Teen Talk: The Good, the Bad, and the Neutral of Adolescent Social Media Use

Prior research has documented various benefits of teens’ social media usage, such as connecting and socializing to break their isolation (Maheux et al., 2021), learning (Asterhan and Bouton, 2017), and building support networks (Sanger, 2022). These positive effects were found to be particularly salient during and after the physical isolation due to the global pandemic  (Haddock et al., 2022). For instance, Maheux et al. (Maheux et al., 2021) conducted a longitudinal study with 704 adolescents to examine the impact of social media use on their gratitude before and during the pandemic over 15 months. In this study, a significant positive association was found between adolescents’ social media usage for meaningful conversations with their friends and their gratitude over time. Additionally, social media facilitated online peer support, which also benefits youth by providing users the ability to connect with similar others without fear of stigma or judgment, to create supportive networks and communities to discuss sensitive topics while choosing how much to share with others (Sanger, 2022). Online peer support platforms were found to offer benefits for teens such as emotional support, the availability of advice, engagement in valuable social interactions, and the opportunity to disclose and express feelings and views (Rayland and Andrews, 2023). Especially, online peer support for individuals with mental health challenges experienced increased empowerment, self-efficacy, and better management of depression, enhanced coping strategies, and reduced social isolation  (Melling and Houguet-Pincham, 2011). Overall, it is important to note that while social media provides a wide array of benefits, responsible and mindful use is crucial to have a safer engagement online.

Therefore, over the years, a major number of prior works have extensively studied the negative impact of social media usage on adolescents, shedding light on its multifaceted consequences (Valkenburg et al., 2022; Paat and Markham, 2021). For instance, researchers have consistently pointed to numerous concerns related to the relationship between teens’ social media usage and their mental health issues, such as increased feelings of loneliness, anxiety, and depression (Keles et al., 2020). Given the prevalence of cyberbullying among adolescents due to the usage of social media platforms, several studies have primarily focused on examining the detrimental impact of cyberbullying on teens, uncovering how these experiences had severe psychological effects on teens such as increased depression, anxiety, and suicide ideation (Kwan et al., 2020; Levine et al., 2022; Doumas and Midgett, 2023). In addition, several studies have explored the negative impact of social media through the lens of social comparison, a process where individuals assess their opinions, values, achievements, and abilities in relation to those of others (Powdthavee, 2014). For example, past research found that using Instagram may heighten social comparison, subsequently leading to a decrease in self-esteem (Jiang and Ngien, 2020). Further studies have also indicated that the use of Instagram could lead to feelings of insufficiency due to upward social comparison (Hwnag, 2019). Other researchers have looked at the negative aspects of social comparison shedding light on other platforms. For example, Vogel et al (Vogel et al., 2014) found that frequent Facebook use is correlated with lower self-esteem, particularly when users compare themselves to others who appear more successful or happier on the platform. In a similar vein, Burke et al (Burke et al., 2020) found that people who often compare themselves to others on Facebook tend to be highly active on the site and have larger friend networks. However, despite being exposed to more positive and socially engaging content, a substantial portion of these users experience negative feelings after encountering posts that make them feel worse about themselves. Other studies have also indicated that social comparison on social media has been linked to various negative outcomes such as depression (Samra et al., 2022) and body dissatisfaction (Jiotsa et al., 2021). Pitfalls of social media-based online peer support have also been highlighted, such as personal distress due to others’ experiences (Easton et al., 2017), unhelpful interactions with others  (Griffiths et al., 2015), social exclusion  (Easton et al., 2017), and feeling of vulnerability when talking to strangers online (Breuer and Barker, 2015). Emerging evidence also documented other adolescent online safety issues including online sexual risks (Hornor, 2020; Alsoubai et al., 2022; Razi et al., 2023), exposure to explicit media content (Chatterjee and Kar, 2023; Park et al., 2023a; Rostad et al., 2019), and problematic media use with mental health issues  (McHugh et al., 2018; Park et al., 2023b; Patchin and Hinduja, 2017).

