Adolescents who are nonusers of fashionable social networking platforms

Introduction

The influence of social networking platforms on modern adolescents’ interactions with peers has been well-documented (Underwood et al., 2018). These digital mediums have rapidly assumed a central role in adolescents’ social worlds (Mikami et al., 2010). There is a wide array of social networking platforms available, with most options including tools that allow users to communicate through customized profiles, virtual communities, messaging, and shared media (Shapiro & Margolin, 2014). Adolescents’ interest in specific platforms can shift rapidly with technological developments and changing fads, but the vast majority of American youth acknowledge frequent use of the most fashionable options. In fact, many of today’s adolescents access popular social networks on a nearly constant basis (Anderson & Jiang, 2018).

The ubiquitous nature of social networks in adolescent peer groups has motivated considerable empirical interest (Nesi et al., 2018a, 2018b). Much of this work has focused on the attributes of adolescents who are avid consumers of social media (e.g., Marino et al., 2020). Conversely, relatively little attention has been paid to those adolescents who are not users of these communication modalities. In survey research, a small percentage of youth report that they do not have an active presence on any of the networking platforms that are commonly accessed by their peers (Vannucci & McCauley Ohannessian, 2019). The goal of the current project is to investigate the psychosocial adjustment of this understudied subgroup. We will rely on a school-based data collection to examine the social and academic functioning of these adolescents.

Peer Group Relationships

Lack of activity in digital realms could have significant social implications for today’s adolescents. Nesi and colleagues (2018a, 2018b) emphasized the transformative impact of mobile communication technologies on the organization of modern school peer groups. These theorists note that adolescents’ interactions with peers are now easily observable by wide audiences, accompanied by quantifiable indices of prominence (e.g., number of likes or followers) and documented using new media (e.g., photos and videos). Activity on social networking platforms may be especially impactful because digital interactions can be continually accessed and follow adolescents across contexts (e.g., home and school).

The centrality of social networks to the secondary school peer environment is also likely to accelerate over time. At an earlier stage in the evolution of the social networking phenomenon, Valkenburg and Peter (2009a) hypothesized that ideas about online activity displacing offline connectedness would become increasingly obsolete as new communication modalities achieved widespread acceptance. Indeed, the significance of social networks for day-to-day interactions has grown as specific platforms have gained influence in adolescent peer groups (Nesi et al., 2018a, 2018b). Social networks have evolved from modalities that facilitate interactions with unfamiliar peers to essential tools for intensifying connections to the existing peer group (Valkenburg & Peter, 2009a).

Popularity and Social Preference

The emergence of social networking platforms as a dominant communications modality might have particular significance for the dynamics that maintain social standing in adolescent peer groups. During adolescence, popularity requires a degree of compatibility with the values and interests of peers (Santor et al., 2000). Youth who do not participate in interactions on the social networking platforms that are most widely accessed by other adolescents in the immediate setting may be viewed as violating peer group norms and could experience sanctions as a result (Schwartz et al., 2013). These adolescents could also be deprived of opportunities that might otherwise promote integration and visibility with peers. They might miss out on group invitations, information about social events, or exposure to aspects of youth culture.

Likewise, an absence from fashionable social media could be one manifestation of a subordinate position in the peer group hierarchy. Adolescents affiliate with specific peers based partially on the presumption that social reputation can be enhanced through association (Dijkstra et al., 2013; Marks et al., 2012). Social networking platforms offer a publicly visible communication modality through which youth could demonstrate their membership in high status cliques (Yau & Reich, 2019). In contrast, status-seeking adolescents might eschew digital interactions with unpopular classmates as a means of avoiding degradation of their own social image.

Limited activity on fashionable networking platforms could also be indicative of low social impact rather than an aversive social reputation. Adolescents may lack opportunities for online interactions because they do not have high visibility in the peer group. This argument is reminiscent of the sociometric classification systems that were once widely influential in the literature on children’s peer relationships (Newcomb et al., 1993). Sociometric researchers characterized a subset of youth as “neglected” (neither well-liked nor actively rejected).

A related point is that status in the peer group is not likely to be an overriding goal for all youth. Although popularity with peers can emerge as a central objective by early adolescence, other priorities may take precedence for a select group of youth (Dawes & Xie, 2014). For some adolescents, disinterest in fashionable social networks can indicate a lack of interest in adopting behavioral styles that promote popularity (Cillessen & Mayeux, 2004). The link between unpopularity and avoidance of trendy social media could then reflect broader value systems.

