as they seek to reestablish their positions in a new peer hierarchy (Cillessen & Mayeux, 2004; Kinney, 1993).

Participants were included if data on their teacher-reported levels of social aggression and socially aggressive texting were available for at least one of the two of the time points. Participants who met these criteria included 221 adolescents (46.7% female, Grade 9 M_age = 15.02, SD_age = 0.51). Of these, 185 (84%) participated in Grade 9, and 207 (94%) participated in Grade 10 (172 participated at both time points). These 221 participants did not differ from the non-participating portion of the original sample (n = 76) in terms of race, χ² (4) = 8.22, p = .08, gender, χ² (1) = 2.86, p = .09, or teacher-rated social aggression, t (198) = .734, p = .46. Income differences between these groups of participants were found, χ2 (4) = 12.05, p = .02, such that fewer participants in the lowest income bracket were included than should be expected, and more than expected were in the highest income bracket. The final study sample was racially diverse (52.3% Caucasian, 21.5% African-American, 18.7% Hispanic or Latino, 4.6% Biracial and 1.4% Asian), and, at the time of the study, reflected the demographic composition of the region from which the participants were sampled (U.S. Census Bureau, 2010). Furthermore, participants in the present study were from economically diverse families, as indicated by contemporaneous parent reports of annual household incomes, with 11% reporting less than $25,000, 17% reporting $26,000 – $50,000, 15.9% reporting $51,000 – $75,000, 16.8% reporting $76,000 – $100,000, and 24% reporting $100,000 or more. Fifteen percent did not report annual income.

Procedure

Participants were given smartphones prior to beginning the ninth grade. These devices had data plans and unlimited texting paid for by the researchers, and were replaced each year or when lost or broken. All text messages sent to and received by participants every day over four subsequent years were stored using a secure server, which was maintained by Ceryx and archived by Global Relay. Daily digests of each participant’s text messages were provided by Global Relay; these included content from every text message sent and received by a participant over a 24-hour period, as well as the date, time, and phone number of the sender and the receiver for each text. Contact information for each communication partner was also stored. It is important to note that participants were not restricted to using the smartphones provided solely for text messaging. A full description of the methodologies used to capture and archive participants’ text messages can be found in Underwood, Rosen, More, Ehrenreich, and Gentsch (2012).

The Blackberry Project has the ongoing approval of the institutional review board at the University of Texas at Dallas (IRB protocol number: 07–36). In adherence to ethical principles, efforts were made to ensure that all participants and their parents understood and consented to the adolescents’ devices being monitored prior to receiving the phones. The archive was routinely monitored for language that indicated possible harm to the self or to others, and the principal investigator, a clinical psychologist, intervened when necessary. Due to the potential for text messages between participants and others to involve antisocial, rule-breaking, and drug-related behavior, a Federal Certificate of Confidentiality was obtained from the National Institutes of Health. This certificate served as protection from having to report such behavior to authorities. A full description of ethical considerations for this project can be found in Underwood et al. (2012).

There was evidence that observations of participants’ text messages reflected their true text messaging habits, an important consideration when examining constructs susceptible to social desirability bias such as social aggression. Most participants reported using their smartphones almost all the time (Underwood et al., 2012). Furthermore, the rates of profanity and sexual talk observed in our data are comparable with the rates seen in unmonitored online chat rooms (Subrahmanyam et al., 2006), suggesting that they did not adjust their text messaging behavior due to being monitored (Underwood et al., 2012).

Measures

Demographics.

Parents provided information on adolescents’ gender and race, as well as their family income (as a marker of socio-economic-status). For analytic purposes, dummy coding was used to quantify gender (Male = 0, Female = 1) and race (“White” was used as the reference group). Demographic information was collected at the end of participants’ ninth-grade year.

In-person social aggression.

Teachers who knew the participants best, as identified each year by participants and their parents, provided ratings of adolescents’ social aggression near the end of their ninth- and tenth- grade years by completing a modified version of the Children’s Social Behavior Scale – Teacher Form (CSBS-T; Crick, 1996). In most cases, the teachers reporting on the students’ aggression differed from Grade 9 to Grade 10. The CSBS-T was modified by adding social aggression items for gossip and nonverbal social exclusion. Four items were used to capture social aggression (e.g., “This student gossips or spreads rumors about people to make other kids not like them”). All items were rated on a Likert-type scale, ranging from 1 (“This is never true of this student”) to 5 (“This is almost always true of this student”). In the present study, these items showed high internal consistency, both in Grade 9 (α = .84) and Grade 10 (α = .86).

Teacher-reported measures were preferred to self-reported measures for two reasons. First, self-reports can be prone to self-enhancement biases, in which participants report themselves as being less aggressive than they objectively are (Underwood, Ehrenreich, & Meter, 2018). Social aggression often involves subtle behaviors that youth engage in covertly. Precisely because youth do not want these behaviors to be known about, it may be unreasonable to expect that their reports will be truthful and accurate. Second, teacher-reported measures were used because they allowed us to more clearly distinguish in-person social aggression from socially aggressive texting. Teachers’ reports were likely based on what they observed in and between classes rather than their students’ virtual communications, so these measures of social aggression were interpreted as reflecting social aggression occurring in face-to-face interactions. By contrast, self-reports of social aggression may have captured an amalgam of in-person and text-based social aggression, thus conflating the two.