Given the disagreement within the literature, whether social media is beneficial or detrimental for adolescents remains an open question to explore that requires a nuanced picture of its potential benefits and detriments. The review and meta-analysis also showed that the association between social media use and teens’ psychological well-being is still unclear as effects have been found to exist in both (positive and negative) directions (Orben, 2020). Yet, positive and negative effects can co-exist as effects can vary across different adolescents; some adolescents experience net positive effects, while others experience heightened risk  (Odgers et al., 2022). For instance, through analyses of 2,155 real-time assessments of the adolescents’ unique susceptibility to the effects of social media, Beyens et al. (Beyens et al., 2020) illustrated that the impact of passive (but not active) social media use on their well-being differed significantly from teen to teen, with 44.26% of teens did not feel better or worse, 45.90% felt better, and a small group felt worse (9.84%). Therefore, efforts to holistically understand what adolescents are going through on social media are imperative to provide appropriate support systems for them and to inform design implications to provide a safer online environment, keeping in mind to maximize the benefits while mitigating potential risks and negative outcomes. This study takes this nuanced approach by analyzing teens’ disclosures about their positive and negative experiences on various social media platforms.

How youth seek and receive online social support has been an important area of study within the CSCW and HCI community. Previous research showed that adolescents go online to seek social support for sensitive topics such as mental health (Suzuki and Calzo, 2004), romantic relationships (Kim et al., 2017), and sexual experiences (Razi et al., 2020). This is because adolescents are hesitant to receive professional help or talk to their parents about such sensitive topics  (Forte et al., 2014; Hartikainen et al., 2021) because of perceived stigma and embarrassment (Radez et al., 2021). Most teens who face relationship difficulties do not seek help from people they know in person due to fear of judgment or concerns about confidentiality  (Kim et al., 2017). Therefore, they turn to online social support to feel more comfortable interacting with people they do not know offline (Ellison et al., 2016). Recently, researchers utilized digital trace data from social support platforms to understand adolescents’ online social support-seeking behaviors (Razi et al., 2020; Hartikainen et al., 2021; Prescott et al., 2017). For instance, by conducting a thematic content analysis of 622 initial posts shared by young people aged 11-25 years, Prescott et al. (Prescott et al., 2017) explored how youth seek support on the online forum for emotional and mental health issues. They observed two distinctive ways of youths’ online support-seeking: one was a direct request for advice, with a themed heading followed by a post with details about the specific advice they sought, while the other was finding other young people on the forum who shared similar experiences to themselves. Another main theme they found was that in many posts, youth shared their personal experiences and offered support for others  (Prescott et al., 2017). More recently, Razi et al. (Razi et al., 2020) conducted a thematic analysis of over 4,000 teen posts on a peer support platform to understand how youth talk about their online sexual experiences. They found that most youths have utilized the platforms to seek support (85%), while 15% were more interested in connecting with others, and 5% gave advice on different sexual matters. Although previous work has provided valuable insights into adolescents’ support-seeking behaviors, most have focused on a particular type of negative online experiences (e.g. online sexual experiences, romantic relationship issues, digital stress) or risks experienced on a specific social media platform. Our work expands on previous studies by understanding adolescents’ experiences on various social media platforms for which they seek online peer support and the differences in those experiences depending on different social media platforms. To do so, we analyzed posts that adolescents shared on an online social support platform for teens, where they explicitly mentioned the name of popular social media platforms.

Meanwhile, adolescents use a variety of different social media platforms for different purposes including learning, entertainment, networking, and communicating with others (Anderson et al., 2022a; Moreno and Uhls, 2019). For instance, teens use TikTok for entertainment, learning, and discovering trends, while, they use Snapchat for socializing, and go to Instagram for sharing pictures and checking what others are doing (Anderson et al., 2022a). Within the HCI literature, users’ behaviors and experiences on social media platforms have been studied through the lens of “affordances,” a relational structure between the social media platforms and the users (Faraj and Azad, 2012). As a concept that captures the relationship between technology and users, affordances have been applied in social media research as a key concept to examine the perceived features of social media platforms that enable or constrain users’ behavior on the platform (Park et al., 2023b; Ronzhyn et al., 2022). For instance, on social media platforms that value revealing real identities (e.g., Facebook), users tend to be more careful with how they present themselves, while on platforms that afford low identity affordances (e.g., Reddit), users disclose sensitive information without repercussions of personal identity (Leavitt, 2015). With their low identity affordances, (pseudo) anonymous platforms provided users with online spaces to seek social support regarding sensitive topics such as mental health, sexual encounters, and more. As such, affordances can aid us in understanding how adolescents interact on different social media platforms and in guiding them on the risks and benefits of each platform.