As researchers consider the potential link between social standing and lack of activity on fashionable social networking platforms, it will be important to recognize that adolescents’ peer relationships are multifaceted. In particular, developmentalists have demonstrated that popularity and social preference are distinct aspects of social standing with peers (Cillessen & Mayeux, 2004). Popularity is an indicator of status, prestige, and dominance (Parkhurst & Hopmeyer, 1998). Conversely, social preference is operationalized as a high level of positive regard from peers combined with low levels of disliking (Cillessen & Marks, 2011). Popularity and social preference are not synonymous constructs, particularly when youth rely on aversive strategies (e.g., gossip, rumor spreading, and manipulation) to solidify their position in the peer hierarchy (Prinstein & Cillessen, 2003). Some popular adolescents may even be actively disliked by a large percentage of peers (Cillessen & Rose, 2005).

Popularity is the index of peer group standing that has been emphasized in the extant literature on electronic communications (e.g., Nesi & Prinstein, 2015). This direction in the literature seems sensible insofar as popularity requires a prominent image that might effectively be demonstrated through a visible presence in the digital world (Yau & Reich, 2019). Social preference, in contrast, is determined by prosocial attributes and the absence of aversive behaviors (e.g., aggression; Coie et al., 1982). Preference does not require a carefully cultivated persona. For these reasons, a limited presence on fashionable networking platforms might be linked to low popularity while not having strong associations with social preference.

Friendship

Although popularity and other forms of standing with peers have critical implications for adjustment (Schwartz & Gorman, 2011), adolescence is a developmental period that brings an increased focus on friendship (McNelles & Connolly, 1999). Friendship is a dyadic relationship that incorporates warmth, intimacy, and emotional support (Hartup, 1996). For many adolescents, social networking platforms serve as the primary forum for day-to-day interactions with friends (Anderson & Jiang, 2018; Yau & Reich, 2018), and can be important tools for intensifying these relationships (Valkenburg & Peter, 2009b). One reason some youth might not gravitate toward digital communication modalities is that they have relatively few friends and, hence, limited options for online partners.

Fortunately, youth who do not occupy a central role in the peer group social system can still have friends. A defining feature of friendships in adolescence and across the lifespan is homophily (i.e., similarities within dyads; J. V. Hamm, 2000). Marked social difficulties in the broader peer group will not preclude connections with a select group of behaviorally similar classmates (Flannery & Smith, 2017). Adolescents who are not embedded in a dense social system that supports digital interactions might still enjoy a small number of close friendships at school (Antheunis et al., 2012). Accordingly, our expectation was that youth who do not have a presence on fashionable social media platforms would succeed in establishing small networks of behaviorally similar peers.

Peer Victimization

Another aspect of social adjustment with peers that merits consideration is the potential link between victimization in the school peer group and abstention from social networking platforms. Adolescents describe mistreatment online as an important motivation for avoiding electronic interactions with particular peers (M. P. Hamm et al., 2015). This harassment in the digital world can reflect spillover of existing peer group dynamics. Indeed, youth who report being victims of cyberbullying also endorse experiences of peer mistreatment at school (Jose et al., 2012). Additionally, the negative social ramifications of an absence from the digital world might result in an adolescent being bullied at school (Brewer et al., 2017).

A caveat to this discussion is that there may be significant differences in the pattern of associations for distinct subtypes of peer victimization. Frequent overt victimization (e.g., being hit, threatened, or insulted) is likely to identify youth who occupy the bottom rungs in the peer group hierarchy (Sheppard et al., 2019). To the extent that lack of activity on fashionable social networking platforms is indicative of low social status, nonusers may be particularly likely to experience this form of mistreatment by peers. The corresponding effects for relational forms of victimization could be more equivocal. Gossip, rumors, and other relational forms of victimization can be markers of popularity (at least for some youth) and often occur as a manifestation of day-to-day jockeying for social power (Malamut et al., 2020).

Academic Competence and Behavioral Adjustment at School

The importance of peer group processes notwithstanding, some adolescents may refrain from social media because they have other priorities. Academically-inclined youth, who are busy with schoolwork and studying, may be less likely to focus on digital interactions than some of their peers. In that case, a lack of activity on social networking platforms could identify youth who are oriented toward academic achievement.

The idea that high-achieving adolescents may have limited enthusiasm for some forms of social media is consistent with uses and gratification theory (e.g., Sundar & Limperos, 2013). This theoretical perspective posits that choices in digital media will be optimized to address an individual’s goals and interests. Thus, adolescents’ enthusiasm for specific digital modalities is expected to reflect their own personal objectives. Academically-oriented students may gravitate toward electronic modalities that enhance learning (e.g., e-mail exchanges with teachers) while being disinterested in the more socially-oriented modalities that are in vogue with peers.