Text messaging communication with peers.

Text-messaging data captured over 4 days during both ninth and 10^th grade were microcoded for content. For both years, two days were coded near the end of fall, during Homecoming week, and two days were coded in winter, during Valentine’s Day week. These two 2-day periods were chosen for coding because an increase in text-based social interaction (both positive and negative) was expected to coincide with the social activities associated with Homecoming and Valentine’s Day (e.g., school-sponsored dances). Alternative dates were used for participants for whom archived communication during these defined periods could not be captured, either because they did not use their phone or because it malfunctioned. These alternative dates were chosen by incrementally expanding the bounds of the search up to a week before and after the given 2-day periods.

Once the target days of communication were selected, transcripts of all the text messaging communication that occurred during these days were formatted for coding and were then randomly distributed to a team of 24 trained coders. Twenty percent of the transcripts that were coded by this team were also coded for reliability by an independent coder. All text messaging communications (including both texts sent from the participant and received by the participant) were coded. Coders read through the transcript and coded each text to (a) identify who the participant was sending text messages to and receiving text messages from, (b) determine how many utterances (i.e. complete thoughts) were contained within each text, and (c) capture the agent, target, and content of each utterance. To identify text messaging partners, coders used information gleaned either from the contact name (when present in the contact list on participants’ phones) or from the context of the communication. For the purposes of this study, only communication with individuals identified as peers (κ = .87) was examined. It should be noted that most peers with whom participants texted were not participants in the study. To establish the number of utterances in each text, coders noted how many complete thoughts were contained within each message. Finally, the content of each utterance was coded for content that denoted social aggression as well as the agent (i.e., the aggressor) and target (i.e., the aggressed) of the utterance.

Socially aggressive text messaging.

Socially aggressive texts included those that contained utterances that expressed negative talk (i.e., slanderous name calling or rumor spreading; e.g., “he’s an effing douche”; κ = .65 to .75), exclusion (e.g., “okay awesome. i invited all of them except matt”; κ = .63 to .83), or friendship manipulation (“Only reason I tlk to her is to get the homework”; κ = .66 to .88) targeting other, out-dyad peers. Instances of socially aggressive texts received by the participant in which they were the target were not counted to avoid capturing texts that reflected being the victim of peers’ social aggression as opposed to witnessing it. Likewise, texts sent by the participant in which the peer recipient was the target were not included because they did not capture social aggression—where the intent is to damage an individual’s relationships or social status among others.

Negative talk comprised a vast majority of the socially aggressive texts that adolescents sent to peers in Grade 9 and Grade 10 (95% and 98%, respectively). Socially aggressive texts received from peers were similar in composition (94% negative talk in Grade 9, and 98% in Grade 10). Due to the relatively low frequencies of exclusions and friendship manipulations, analyzing these forms of social aggression separately was not possible. Thus, data on these three modes of social aggression were combined into one single count variable capturing socially aggressive texting. Target adolescents’ socially aggressive texting was operationalized as the number of socially aggressive texts that were sent by the target adolescent to a peer. Similarly, exposure to peers’ socially aggressive texting was captured as the number of socially aggressive texts received by the target adolescent from peers.

Texting network characteristics.

Additional features of participants’ peer texting networks were identified. Peer texting network size (PTNS) denoted the number of peers with whom adolescents sent or received text messages during the 4 observed days. Peer texting network aggressiveness (PTNA) was calculated as the percentage of adolescents’ networks that had sent at least one socially aggressive text. Considering both of these features, by including them as covariates, allowed us to account for the focused or diffuse nature of adolescents’ exposure to peers socially aggressive text messaging. For example, an adolescent who only receives socially aggressive text messages from one of the six peers with whom she interacts may be affected differently by this exposure than if she had been exposed to the same number of socially aggressive texts but received some from each one of her peers. Finally, peer network texts (PNTs) captured the total number of texts received by peers, and was used as a control to account for the effects of the use of texting technology itself.

Analysis Plan

Two cross-lagged panel models examined the extent to which peers socialize adolescents’ in-person and text-based social aggression via text messaging interactions (Figure 1). The first model (i.e., Model 1) explored whether adolescents’ socially aggressive texting behavior was socialized through exposure to peers’ socially aggressive texting behavior. In this model, exposure to peers’ socially aggressive texting in ninth grade was used as a predictor of adolescents’ socially aggressive texting a year later in 10^th grade, controlling for their prior socially aggressive texting in ninth grade (and other control variables that are identified at the end of this section). This model also examined whether adolescents own aggressive texting behavior in ninth grade predicted changes in their peers’ socially aggressive texting from Grade 9 to Grade 10. The second model (i.e., Model 2) examined whether adolescents’ in-person social aggression was also socialized through exposure to their peers’ socially aggressive text messages. This model examined simultaneously whether peers’ socially aggressive texting in Grade 9 predicted changes in target adolescents’ teacher-rated, in-person social aggression from 9^th to 10^th grade, and whether adolescents’ in-person social aggression in Grade 9 predicted changes in peers’ socially aggressive texting from 9^th to 10^th grade. In both models, adolescents’ gender and race were included as covariates to account for differences in social aggression based on these variables. Peer texting network characteristics were also included as covariates, to further isolate the effect of the aggressiveness of peers’ texting from the more general features of their texting behaviors. All models were tested using AMOS 24 (Arbuckle, 2016).