Prior research has revealed that user perceptions and interactions with social media platforms can vary significantly depending on the platforms’ affordances. For example, affordances can influence the selection of social media platforms (Shane-Simpson et al., 2018); Instagram is favored for its visual capabilities in sharing and enhancing photos and videos, Facebook for connecting with friends and family through personal interactions, and Twitter is implied to be appreciated for its real-time, concise communication ideal for disseminating information and participating in public dialogues. Another study comparing affordances between TikTok and YouTube platforms revealed that users prefer YouTube for its ability to control the sharing of information and manage who sees their information, activities, or profiles; on the other hand, TikTok was favored for helping users stay informed about information and the presence of others, allowing customization to suit personal preferences and interests, and making it easier to share content within the platform and to other platforms (Lee et al., [n. d.]). Furthermore, prior research suggests that individual differences affect how users interact with and perceive social media. One study (DeVito et al., 2017) found that people’s traits, such as their experience with the platform and personality affect how they present themselves online. For instance, experienced users of a platform are likely to be more skilled at navigating the platform’s features to manage their online presence more effectively. Personality also plays a role; individuals open to exploring diverse creative tools may express themselves more freely, while those higher in neuroticism may prioritize privacy controls and worry about negative feedback, affecting their online presentation (DeVito et al., 2017).

As such, previous literature provides an understanding of the concept of affordances and how it can be applied in social media research with general populations, yet the application of affordances in the adolescent online safety context remains under-studied. The motivation to explore social media affordances in the context of teens lies in understanding how the features of social media platforms shape teenagers’ behaviors, experiences, and well-being. This exploration is important because teens may use social media differently than adults due to their unique developmental needs (Senekal et al., 2023), social dynamics (Jang et al., 2015), and technological skills (Throuvala et al., 2019). In this study, we examined teens’ differing experiences across multiple social media platforms and interpreted some of our findings in light of the unique affordances of social media platforms. To do so, we explored the posts teens shared on an online peer-support platform with low identity affordances (i.e., registered users use pseudo-anonymous user names) to understand in what ways teens discuss social media platforms when they seek online support. By analyzing adolescents’ posts on a pseudo-anonymous platform, we provide an understanding of adolescents’ experiences on social media that they may not otherwise share with their parents or peers. In doing so, we investigate whether the way teens discuss social media platforms varies across different platforms.

We licensed an existing dataset from an online peer support platform that aims to offer youth and young adults a safe environment to discuss topics such as mental health, sexuality, religion, and more. The name of this platform was anonymized to protect the teens’ identities whose data we analyzed. On this platform, users can create posts, leave comments on posts, and choose whether to publish the posts anonymously or using their username. Hence, users have the ability to share their screen names and profile photos with other users when they intend, making their posts pseudo-anonymous; thus, safeguarding users’ personal privacy. The platform is designed exclusively for sharing text-based posts, hence no images were collected in our dataset. While the dataset did not include users’ country of origin, most posts were in English. The licensed dataset consisted of about five million posts from 400 thousand users, including metadata (e.g., likes, hearts, and moods). The dates of the posts in the dataset ranged from 2011 to 2020. More details about the characteristics of our sample can be found in section 4.1.

The Institutional Review Board (IRB) of our university determined that this study was exempt from the review for human subject research because personally identifiable information (e.g., usernames, contact information) was removed from the dataset prior to being shared with the research team. Nevertheless, we ensured that all authors completed their IRB Collaborative Institutional Training Initiative (CITI training) on human subjects research before accessing this dataset. We also took great care to remove any potentially personally identifiable information from the posts themselves (e.g., screen names for other platforms, hyperlinks) prior to sharing posts as exemplar quotations in the paper.

Our primary goal was to examine how teens (ages 13-17) discussed different social media platforms in their support-seeking posts. To accomplish this, we first ran a SQL query that searched for posts created by users whose ages were between 13 and 17 years old. Then, we narrowed our search criteria based on a keyword search of the ten most popular social media platforms among teens in the U.S., which included Instagram, Twitter, Facebook, Snapchat, YouTube, TikTok, WhatsApp, Reddit, Twitch, and Tumblr (Vogels et al., 2022). Our search resulted in posts mentioning nine of the top ten platforms (Twitch did not appear in the search results). Also, to focus on more recent trends, we only included posts from the last two years of the dataset (2019 and 2020), which resulted in a total of 2,465 posts. Some of these posts (n = 205) were coded as irrelevant due to the alternative use of some of the keywords we used in our search, which can be found in Table 1. For example, the post “well ig y’all don’t understand” appeared in our search results because of the word in bold. This post was deemed irrelevant because of the keyword “ig,” which in this context means “I guess”. Lastly, after removing irrelevant (8%), duplicate (7%), and non-English posts (1%), the final dataset contained 2,061 posts. We only focused on the original posts, excluding the comments on the posts, since we aimed to understand the support-seeking behavior of teens through their posts, rather than the support and advice they received as comments.