Related aspects of classroom functioning might also warrant consideration. It seems reasonable to predict that adolescents who have a strong academic orientation will be compliant with rules and avoid unexcused absences or disciplinary incidents. Information on these aspects of school adjustment could further enhance current understanding of the underlying attributes of youth who are not active users of fashionable social networks.

Depressive Symptoms

Although our primary focus was on the academic functioning and peer relationships of fashionable social networking nonusers, we also conducted descriptive analyses that targeted symptoms of depression. The extant work has not produced consistent findings and is generally more focused on frequent users of social media than nonusers. A number of investigators have concluded that frequent digital activity could be linked to depression and related difficulties. In their recent review, Vidal et al. (2020) noted that increases in social media use among adolescents have paralleled increased rates of depression and suicide. On the other hand, Valkenburg and Peter (2009a) hypothesized that the implications of social media have evolved over time, with frequent use promoting connectedness and positive adjustment rather than isolation. Przybylski and Weinstein (2017) presented a nuanced perspective suggesting that there may be a “just right” level of use that is neither indicative of disengagement from peers nor excessive enough to be disruptive.

Research on the adolescents who are infrequent users of social media is less extensive, but does not suggest risk for depression. For example, Nesi and Prinstein (2015) found that unpopular adolescents who avoid social media are not characterized by high rates of depression. It is possible that some unpopular adolescents are buffered by the social support provided by a small group of friends, and they may also achieve benefits to self-esteem associated with high achievement (Schwartz et al., 2008). As a note of caution, the limited findings that are available focus on infrequent users rather than nonusers. A complete absence from fashionable social networking platforms could still be indicative of processes that intensify isolation from peers and exacerbate personal distress (Przybylski & Weinstein, 2017). In light of these competing possibilities, we did not form strong a priori hypotheses, but adopted an exploratory stance.

Gender and SES as Potential Confounders

To increase the precision of our analyses, we included gender as a covariate in all models. Girls tend to be more active on most forms of social media than boys (Anderson & Jiang, 2018). There are also well-documented gender differences that relate to both academic functioning (Scheiber et al., 2015) and social adjustment (Rose & Rudolph, 2006).

Likewise, we considered socioeconomic status (SES) as a potential confounder. Survey research indicates that SES has only limited significance for adolescents’ use of social networking platforms (Associated Press-NORC Center for Public Affairs Research, 2017). Nonetheless, economic resources can create imbalances in access to smartphones and high-speed data (Anderson & Kumar, 2019). Accordingly, youth from disadvantaged backgrounds may tend to avoid social networking tasks that are resource intensive (e.g., uploading videos or photos).

The Current Investigation

This project used a cross-sectional design with data from peer nominations, self-report, and review of school records. We targeted the middle adolescent years, a developmental period that is marked by frequent use of electronic communications (Anderson & Jiang, 2018). We expected to identify a small subgroup of youth who are not active on any of the fashionable social networking platforms that are favored by peers. Our hypothesis was that these social network nonusers would differ from peers who are active on fashionable social networks along a number of key dimensions. Specifically, we expected that these youth would be less popular and more overtly victimized than their classmates. Additionally, we predicted that nonusers would establish friends with a relatively small number of behaviorally similar peers. We also anticipated that nonusers would have friends who are comparatively high achieving and compliant with school rules.

Method

Overview

This investigation was conducted as part of a larger project, completed in a semi-urban school district located in the greater Los Angeles region. Analyses based on this project have appeared in past reports (Schwartz et al., 2019), although our previous publications did not consider the attributes of social networking nonusers. The participating school had a low dropout rate (approximately 4%) at the time of data collection. Achievement tests were not administered by the school district while this research was underway because the statewide batteries were being revised. In previous years, school-wide test averages were above mean levels for the state, but were similar to secondary schools in proximal school districts.

The surrounding community is ethnically diverse, with Latinx and European American families particularly well represented. The majority of families living in these neighborhoods are from middle class backgrounds, but there is considerable variability reflected in a range of housing options (e.g., apartment rentals as well as high priced single-family homes).

This study was conducted in compliance with the ethical standards of the American Psychological Association and was approved by our institution’s Internal Review Board.

Participants

Participants were recruited from ninth grade (first year of high school) English classes. In the spring of 2016, 659 eligible ninth graders were invited to participate and 413 returned positive parental consent. Thirty-three potential participants transferred from the school district before the data collection began. Of the remaining youth, 376 (M = 14.4 years of age; 209 girls, 167 boys) assented to participate in writing, were present in school during the data collection and provided data regarding electronic communications. These participation rates are typical for schools in diverse urban settings (Schwartz et al., 2016), and reflect the complexities of data collection inherent in modern urban environments. The self-reported ethnic/racial composition of participants was 29.14% Latinx, 27.27% White, 9.89% Asian/Pacific Islander, 2.14% African American, .27% American Indian, 27.54% mixed race/ethnicity, and 3.74% not classified.