Figure 1.

Models Testing Peer Socialization of Adolescents’ In-Person and Text Messaging Social Aggression

Results

Descriptive Statistics and Correlations

Descriptive statistics and correlations are shown in Table 1. In all analyses, a full information maximum likelihood method was used to account for missing data. On average, participants were low in teacher-rated social aggression in Grade 9 (M = 1.47, SD = .63) and Grade 10 (M = 1.42, SD = .63; r = .18, p = .04). Participants sent an average of 217 texts to peers over 4 days in Grade 9, of which an average of five were socially aggressive. This amount increased to approximately 373 texts sent to peers over 4 days in Grade 10, of which an average of six were socially aggressive. Exposure to peers’ socially aggressive texting behavior was also relatively minimal. On average, participants received about 185 texts from peers over 4 days in Grade 9, of which an average of four were socially aggressive. This also increased to approximately 360 texts received from peers over 4 days in Grade 10, of which an average of five were socially aggressive. High concurrent correlations were found between participants’ own socially aggressive texting and their peers’ socially aggressive texting both in Grade 9 (r = .78, p < .001) and Grade 10 (r = .74, p < .001). Links between participants’ in-person social aggression and their peers’ socially aggressive texting were significant in Grade 9 (r = .17, p = .04) and marginal in Grade 10 (r = .15, p = .06). Furthermore, a significant time-lagged correlation was found between participants’ exposure to peers’ socially aggressive texts in Grade 9 and their own socially aggressive texts in Grade 10 (r = .37, p < .001). A marginal time-lagged association was found between participants’ exposure to peers’ socially aggressive texts in Grade 9 and their own in-person social aggression in Grade 10 (r = .16, p = .10).

Table 1.

Correlations and Descriptive Statistics

	1	2	3	4	5	6	7	8	9	10	11	12	M	SD
1. In-person SA Grade 9													1.47	.64
2. In-person SA Grade 10	.18*												1.42	.63
3. SA Texting Grade 9	.19*	.13											5.20	8.23
4. SA Texting Grade 10	.13	.11	.36***										6.19	7.13
5. Peers’ SA Texting Grade 9	.17*	.16†	.78***	.37***									4.15	6.24
6. Peers’ SA Texting Grade 10	.17*	.15†	.24**	.74***	.31***								5.33	5.90
7. PTNS Grade 9	.20*	.13	.45***	.16†	.44***	.11							11.05	8.58
8. PNT Grade 9	.21*	.09	.56***	.39***	.58***	.37***	.66***						184.59	185.08
9. PTNA Grade 9	.05	−.01	.40***	.18*	.53***	.10	.04	.28***					.13	.15
10. Gender	−.04	.00	.18*	.09	.14†	.06	.20**	.12	.17*				.48	.50
11. Race: Hispanic	.05	.09	.00	.04	−.04	.08	.01	.02	−.09	.00			.18	.39
12. Race: Black	.10	.08	.04	.11	.08	.21**	.23**	.26**	−.07	−.06	−.24***		.21	.41
13. Race: Other	−.08	−.09	.00	−.09	.00	−.12†	−.18*	−.11	.01	.00	−.13†	−.14*	.07	.25

Note. N = 221. Missing data were accounted for using full information maximum likelihood estimation. SA = social aggression; PTNS = peer texting network size; PNT = peer network texts; PTNA = peer texting network aggressiveness.

^†p < .10.

^*p < .05.

^**p < .01.

^***p < .001.

Gender differences in teacher-rated social aggression were not significant in Grade 9, t(219) = −.46, p = .65, or Grade 10, t(219) = −.00, p = .99; however, on average, girls sent more socially aggressive texts to peers in Grade 9 (M = 6.90, SD = 9.80) than boys (M = 3.98, SD = 6.35), t(219) = 2.33, p = .02, despite there being no observed differences in overall frequency of texts with peers, t(219) = 1.53, p = .13 (M = 213.81, SD = 178.05, for girls; M =170.04, SD = 190.97, for boys). Differences between boys and girls in the number of socially aggressive texts sent to peers were not found in Grade 10, t(219) = 1.20, p = .23. Differences between girls and boys in their exposure to peers’ socially aggressive texts were nonsignificant in Grade 9 (M = 5.15, SD = 6.60, for girls; M = 3.44, SD = 5.80, for boys), t(219) = 1.71, p = .07, and Grade 10 (M = 5.69, SD = 5.77, for girls; M = 5.02, SD = 6.00, for boys), t(219) = .76, p = .45. On average, girls texted with a greater number of peers (M = 12.85, SD = 9.60) than did boys (M = 9.48, SD = 7.29), t(219) = 2.60, p = .009; furthermore, girls received socially aggressive text messages from a greater proportion of their peers (M = .16, SD = .17) than boys (M = .11, SD = .14), t(219) = 2.20, p = .03. Racial differences in teacher-rated social aggression and socially aggressive text messaging behavior were non-significant at both time points.