We first conducted a thematic analysis (Braun and Clarke, 2012) of teens’ posts about their experiences on social media platforms (RQ1). To do this, we familiarized ourselves with the data by reading through the posts. The first two authors then discussed the main topics presented in the posts to create initial codes. Subsequently, they divided the posts equally and coded using the same set of initial codes, while actively identifying new themes. Whenever the two authors identified potential new codes, other co-authors deliberated on whether to include them in the codebook. If they reached an agreement, the first two authors went back and re-coded all previous posts to incorporate the new code. This process was iterative, involving constant communication and consensus among the researchers. Once all the data coding was completed, the first two authors worked with the last author to conceptually group the codes into cohesive themes that aligned with our RQ1. Finally, we reviewed our themes alongside our dataset to confirm that they actually captured important meanings within the coded data and named the themes. Table 2 presents the themes and their corresponding codes along with illustrative quotations. Since we employed a double-coding approach for the posts, the total count of codes may exceed the overall number of posts. For example, teens sometimes mentioned more than one social media platform with varying experiences across the differing platforms. Therefore, the count of all codes can be totaled to be more than 100% of our total post count.

After completing qualitative analysis (RQ1), we leveraged our codebook for statistical tests to identify patterns and trends in the data. While RQ1’s qualitative analysis provided in-depth insights into teens’ complex online experiences, a limitation lies in the challenge of integrating information across observations or assessing associations between observations (Kirk and Miller, 1986). Conversely, the strength of the quantitative approach lies in its ability to compare observations and examine associations (Castro et al., 2010). Our mixed-methods design, crucial for addressing RQ2 (relationship between platforms and teens’ support-seeking behavior), combines qualitative and quantitative data (Creswell et al., 2003). This approach facilitates accurate insights from confirmatory statistical analysis and deep explanatory descriptions derived from thematic analysis. For the qualitative examination of the relationship between social media platforms and teens’ support-seeking behavior, we conducted $\chi^{2}$ tests, a between-group test for two or more variables (Sharpe, 2015). This method ensures that observed relationships between social media platforms and codes/themes from thematic analysis are not due to random chance, enhancing the rigor of thematic analysis findings. Prior research also supported such use of between-group tests in studies involving qualitative analysis of online content (Unni and Weinstein, 2021; Basch and MacLean, 2019; Harvey et al., 2019; Lee et al., 2019). To demonstrate significant differences between social media platforms, standardized residuals were used, calculated by ”dividing the product of subtracting expected from observed values by the square root of the expected value” (Sharpe, 2015). Through these tests, we interpret nuanced differences between platforms based on youth disclosures of their experiences.

Prior to addressing our RQs, we first assessed for potential effects of the COVID-19 pandemic, which occurred in 2020. Therefore, we conducted a chi-squared ($\chi^{2}$) between the years 2019 and 2020 based on our main themes from RQ1. This analysis revealed no significant differences in the described themes in Table 2, except for seeking information about social media. Standardized residuals showed that, during the pandemic, youth in our dataset were more likely to discuss seeking information related to social media platforms in 2020, compared with 2019. We highlight this difference in our results (section 4.5). Next, we conducted a Chi-squared between-group analysis ($\chi^{2}$) to examine any significant differences between the online platforms based on the codes for each sub-theme (RQ2). We first categorized the posts based on the different social media platforms teens mentioned in their posts. Some posts (n = 122) had multiple social media platforms mentioned, so we labeled the same post for each platform cited. For two of the social media platforms, Reddit and Tumblr, the code counts were notably small, with over 20% of the codes having counts below five. This rendered our data inconsistent with the assumption of the $\chi^{2}$ test of having the expected values cells to be 5 or more in at least 80% of the cells (McHugh, 2013). Therefore, we excluded these two platforms and conducted $\chi^{2}$ tests among the social media platforms that were more frequently discussed among the youth.