Procedure

Participants completed measures in group-administered sessions led by trained research assistants. The research assistants read standardized instructions to the participants and circulated the classroom to answer questions. School records were reviewed in the following summer.

Measures

Electronic Communication and Social Media Use

Our study objectives were concerned with youth who do not access any of the social networking platforms that are widely used by their school peers. Potential platforms for consideration were selected based on existing survey data (Lenhart, 2015). We also met with recent high school graduates from this region, parents of current students, and school teachers. Following these sources, we identified the following online social networks for assessment: Facebook, Google+, Instagram, Pinterest, Reddit, Snapchat, Tumblr, and Twitter. Study participants were asked to rate the frequency with which they accessed each of the identified platforms on a five-point Likert scale (0 = never, 1 = once a month, 2 = once a week, 3 = once a day, 4 = several times a day).

Table 1. Percent of Participants Using Each Social Networking Platform.

As shown in Table 1, the adolescents in our sample tended to be heavy users of Instagram, Twitter, and Snapchat. The remaining social platforms were each used by less than 20% of the participants. The pattern for Facebook was notable as this platform was a popular option for American adolescents in the years immediately prior to our data collection (Lenhart, 2015). Still, only a small percentage of youth reported any use of Facebook. Furthermore, an exploratory factor analysis of the frequency ratings for all of the assessed social networks highlighted the peripheral role of Facebook. The factor that accounted for the most variance included high loadings for the three most commonly accessed platforms: Twitter (.69), Snapchat (.85), and Instagram (.82). In contrast, the loading for Facebook was modest (.23).

Guided by uses and gratification theory (e.g., Sundar & Limperos, 2013), we sought to identify the social networking platforms that adolescents would be likely to utilize as means of reaching or maintaining prominence in the peer group. Accordingly, we based classification of nonuser/user status on the three platforms (i.e., Snapchat, Twitter, and Instagram) that were widely used in this sample. Youth who reported that they never used Snapchat, Twitter, or Instagram were classified as fashionable social network nonusers. Youth who were active on these social networking platforms with any degree of frequency were classified as fashionable social network users.

As Table 1 shows, adolescents who did not use fashionable social networks may still have an active digital presence on platforms that were not often selected by peers. A small percentage of these youth were occasional users of social networks that are not optimized to foster connections in the existing peer group but are likely to facilitate interactions between unfamiliar users (e.g., Reddit and Pinterest). Platforms that serve primarily informational purposes may have also been selected by some of these youth. For example, Facebook did not appear to be a popular outlet for peer-to-peer communication during the period of data collection but may have been a useful platform for viewing announcements regarding school activities.

Past researchers have tended to use less strict cutoffs when conducting subgroup analyses. Typically, infrequent users (including youth who report that they occasionally use social networking platforms) are identified based on cutoff scores (Rideout, 2012) or empirically derived classifications (Vannucci & McCauley Ohannessian, 2019). Our classification system was based on a theoretical interest that focused specifically on a more extreme subgroup of adolescents who have no presence on the social networks widely used by other adolescents.

Of the 376 participants in this study, 48 (12.77%) were classified as fashionable social network nonusers. Fourteen (26.17%) of the nonusers were girls; 31.25% identified as Latinx, 27.08% identified as European American.

Academic Performance

Grades in English, math, and science were obtained from school records in the summer following each school year. The grades were coded on a five-point scale (0 = F and 4 = A). A grade point average (GPA) was calculated from the mean of grades (α = .94).

Academic competence has been widely discussed in the extant literature on adolescent digital communications (e.g., Schwartz et al., 2019), although past researchers have relied nearly exclusively on self-report data. Self-report estimates can be quite problematic insofar as meta-analyses suggest poor agreement with actual school outcomes (Kuncel et al., 2005).

Compliance with School Rules

A disciplinary events score was created for each participant by summing the total number of incidents involving after-school detention, suspension, detention during recess/lunch periods, and referrals to the principal’s office for incidents that occurred during the course of the school day (e.g., academic dishonesty, violations of dress code, disrespectful behavior toward teachers).

We also calculated a separate unexcused absence index based on the number of days each student was absent from school without a valid excuse.

Social Reputation

A peer nomination inventory was utilized to assess youths’ reputations with peers. Participants were provided with an alphabetized roster of randomly selected consenting peers. The lists were approximately 50 names in length, and there were eight different lists. Names could only appear on one list. Based on these rosters, students were asked to identify peers who fit a series of descriptors. Participants could identify as many peers as they desired for each descriptor. Because the roster lists were distributed randomly, a small number of adolescents could have nominated themselves. However, self-nominations were removed during calculation of the peer nomination indices. This assessment approach reflects de facto standards for peer nomination research with adolescents in secondary schools (e.g., Schwartz et al., 2006).