Peer socialization of youths’ socially aggressive text messaging.

A cross-lagged panel model (Figure 1, Model 1) was used to test the hypothesis that adolescents’ socially aggressive text messaging behavior is socialized through exposure to peers’ socially aggressive text messages. Results are shown in Table 2. Peer texting network size was a negative predictor of changes in adolescents’ socially aggressive texting (β = −.30, p < .001). The concurrent association between participants’ own social aggression and peers’ socially aggressive texting in Grade 9 was positive (β = .78, p < .001), suggesting that youth who were exposed to high levels of peers’ socially aggressive text messaging were themselves highly aggressive in their own text messaging behavior. Adolescents’ socially aggressive text messaging behavior was not stable across the year (β = .08, p = .50), accounting for the other variables in the model. Contrary to our expectations, adolescents’ socially aggressive texting in Grade 9 was not a predictor of changes in their exposure to peers socially aggressive texting (β = −.07, p = .55). Peer texting network aggressiveness (i.e., the proportion of peers who sent socially aggressive texts) was a marginally significant predictor of changes in adolescents’ socially aggressive texting (β = −.18, p = .09). However, frequency of exposure to peers’ socially aggressive texting behavior in Grade 9 was a positive predictor of adolescents’ own socially aggressive texting behavior a year later in 10^th Grade (β = .29, p = .02), controlling for adolescents’ gender, race, texting network size, peer text frequency, and peer network aggressiveness, as well as their prior levels of in-person and text-based social aggression.

Table 2.

Exposure to Peers’ socially aggressive texting in 9^th grade predicts changes in adolescents’ own socially aggressive texting from 9^th to 10^th grade.

	Socially Aggressive Texting Grade 10			Peers’ Socially Aggressive Texting Grade 10
Predictor in 9th Grade	b (SE)	95% CI	β	b (SE)	95% CI	β
Peer texting network size	−.25 (.08)	[−.41, −.08]	−.30**	−.25 (.07)	[−.39, −.12]	−.37***
Peer texting network aggressiveness	−8.32 (4.86)	[−17.90, 1.25]	−.18†	−8.52 (3.87)	[−16.16, −.89]	−.22*
Peer network texts	.01 (.004)	[.006, .022]	.37***	.01 (.003)	[.007, .020]	.43***
Gender	1.38 (1.00)	[−.58, 3.34]	.10	1.05 (.80)	[−.53, 2.63]	.09
Race: Hispanic	.76 (1.38)	[−1.97, 3.48]	.04	1.51 (1.11)	[−0.67, 3.70]	.10
Black	.76 (1.28)	[−1.76, 3.29]	.04	2.07 (1.03)	[0.03, 4.11]	.14*
Other	−2.33 (1.93)	[−6.14, 1.48]	−.08	−4.16 (1.76)	[−5.72, 0.44]	−.11†
Socially Aggressive Texting Grade 9	.07 (.10)	[−.13, .27]	.08	−.05 (.08)	[−.21, .11]	−.07
In-person SA Grade 9	.59 (.91)	[−1.20, 2.37]	.05	.59 (.79)	[−.97, 2.15]	.06
Exposure to Peers’ Socially Aggressive Texting	.33 (.15)	[.046, .621]	.29*	.35 (.12)	[.12, .58]	.37**

Note. N = 221. Missing data were accounted for using full information maximum likelihood estimation. SE = standard error; CI = confidence interval.

^†p < .10.

^*p < .05.

^**p < .01.

^***p < .001.

Gender differences in the association between adolescents’ exposure to peers’ socially aggressive texting and changes in their own socially aggressive texting were examined using multiple-group analysis. A configural model, in which model parameters were estimated freely for boys and girls, indicated little difference in the association between exposure to peers’ socially aggressive texting and changes in the socially aggressive texting of boys (β = .48, p = .01) and girls (β = .42, p = .01). Imposing an equality constraint on the model parameter representing peers’ socialization of socially aggressive texting did not lead to a significant reduction in model fit, Δχ²(1) = .209, p =.65, suggesting that exposure to peers’ socially aggressive texting predicted changes in adolescents’ own socially aggressive texting similarly for boys and girls.

Peer socialization of youths’ in-person social aggression.