Our study involved an analysis of 2,061 posts generated by 1,038 distinct users between the age range of 13 to 17 years. However, all the posts found were made by adolescents who fell within the age group of 15 to 17 years old. Notably, a significant proportion of the posts (70%) were created by teens who were 17 years old, while about a third of posts came from those aged 16 (24%), and only 1% of posts came from those who were 15 years old at the time of posting. With respect to gender distribution, the majority of the posts were made by adolescents who identified as female (53%), followed by those who did not specify their gender (33%), and lastly, by those who identified as male (14%). As shown in Table 3, the most mentioned social media platform in adolescents’ posts, Instagram was the highest with 35% (n = 720) of total posts. Snapchat ranked second with 20% (n = 412). YouTube was the third most mentioned platform among teens with 14% (n = 282), followed by TikTok (10%, n = 207), Facebook (7%, n = 147), WhatsApp (n = 127) and Twitter (n = 125) with the same frequency level (6%). Lastly, Tumblr (n = 28) and Reddit (n = 13) were only mentioned in 1% of the posts. Keeping in mind the relative percentages expressed by platform, there are some notable trends in Table 3 to highlight. For instance, negative experiences were most frequently shared when teens referred to Instagram, Facebook, Twitter, Tumblr, and Reddit. Snapchat and Whatsapp had the highest relative proportion of posts related to using social media to connect with people, while YouTube had the largest proportion of posts related to information seeking. YouTube was the only platform where the number of positive posts exceeded the negative, possibly speaking to its overall popularity among youth. Tables 2 and 3 serve as the organizing structure for reporting our findings below. The $\chi^{2}$ results for RQ2 are presented at the beginning of each subsection and helped guide the selection of illustrative quotations presented in our qualitative results (RQ1). We use illustrative quotations to describe each theme and subtheme of our results. Each quotation is contextualized with the teen’s age and gender information.

We found that in 43% of the posts (n = 891), teens mentioned different social media platforms included in our dataset to share the negative experiences that they had online. The $\chi^{2}$ test indicated significant differences between these platforms based on relative expected proportions of the total number of posts $(\chi^{2}(df=36)=120.11,p<0.001$). Below, we present how teens explained the adverse experiences they faced on different social media platforms.

Overall, we found that 570 teens mentioned different social media platforms to connect with others in their posts (45%, n = 922). To this end, they invited others to connect on a variety of different social media platforms. Based on the $\chi^{2}$ test, there were significant differences between these online platforms based on the codes that emerged from teens’ discussions about connecting with people theme ($\chi^{2}(df=18)=153.53,p<0.001$). Below, we present how teens invited others to increase their social connections.

Meanwhile, 151 teens talked in a portion of their posts (20%, n = 409) about the positive aspects of different social media platforms. Statistically significant differences between the social media platforms based on the codes emerged from the positive sides of social media theme ($(\chi^{2}(df=6)=65.72,p<0.001$). About 11% of the posts made by adolescents (n = 234) emphasized how they utilize social media as a coping mechanism during moments of low mood or emotional distress. YouTube had the highest relative proportion of posts describing social media as a coping mechanism, followed by Snapchat and then TikTok as shown at Figure 4. Yet, the standardized residuals showed that when teens posted about how they used social media as a coping mechanism, these posts had the highest probability to involve Snapchat, Twitter, and Facebook and the least probability to involve Instagram. Within these posts, some teens explained how engaging with others on social media platforms provided them with a sense of belonging and connectedness during hard times. For instance, some teens mentioned that they share funny selfies on Snapchat when they feel stressed during studying. Other teens found Snapstreak on Snapchat (i.e., tracks how many days in row users and their friends have exchanged Snaps and provides users extra points for their Snapchat score) rewarding enough to continue doing it as an important daily routine, as shown in the following post.

“This Snapstreak on Snapchat is life saving. It’s the most important daily-routine of my life right now. Gives a purpose.” - Female, 16 years old

Others shared that consuming positive content on social media helped them relax when they felt stressed. For instance, some teens manage their Twitter accounts to push funny videos so that when they feel stressed, they can visit Twitter and ease their stress by watching those videos. For instance, in the following post, a teen shared that they filtered their Twitter content by setting preferences to display only humerous videos. When feeling sad, they turned to those curated feed of videos which served as a form of distraction or mood enhancement for them.