The inventory included 17 items, with descriptors addressing a range of both negative and positive aspects of social functioning. Of relevance to the current study, respondents were asked to nominate peers who are popular (“students who are popular”), unpopular (“students who are unpopular”), liked (“students who you really like”), disliked (“students you don’t like”), overtly victimized (“students that get hit, pushed, or bullied by other students”; “other kids start fights with them”; r = .37, p < .0001) and relationally victimized (“get mean things said about them”; “other students try to be mean to them by ignoring them or excluding them”; r = .53, p < .0001). The correlation between the two overt victimization items was modest, but we opted to combine the items into a summary index given consistency in third-variable correlates. We also note that the overt victimization items were worded in a manner that was intended to capture both verbal (e.g., insults and threats) and physical forms of overt victimization.

The total number of nominations received for each item was calculated. As per Cillessen and Mayeux (2004), we created a popularity score based on the difference between the popularity and unpopularity score standardized within the list. A social preference score was also calculated based on the standardized difference between the liking and disliking totals (Coie et al., 1982). In addition, we calculated overt victimization and relational victimization scores based on the sum of the nominations received for the relevant items standardized within list.

Friendship and Attributes of Friends

Following completion of the peer nomination inventory, the participating adolescents were given a second roster list that included all participants in the project. Thus, friendship selections were based on the full grade rather than on the randomized lists. They were then asked to circle the names of any students who were their “really good friends.”

Friendship choices were unlimited so that the participants could circle as many or few names as desired. Youth were classified as friends if they reciprocally identified each other (e.g., Child A identified Child B as a friend, and Child B identified Child A). The total number of friendships involving each adolescent was then generated. 95.19% of all study participants had at least one friend, 96.63% of the social network users had at least one friend, and 85.42% of the social network nonusers had at least one friend.

Our focus on reciprocated friendship selections warrants attention given that a defining feature of friendship is mutuality. Perhaps, for reasons of expedience, investigators do not always require dyadic ties and rely on unilateral nominations instead. These less stringent assessment approaches can compromise validity and lead to overestimates (Badaly et al., 2012). For some adolescents, systematic biases in awareness of friendship can reflect specific social goals (Garandeau & Lansu, 2019) or underlying schemas (Badaly et al., 2012).

Next, a series of indices were generated to assess the attributes of the adolescents’ friends, including mean levels of social adjustment among friends (popularity, social preference, relational and overt victimization), mean levels of school competence among friends (GPA, unexcused absences, and disciplinary events), and mean levels depressive symptoms among friends. These scores were calculated for the 356 adolescents who had at least one friend with missing values assigned to the remaining adolescents.

Socioeconomic Status (SES)

Students reported on both parents’ occupations and education levels as per Hollingshead (1975). Students could select an occupation level from structured choices, or enter an open-ended response. Open-ended responses were coded by trained research assistants with discrepancies resolved through discussion. Data were double-weighted for single parent homes. The Hollingshead Four Factor Index was then calculated based on these data. SES is often estimated with data derived primarily from parental reports (e.g., Dodge et al. 1994), but past investigations have validated estimates derived from adolescents’ report (Duong et al., 2014).

Results

Overview

Our study objective was to examine the academic and social adjustment of adolescents who are not active users of Instagram, Snapchat, or Twitter. To this end, we conducted analyses in several distinct steps. For descriptive purposes, we began by examining univariate and bivariate statistics for all constructs. We also considered group differences for each construct (nonuser vs. user, boys vs. girls) as well as correlations between the adjustment variables.

Our inferential analyses were based on a series of logistic regression analyses. Separate models (four total) were specified with odds of nonuser status predicted by adolescents’ own social adjustment with peers (Model 1), adolescents’ academic and behavioral adjustment at school (Model 2), friends’ social adjustment with peers (Model 3), and friends’ academic and behavioral adjustment at school (Model 4). Logistic regression is an efficient tool for examining simultaneous associations between multiple indicators and a dichotomous construct. Nonetheless, we are explicitly not conceptualizing social media nonuse as the dependent construct given our cross-sectional design and our theoretical assumption that relations are reciprocal. We use the term “prediction” in a statistical sense (i.e., to denote whether variables are on the right or left side of a model), and are not implying causal relations.

Descriptive Statistics

Table 2. Means and Standard Deviations for All Study Variables.