The same cross-lagged panel modeling strategy was used to test the hypothesis that adolescents’ in-person social aggression is socialized through exposure to peers’ socially aggressive text messages (see Table 3). The concurrent association between participants’ in-person social aggression and their exposure to peers’ socially aggressive texting was marginally significant (β = .14, p = .10). Teacher reports of adolescents’ social aggression in Grade 9 did not predict subsequent teacher ratings of their social aggression in Grade 10 (β = .11, p = .21), suggesting instability in adolescents’ in-person social aggression. These findings mirror the instability observed in the objective text messaging data. Adolescents’ in-person social aggression in Grade 9 did not predict changes in their exposure to peers socially aggressive texting (β = .07, p = .38). Exposure to peers’ socially aggressive texting behavior in Grade 9 was a marginal predictor of participants’ own in-person social aggression in 10^th Grade (β = .29, p = .06), controlling for adolescents’ gender, race, texting network size, peer text frequency, and peer network aggressiveness, as well as their prior levels of in-person and text-based social aggression.

Table 3.

Exposure to Peers’ socially aggressive texting in 9^th grade predicts changes in adolescents’ in-person social aggression from 9^th to 10^th grade.

	In-person Social Aggression Grade 10			Peers’ Socially Aggressive Texting Grade 10
Predictor in 9th Grade	b (SE)	95% CI	β	b (SE)	95% CI	β
Peer texting network size	−.00 (.01)	[−.02, .02]	−.03	−.25 (.07)	[.12, .39]	−.37***
Peer texting network aggressiveness	−.63 (.43)	[−1.47, .21]	−.16	−8.62 (3.41)	[−15.34, −1.90]	−.23*
Peer network texts	−.00 (.00)	[−.001, .001]	−.07	.01 (.00)	[.007, .020]	.43***
Gender	−.01 (.10)	[−.20, .18]	−.01	1.04 (.82)	[−.57, 2.65]	.09
Race: Hispanic	.18 (.13)	[−.07, .43]	.11	1.52 (1.07)	[−.58, 3.63]	.10
Black	.13 (.13)	[−.12, .38]	.08	2.06 (1.07)	[−0.04, 4.16]	.14†
Other	−.19 (.19)	[−.57, .18]	−.08	−2.63 (1.60)	[−5.79, .52]	−.11‡
Socially Aggressive Texting	−.00 (.01)	[−.02, .02]	−.01	−.05 (.08)	[−22, .11]	−.08
In-person SA	.11 (.09)	[−.06, .28]	.11	.64 (.73)	[−.79, 2.07]	.07
Exposure to Peers’ Socially Aggressive Texting	.03 (.02)	[−.001, .058]	.29†	.36 (.12)	[12, .59]	.38***

Note. N = 221. Missing data were accounted for using full information maximum likelihood estimation. SE = standard error; CI = confidence interval.

^†p < .06.

^‡p < .10.

^*p < .05.

^***p < .001.

Gender differences in the association between adolescents’ exposure to peers’ socially aggressive texting and changes in their own in-person social aggression were examined, again using multiple-group analysis. A configural model, in which model parameters were estimated freely for boys and girls, revealed gender differences in the link between exposure to peers’ social aggression and changes in adolescents’ own in-person social aggression. Specifically, exposure to peers’ socially aggressive texting was a positive predictor of changes in adolescents’ in-person social aggression among girls (β = .49, p = .01) but not among boys (β = −.06, p = .82). Imposition of an equality constraint on the model parameter representing peers’ socialization of adolescents’ in-person social aggression lead to marginally significant reduction in model fit, Δχ²(1) = 3.480, p =.06, suggesting that differences between boys and girls in the association between exposure to peers’ socially aggressive texting and changes in their own in-person social aggression trended toward significance.

Discussion

Albert Bandura (1971) insightfully argued that “most of the behaviors that people display are learned, either deliberately or inadvertently, through the influence of example” (p. 5). Guided by this theoretical framework, this study used data on the content of adolescents’ text messages with peers to examine whether adolescents’ socially aggressive behavior, as expressed both in person and through text messaging, is socialized through exposure to their peers’ socially aggressive text messaging. Socialization processes that occur via text messaging may mirror those that occur in in-person settings (Subrahmanyam et al., 2006) and likely involve many of the same mechanisms. Through frequent exposure to peers’ socially aggressive texts, adolescents may begin to conform to group norms in order to preserve their group membership (Cialdini & Trost, 1998). At the same time, adolescents may model their peers’ socially aggressive behavior, particularly if they view those behaviors as socially rewarding. Finally, the constancy and immediacy of text messaging may also provide peers an enhanced opportunity to reinforce adolescents’ expressions of social aggression. Results from this study suggest that adolescents’ social aggression is socialized through exposure to peers’ socially aggressive texting. However, we were not able to directly identify which mechanisms of peers’ influence were most at play.