Variables tended to be distributed with modest skew with the exception of the unexcused absence and disciplinary event scores. As might be expected, a relatively small percentage of children were either frequently absent without a valid excuse or had frequent disciplinary events. We applied inverse and log transformations to these scores (Tabachnick & Fidell, 2007) to normalize distributions. However, transformation had no impact on the full pattern of results. To facilitate interpretation of odds ratios in logistic regressions, we retained the untransformed scores in our final models.

Table 2 presents the means and standard deviations for all variables. Differences in mean scores for boys and girls, and for fashionable social network nonusers and users, are also summarized (as assessed via a series of independent samples t-tests). Note that bivariate group differences are summarized for descriptive purposes only with our eventual conclusions based primarily on multivariate models. Compared to boys, girls had higher GPAs, were less overtly victimized, had higher levels of depressive symptoms, and had fewer friends. Moreover, girls’ friends had higher levels of depression and were less overtly victimized than boys’ friends.

The findings with regard to differences between users and nonusers of fashionable social networking platforms were somewhat more substantial. Compared to fashionable social network users, nonusers had lower scores for popularity, social preference, total number of friends, unexcused absences, disciplinary events, popularity among friends, disciplinary events among friends, and social preference among friends, as well as higher scores for relational and overt victimization, GPA, and GPA among friends. On the other hand, there were no significant differences between social network nonusers and users with regard to socioeconomic background or depressive symptoms.

Girls were underrepresented in the fashionable social network nonuser group and were also more depressed than boys, overall. Accordingly, null effects for nonuser/user status on depression scores might be confounded by gender differences. To consider this possibility we also examined nonuser/user difference for depression separately by gender group. However, differences between users and nonusers were negligible for both boys (Ms = 0.31 and 0.32, respectively) and girls (Ms = 0.45 and 0.49, respectively).

Bivariate Correlations

Table 3 summarizes correlations between all study variables. Table A1 in the Appendix reports the correlations by gender. These effects are, again, presented for descriptive purposes only. Particular caution is warranted with regard to the correlations between adolescents’ own attributes and the attributes of their friends (e.g., the correlation between an adolescent’s popularity and the popularity of his/her friends). Although similar correlations are presented elsewhere in the extant literature, friendships are interchangeable dyads so the relevant effects can be inflated due to nonindependence of observations (Kenny et al., 2006).

Nonuser Status and Adolescents’ Psychosocial and Academic Functioning

In order to examine the adjustment of social networking nonusers, we specified two separate logistic regressions (Table 4). The first of these models (Model 1) focused on social functioning in the peer group. Nonuser status was predicted from SES, gender, popularity, social preference, total number of friends, relational victimization, and overt victimization. All variables were entered into the model simultaneously. As depicted in Table 4, there was a significant positive effect for gender and significant negative effect for popularity and number of friends. Odds of being a nonuser were increased among adolescents who were low in popularity, had relatively few friends, and were male.

The second logistic regression model focused on indicators of academic adjustment (Model 2). Nonuser status was predicted from gender, SES, GPA, unexcused absences, and disciplinary events. There was a significant positive effect for GPA and a marginal negative effect for unexcused absences. Odds of being a nonuser were increased among adolescents who had higher GPAs and marginally fewer unexcused absences than classmates.

Table 3. Correlations Among All Study Variables.

Table 4. Odds of Nonuser Status and Adolescents’ Own Social and Academic Adjustment.

Nonuser Status and Friends’ Psychosocial and Academic Functioning

Attributes of social networking nonusers’ and users' friends were compared using two separate logistic regression models (see Table 5). We first specified a model (Model 3) predicting nonuser status from adolescents’ gender, adolescents’ SES, popularity among adolescents’ friends, social preference among adolescents’ friends, overt victimization and relational victimization levels among adolescents’ friends. As summarized in Table 5, there was a negative association between friends' popularity and odds of being a nonuser.

Similarly, we specified an additional model (Model 4) predicting nonuser status from adolescents’ gender, adolescent’s SES, average GPA for adolescents’ friends, average disciplinary events for adolescents’ friends, and average unexcused absences for adolescents’ friends. This model yielded a significant positive effect for friends’ GPA and significant negative effect for friends’ unexcused absences. Odds of being a nonuser were increased among adolescents whose friends had high GPAs and high unexcused absences. Note that the effect for unexplained absences is opposite in direction to our hypotheses and also inconsistent with the bivariate effects summarized earlier. This pattern is likely to be an artifact of correlations between predictors and is not easily interpreted.

Table 5. Odds of Nonuser Status and Friends’ Social and Academic Adjustment.