Surprisingly, the proportion of peers who sent socially aggressive texts (i.e., peer texting network aggressiveness), which may capture peer norms for social aggression, was not a significant predictor of changes in adolescents’ socially aggressive texting. However, frequency of exposure to peers’ socially aggressive text messaging in ninth grade was a positive predictor of adolescents’ own socially aggressive texting behavior a year later, controlling for prior levels of social aggression. Taken together, these results suggest that adolescents’ socially aggressive texting is socialized by peers via texting, but that group norms may not be the predominate mechanism through which peer influence is conveyed. Exposure to peers’ socially aggressive text messaging was a marginally positive predictor of changes in adolescents’ own in-person social aggression, as observed and rated by teachers, from ninth to tenth grade. These results build on prior research that has found evidence of peer socialization of adolescents’ social aggression in face-to-face contexts (Brendgen et al., 2008; Ellis & Zarbatany, 2007; Shi & Xie, 2012; Shi & Xie, 2014), by demonstrating how these socialization processes also occur through adolescents’ use of digital communication technologies, specifically text messaging.

Results from this study also indicated positive concurrent associations at both time points between adolescents’ own social aggression (in person and via text messaging) and their exposure to peers’ socially aggressive texting. These results suggest that exposure to peers’ social aggression tended to be higher among socially aggressive adolescents, perhaps reflecting processes of peer selection whereby socially aggressive youth chose to interact with similarly aggressive peers. However, counter to our expectations, neither adolescents’ in-person nor their text-based social aggression in Grade 9 predicted changes in their exposure to peers’ socially aggressive text messaging from ninth to 10^th grade. These results suggest that initial similarities in social aggression between adolescents and their peers might be maintained over time.

Although it was not a major focus of the present study, it should be noted that peer texting network size showed strong concurrent links to adolescents’ in-person and text-based social aggression, as well as their exposure to peers’ socially aggressive texting in Grade 9. These findings are in line with the general assumption in the field of peer relations that those who are highly central (or popular) tend to be the most socially aggressive (Cillessen & Mayeux, 2004). However, peer texting network size was a negative predictor of changes in adolescents’ socially aggressive texting. These results indicate that adolescents who texted with fewer peers in Grade 9 actually showed steeper increases in social aggression over time, perhaps suggesting that they became more aggressive as they sought to improve their status among their peers and grow their texting network.

That peer socialization of social aggression occurs via text messaging is important for both theoretical and practical reasons. The fact that peer socialization of social aggression occurs in digital as well as in-person environments is consistent with co-construction theory (Subrahmanyam et al., 2008), according to which “adolescents are not at the mercy of an externally created environment; they are creating, and more to the point, co-creating their Internet environment through processes of social interactions” (Subrahmanyam et al., 2006, p. 396). The findings from this study suggest that social interactions in digital communication are involved in peer socialization of adolescents’ social aggression, perhaps mirroring processes of socialization that occur during in-person peer interactions. Co-construction theory goes further and posits that adolescents navigate the same developmental challenges in their online and offline worlds (Subrahmanyam et al., 2006). Although adolescents socialize one another to engage in social aggression in their face-to-face interactions, their influence on each other’s social aggression is likely also facilitated by text messaging communication. Coconstruction theorists have suggested that, for youth, “physical and virtual worlds are psychologically connected” (Subrahmanyam & Greenfield, 2008, p. 124) and that online social lives may be “psychologically continuous” with their offline social worlds (Subrahmanyam et al., 2008, p. 421). This co-construction hypothesis is supported by our findings that exposure to peers’ socially aggressive text messaging predicted changes in target participants’ social aggression—in their in-person as well as in their text messaging interactions.

That peer socialization of social aggression occurs in text messaging also presents the possibility that the availability of digital communication may enhance or possibly even transform the power of peers to socialize each other with these highly hurtful but subtle behaviors. As research on adolescents’ use of social media has accumulated, a new transformational framework has been proposed that suggests that social media may enhance or transform adolescents’ peer interactions by increasing the constancy of experiences, amplifying experiences, altering the qualitative nature of interactions, and offering new opportunities for compensatory behaviors or novel experiences (Nesi et al., 2018a, 2018b). The findings from this study of text messaging fit with several of their hypotheses for effects of social media on peer influence (Nesi et al., 2018b). Compared to in-person interactions, text messaging may provide more frequent and immediate exposure to a greater volume of peers’ social aggression, and more opportunities for peer modeling and reinforcement of their own aggressive behaviors. Moreover, the relative privacy of digital communications technologies may provide youth a context in which there is less social risk in going along with peers’ social aggression. These features of digital communication may allow processes of peer socialization of social aggression to play out more quickly than they do in in-person interactions.

In addition to the findings about peer socialization of social aggression, the results of the current study add to the growing body of evidence that gender differences in social aggression are not large or consistent (Card et al., 2008). The behaviors that are part of the social aggression construct are consistent with gender stereotypes of girls and women as catty, manipulative, and backbiting (Underwood, Galen, & Paquette, 2001). These stereotypes are so pervasive in our culture that children as young as 3 years of age believe that relational aggression is more characteristic of girls than boys (Giles & Heyman, 2005). Early investigators made strong claims about gender differences in social aggression (for a review, see Underwood et al., 2001), but a thoroughgoing meta-analysis showed that gender differences were so small as to be trivial (Card et al., 2008). In line with these findings, the only gender differences in social aggression that were observed in this study were that ninth-grade girls sent more socially aggressive text messages than ninth-grade grade boys, this despite sending an equivalent number of texts overall. However, by 10^th grade, this gap had disappeared. Furthermore, at both time points, gender differences in participants’ in-person social aggression were nonsignificant.

Results from this study also suggest that boys’ and girls’ level of exposure to peers’ socially aggressive texting was similar. However, results concerning gender differences in the links between this exposure and changes in adolescents’ own social aggression were mixed. As expected, exposure to peers’ socially aggressive text messaging was a positive predictor of changes in boys’ and girls’ own socially aggressive texting, and gender differences in these associations were nonsignificant. However, gender differences in the extent to which exposure to peers’ socially aggressive texting predicted changes in adolescents’ in-person social aggression were found. Specifically, peers’ socially aggressive texting was a positive predictor of changes in adolescents’ in-person social aggression, but only among girls; boys’ exposure did not predict changes in their social aggression across the year. These mixed results match findings from a meta-analysis that found that observed gender differences in social aggression tend to be greatest when parent or teacher reports are used, and tend to be nonsignificant when observational methods are employed (Card et al., 2008). These authors suggest that “popular accounts of indirect aggression among girls likely contributes to a dualistic focus on gender differences (rather than similarities) among teachers, parents, and even researchers” (p. 1204). Thus, gender differences in peers’ text-based socialization of in-person social aggression found in this study may have resulted from differences in how teachers viewed, and thus rated, boys and girls on social aggression. Future work in which adolescents’ in-person social aggression is assessed via peer reports or by observations is needed.

Strengths, Limitations, and Future Directions

This is the first study to use objective data to examine peer socialization of social aggression as it occurs, naturalistically, via adolescents’ text messaging with peers. This strengthens findings from this study in two ways. First, by using objective measures of adolescents’ and peers’ socially aggressive text messaging, this study avoided potential problems caused by bias common to more subjective measures (i.e., self- or adult-reports), which have been used by most prior studies on peer socialization of adolescents’ social or relational aggression (Brendgen et al., 2008; Ellis & Zarbatany, 2007; Shi & Xie, 2012, 2014). Self-reports of social aggression may be prone to self-enhancement bias, and peers and adults may not be fully aware of the extent to which adolescents engage in socially aggressive behaviors because such behaviors are often covert in nature and, particularly in recent years, are enacted in relative privacy—while online or via text messaging. Second, and more substantively, capturing the content of adolescents’ text messages with peers allowed for an examination of peer socialization in an emerging digital context that has become significant in the lives of youth, but nonetheless has been largely understudied. Mobile communications technologies, including text messaging, now play a central role in the social lives of contemporary youth (boyd, 2014). Whereas a decade ago, more of the interactions that youth had with their friends and peer groups occurred in person, today’s youth are spending greater amounts of their time interacting with peers using digital communications technologies, including text messaging (Anderson & Jiang, 2018), and there is no reason to suspect that this will change. As such, contemporary research on peer socialization that relies exclusively on assessments of adolescents’ face-to-face peer interactions (that are observable by teacher, parent, or peer reporters) may miss capturing many meaningful social interactions that adolescents have with peers. Thus, it has become increasingly important for research on peer socialization, particularly during adolescence, to also focus on adolescents’ interactions with peers that occur using digital communications technologies.

Despite these strengths, some important limitations should be noted. First, the use of teacher reports to capture adolescents’ in-person social aggression may limit findings from this study. As already noted, teachers may not be fully aware or accurate in their reporting of the extent to which adolescents engage in socially aggressive behaviors. Self-reports, which were not measured in this study, may circumvent this problem. However, self-reports may be prone to other biases (i.e., social desirability or self-enhancement bias). It remains an empirical question whether findings from this study will replicate in research using self-reports (or other measures; e.g., observations or peer nominations) of adolescents’ in-person social aggression. Thus, future work using other measurement methodologies is needed.

Second, in terms of design, this study may be limited by using data collected at only two discrete time points in time. Future research using multiple years of data would make it possible to use more appropriate models of change, to examine cumulative long-term effects, and to explore nonlinear growth models. In addition, high concurrent correlations between participants’ and peers’ socially aggressive texting were observed at both time points. These results may be taken as evidence of homophily (Kandel, 1978), suggesting that adolescents choose to exchange text messages with peers to whom they are similarly aggressive. Alternatively, this association may reflect that these behaviors occurred sequentially in dyadic exchanges. Both explanations likely underlie this association. Future work examining the sequential nature of adolescents’ conversations with peers is needed and would allow for more direct tests of mechanisms of peer socialization (e.g., reinforcement). Machine-learning methodologies could be used to automatize coding so that many more days of texting could be available for sequential analyses of the more granular dynamics in socially aggressive texting.