Discussion

Interactions that occur in the context of social networks are now central to the organization of adolescent peer groups (Shapiro & Margolin, 2014). Nonetheless, not all adolescents are active users of the platforms that tend to be favored by their peers (Vannucci & McCauley Ohannessian, 2019). These nonusers of fashionable social networks are an under investigated subgroup with only a handful of relevant studies available (e.g., Rideout & Robb, 2018). In fact, the current manuscript reports the first known project to use a multi-informant approach to assess the academic and social functioning of nonusers. The present study also incorporates a unique focus on friendship and the attributes of friends.

Peer Relationships

A small percentage of participants in this study reported that they did not use any of the social networks that were widely preferred by their peers during the period of data collection (i.e., Snapchat, Instagram, and Twitter). Relative to the other adolescents in our sample, nonusers were characterized by low levels of popularity. Moreover, in bivariate analyses, these youth tended to have higher peer victimization scores than their classmates. The effects for victimization did not replicate in multivariate analyses and should not be interpreted in isolation. Still, the full picture suggests that adolescents who are not active users of fashionable social media tend to be marginalized in the peer group ecosystem.

Because this project featured a cross-sectional design, we are not in a strong position to make statements regarding direction of causality. This shortcoming notwithstanding, we would predict reciprocal relations between social media nonuse and peer group outcomes. Adolescents who are unpopular or have difficulty making friends are unlikely to have frequent opportunities for online interactions with peers. Likewise, limited engagement with the digital world might detract from social standing in the peer group.

The underlying processes are likely to be multifaceted and an adolescent’s absence or presence on specific platforms could reflect multiple mechanisms. Nonuse of fashionable social networking platforms may represent a personal choice (e.g., an adolescent who chooses to refrain from online activity as a reflection of personal priorities), limited opportunities due to low social status and friendlessness, and/or external factors that are outside the adolescent’s own control. For example, some adolescents may have parents who restrict access to networking platforms or other digital mediums. The motivations and factors that are associated with nonuse warrant further attention in field designs but have at least been examined in qualitative and interview studies (e.g., Baker & White, 2011).

We also emphasize that our hypotheses focus specifically on the social networking platforms that were most widely used by participating adolescents during the period of data collection (Instagram, Snapchat, and Twitter). Our assumption was that lack of activity with these widely-used electronic modalities would be one indicator of an adolescent’s social goals and experiences with peers. We nonetheless recognize that adolescents who do not use Instagram, Snapchat, or Twitter may still have rich digital lives that are oriented toward the pursuit of other objectives. Uses and gratification theory (Ruggiero, 2000) would suggest that adolescents will make choices regarding the use of digital media that reflect their own priorities and needs. For example, some adolescents may be more focused on success in multiplayer video games than achieving online social prominence with familiar peers (Granic et al., 2014). Other adolescents may gravitate toward platforms that are targeted toward special interests or interactions with peers outside the peer group (e.g., discussion boards that are organized around sports, technology, music, or media, rather than interactions between familiar peers).

Friendship and the Attributes of Friends

Although adolescents who are not active on fashionable social media tended to have low popularity scores, most of these youth did have at least some friends. An unresolved question in the literature on peer relationships is whether the role of friendship is linear. It is possible that beneficial impacts of friendship continually increase as a reflection of total number of friends (e.g., Schwartz et al., 2008). In that case, the potential buffering effects might be limited for those adolescents who have few friends. An alternate model could suggest that one or two friends might be sufficient to buffer youth against the risk associated with other social difficulties (i.e., low social status, rejection, or peer victimization; Laursen et al., 2007).

It is, perhaps, not surprising that nonuser status was positively associated with achievement levels among friends and negatively associated with popularity levels among friends. We are unaware of any existing study that has examined relations between online communication patterns and the attributes of friends, but there is strong evidence in the extant literature regarding homophily effects (Hanish et al., 2005). Across the lifespan, similarity is a defining feature of friends (Youyou et al., 2017). Adolescents are likely to affiliate with peers who share their own interests and values (Rambaran et al., 2017).

To some extent, the composition of their proximal peer environment is likely to be a manifestation of active selection (Van Zalk et al., 2010). Nonusers of fashionable social networks may choose friends who have similar orientations with regard to core developmental domains, including school (Ryan, 2000). Because these youth seem to prioritize classroom performance and school attendance, they will be likely to gravitate toward peers who share these values. Regardless of the causal process, the implication is that youth who are not active on the social networking platforms that are currently in favor with the broader peer group are likely to benefit from exposure to friends who are high achieving.

Academic Functioning

Uses and gratification theory portrays digital communications as a goal-oriented activity (Sundar & Limperos, 2013). Based on this perspective, we might hypothesize that a lack of engagement with fashionable social networking platforms is indicative of value systems that do not emphasize status in the peer group. One possibility is that, for these adolescents, classroom performance carries more weight than high status with peers. Consistent with this expectation, nonusers tended to have higher GPAs than their peers and they were somewhat less likely to have unexcused absences from school.