The naturalistic design may also limit interpretation of findings from the current study. Naturalistic data bolsters ecological validity; however, in the absence of experimental control, we cannot be certain that any number of unmeasured variables would not have affected our results. For example, Shi and Xie (2012) found that the strength of peers’ socialization of social aggression varies depending adolescents’ social status relative to that of their peers’. Specifically, they found that high-status peers are most influential, particularly on lower status individuals. The associations between exposure to peers’ socially aggressive text messaging and adolescents’ social aggression found in the present study may have been attenuated by modeling the influences of low status peers on high status individuals. Future work examining peer influence in digital spaces should include measures of social status.

Third, this study only captured peer interactions that occurred via text messaging. Although texting was a prominently used platform for peer communication when data were collected for this study (Lenhart, Ling, Campbell, & Purcell, 2010), peer interactions that may have occurred using other digital communications platforms that existed at this time (e.g., social network sites, instant messaging applications) were not captured. Thus, to the extent that participants used additional platforms for communicating with peers, the current study may have only captured a slice of their exposure to and engagement in digital forms of social aggression. Furthermore, despite evidence that text messaging remains a central way in which contemporary youth interact with one another using mobile devices (Bailey et al., 2016; Lenhart, 2015; Rideout & Robb, 2018), other modes of digital communication have gained prominence since data for this study were collected (e.g., WhatsApp, Instagram, gaming platforms) and are now widely used among youth. Although it remains an empirical question whether or not the emergence of these new technologies has affected the frequency or quality of youths’ digital communication with peers, findings from this study may be limited in how accurately they capture youths’ technology-based behaviors, today and in years to come. To this point, as communications technologies continue to evolve and more options for digital communication become available, future research will need to account for these new forms and features of digital communication when examining social aggression among adolescents.

Relatedly, the current study did not include observations of participants’ face-to-face interactions with peers. Thus, examining whether adolescents’ socially aggressive text messaging behaviors were also shaped through exposure to peers’ social aggression during face-to-face interactions was not possible. Exposure to peers’ social aggression in digital contexts may overlay, add to, or amplify the socialization effect of being exposed to peers’ social aggression through in-person interactions (Nesi et al., 2018a, 2018b), and thus direct empirical tests of this theory are needed. Toward this goal, future research on peer socialization in digital domains should make use of additional data collection methods designed to identify the peers with whom adolescents interact in face-to-face settings (e.g., socio-cognitive mapping; Cairns, Perrin, & Cairns, 1985; Kindermann, 2007) and capture qualities of these interactions, in addition to the interactions they have using digital communications technologies. The possible additive, amplifying, and transformative effects that digital peer interactions may have on peer socialization can only be assessed empirically by research that includes both types of observational, naturalistic data.

Finally, although modeling and reinforcement provide a reasonable and theoretically sound explanation for why adolescents’ exposure to peers’ socially aggressive texting predicts changes in their own social aggression and socially aggressive texting, correlations between cyber aggression and cyber victimization found in prior work (Varjas, Henrich, & Meyers, 2009) suggest that adolescents’ social aggression may also be born out of cycles of retaliation and revenge. Measures of cyber victimization were not explicitly included in the current study, and objective data on social aggression victimization was not coded using the text messaging data. Thus, future research exploring whether being victimized by socially aggressive texts promotes adolescents’ social aggression is needed. Such research could make use of self-reports or objective data on youths’ cyber victimization.

Conclusion and Implications

Building on prior studies that have investigated peer influence on adolescents’ social aggression in face-to-face interactions (Brendgen et al., 2008; Ellis & Zarbatany, 2007; Shi & Xie, 2012, 2014; Werner & Crick, 2004; Werner & Hill, 2010), this study investigated whether processes of peer influence on adolescents’ social aggression also occur via text messaging. Evidence from this study suggests that they do. These results underscore how important it is for educators and parents to be aware of, and involved in, how the adolescents under their care are using digital communications technologies to interact with their peers. Given the amount of time adolescents spend in school, educators are centrally involved in preventing cyber social aggression among students. Schools should establish policies regarding students’ internet and mobile technology use, and provide staff and faculty with professional development training specifically focused on emerging media literacy, so that they are properly equipped to recognize and respond to instances of cyber social aggression involving students.

For parents, monitoring may be key to preventing or reducing adolescents’ social aggression in digital or online spaces (Vandebosch & Van Cleemput, 2009; Ybarra & Mitchell, 2004). In a study of 1,501 adolescents, Ybarra and Mitchell (2004) found that youth who reported low levels of parental monitoring showed a threefold increase in the likelihood of engaging in cyber aggression. By actively monitoring their children’s digital communications, parents may also be more aware of instances in which their children witness peers’ socially aggressive behavior, and thus may be better able to intervene more efficiently and effectively. To minimize the intrusiveness of their monitoring, parents should foster positive, open lines of communication with their children, specifically regarding appropriate use of these technologies. Parents should frequently engage their children in discussions about their “online” and “offline” peer experiences, taking care to avoid being overly invasive or judgmental (Goldstein, 2015). Indeed, many adolescents are willing to talk to their parents about them (Zhou et al., 2013). Through these practices, educators and parents may be able to decrease adolescents engagement in social aggression, in person and via digital communications technologies, and the effects of their exposure to it.