The link between academic functioning and activity on fashionable social networking platforms may also be reciprocal. In some school peer groups, adolescents can experience social sanctions if they are viewed as being especially studious (Schwartz et al., 2013). A salient developmental task of adolescence is autonomy and an orientation toward school, as an adult institution, can have negative impacts on social reputation. One implication of these processes may be that academically-engaged adolescents have limited opportunities for online interactions.

Depressive Symptoms

Despite their subordinate position in the peer group hierarchy, social networking nonusers were not characterized by high levels of depressive symptoms. Competency models of childhood depression focus on the additive role of challenges in multiple domains of development (Cole et al., 1997). According to these models, an adolescent who is experiencing some social difficulties (e.g., peer victimization, rejection, unpopularity, or friendlessness) can still have other broad competencies with regard to academic functioning. Children and adolescents are also likely to emphasize positive information as they construct their self-images. For example, an unpopular adolescent who has a high GPA will prioritize academic achievement when evaluating self-competence (Schwartz et al., 2008). Under these conditions, proficiency in academic realms could buffer against disappointments in other domains.

Values or personal objectives can also have implications for depression and related outcomes. Social networking nonusers may be youth who simply weigh academic excellence more highly than status with peers or visibility on Snapchat, Instagram, and Twitter. Past research has shown that popularity can be predictive of academic disengagement (Troop-Gordon et al., 2011). Conversely, high achievement may have negative social implications in some peer groups (Ryan, 2000). Lack of interest in trendy social media might be an indicator of particular orientations toward school and peer relationships.

Putting aside speculative comments, our null findings with regard to depressive symptoms are not a strong basis for conclusion. Moreover, our analyses of depression were exploratory and were not driven by strong a priori hypotheses. Future efforts to examine links between social media use and psychological distress could also benefit from more complete assessment of related constructs (e.g., social anxiety, generalized anxiety, loneliness, and low self-esteem).

Gender and SES

Boys were more likely than girls to be nonusers of fashionable social networking platforms (as indicated by significant odds ratios in our logistic regressions). This pattern was not unexpected insofar as national surveys indicate that girls are likely to be more avid users of social networks than boys (Anderson & Jiang, 2018). Accordingly, we entered gender as a control variable in all of our inferential models. The results were supportive of our hypotheses even after we accounted for the main effects of gender.

It should be emphasized that analyses in which gender is operationalized as a covariate are not equivalent to tests of moderation (i.e., tests of gender by predictor interactions). In this project, we did not attempt to ascertain if relations between nonuser status and psychosocial functioning differ for boys and girls. Previous tests of gender differences in associations between social network activity and indicators of academic and social adjustment have generally not yielded significant results (Schwartz et al., 2019). It may be too early for strong conclusions, though and the role of gender as a potential moderator construct could warrant further investigation.

SES was also a covariate in our models but we did not find strong patterns of associations between SES and the constructs examined in this project. A more compelling pattern may have resulted if a wider range of advantage and disadvantage was considered. Our project was focused narrowly on a middle-class community.

Limitations and Future Directions

The design of this study breaks new ground with a school-based data collection and a unique focus on the attributes of friends. Still, several limitations should be highlighted. Our main hypotheses did not emphasize causal direction. Instead, we conceptualized lack of activity on fashionable social networking platforms as a marker of social difficulties and academic achievement. Accordingly, we relied on a cross-sectional data collection. The next steps in this domain of inquiry will need to incorporate prospective designs that focus on longitudinal progressions.

Research that moves toward an exploration of longitudinal (and potentially causal) associations could help move research in this domain in an applied direction. The centrality of social media in young people’s social lives suggests a powerful vantage point for examining mechanisms that can either intensify psychosocial risk or enhance resilience. An empirical focus on subgroups of users, or specific usage patterns, could inform intervention efforts and extend current understanding of risk processes.

Conclusion

The results of the current investigation provide new insight into the attributes of adolescents who are not users of the social networking platforms that are most widely chosen by their peers. These adolescents are not popular with peers, but they are generally characterized by high achievement and compliance with school rules. Moreover, they tend to establish friendships with a small number of peers who share their positive orientation toward school and achievement. The full pattern emerging from these findings suggests that adolescents who are not active on fashionable social media platforms may encounter some social difficulties with peers but still retain academic strengths.

Acknowledgement

The authors gratefully acknowledge the contributions of Dr. Andrea Gorman (Occidental College) who assisted in school recruitment and data collection. Dr. Gorman passed away during the course of the project. This manuscript is dedicated to her memory.

Appendix

Table A1. Correlations Among All Study Variables by Gender.