“I set my Twitter settings to have all funny videos and that helps me to stay away from all BS. When I am sad I just scroll through and that just works like magic.” – Unspecified, 17 years old

As such, our analysis revealed that teens turned to social media for more than just posting content or engaging in Snapstreaks; they used it as a tool for emotional coping and seeking social support. These platforms provided spaces tailored to their emotional and social requirements. Moreover, it allowed teens to control their social interactions, empowering them in times of stress or isolation.

In some cases (8%, n = 175), teens acknowledged social media as a valuable source of inspiration, motivating them through exposure to creative content, success stories, and inspirational quotes. YouTube had the highest relative proportion of posts describing social media as a source of inspiration. However, Instagram reached positive statistical significance as demonstrated at Figure 4. This suggested that teens more frequently included Instagram in their discussions about how social media platforms provided inspiration and support. In their posts, teens most often expressed how this exposure helped them gain a fresh perspective when dealing with challenges and adversity in their offline or online lives. For example, some teens shared how they were motivated by the inspiring content they watched on Instagram:

“”Treat yourself like someone you loved.” - some inspiring dude I saw on Instagram.” – Female, 16 years old

Overall, teens mentioned different social media platforms to appreciate the way they received support and inspiration from others and how social media, in general, helped them with their mental health. Next, we explore the ways in which teens sought information regarding different social media platforms on a peer support platform.

As we examined posts, we found that 235 teens aimed to seek information about various aspects of social media platforms in 20% of their posts (n = 421). Upon closer examination of our standardized residuals between 2019 and 2020, we saw that this trend has increased during the pandemic compared to 2019. Overall, the $\chi^{2}$ test yielded significant differences between the platforms based on the emerged codes within seek information theme emerged from teens’ discussions ($(\chi^{2}(df=12)=92.86,p<0.001$). For the most part, this was noticeable when teens discussed social media content (12%, n = 247). As illustrated in Figure 5, YouTube had the highest relative proportion of teens’ posts about discussing social media content across other platforms, showing also a positive statistical significance. This indicated that teens most frequently discussed content found on YouTube when seeking information about social media content. Within this subset of posts, teens actively sought input from their peers by requesting recommendations for various social media content, e.g., YouTube videos, ranging from makeup tutorials to mental health support. Often times, they expressed a desire to confirm the authenticity and credibility of social media content.

“[Youtube link] is this actually true?.. like for real?” – Unspecified, 17 years old

Next, in 5% of the posts (n = 100), teens were seeking information on how to use different social media features. When comparing the relative proportions of posts asking about features of social media platforms across all platforms, we found that Instagram had the highest relative proportions followed by Snapchat and then Instagram as shown at Figure 5. However, based on the standardized residuals, there were significant associations between posts inquiring about social media features and Twitter. This indicated that teens’ posts asking about social media features had the highest probability to involve Twitter. We saw that many of these posts were centered around content creation, where teens posed questions regarding methods for creating fascinating and visually pleasing content. They primarily sought advice on employing filters, editing tools, and creative strategies to enhance their posts to appeal to their intended audience. In some other instances, teens sought assistance in understanding how to navigate privacy features (e.g., how to connect with others on specific social media platforms) to manage their online presence.

“im sorry, but how are you gaining an audience on an English-language twitter?” – Unspecified, 16 years old

Lastly, a portion of the posts made by teens (4%, n = 74) centered around asking questions about different technical challenges they faced on social media. Figure 5 illustrated that Instagram had the highest relative proportion of posts discussing social media technical challenges in comparison with other platforms in our dataset, reaching also a positive statistical significance. This indicated that when teens seeking technical support, they most frequently referred to Instagram. In many cases, the technical challenges were about server problems or scheduled maintenance, resulting in temporary unavailability or restricted functionality on the platforms. In these instances, teens turned to the peer support platform to seek solutions and reassurances to navigate through the technical difficulties they encountered. For instance, a teen posted to inquire if anyone was experiencing issues with their Instagram, specifically mentioning a problem where the feed was not updating or changing as expected.

“Somebody here has a problem with their Instagram? like the feed won’t change.” – Female, 17 years old

In summary, when teens mentioned various social media platforms in their posts, they predominantly posted to vent about different unpleasant experiences they faced on social media such as cyberbullying and privacy violation, while referring to Instagram, and Facebook. Furthermore, teens often shared how social media adversely impacted them in terms of their self-esteem, time management and arousing their anger particularly when mentioning Instagram, YouTube, and TikTok. In addition, teens often focused on connecting with others in their posts for self-promotion and increasing their followers while referring to Instagram, YouTube, and TikTok. Also, teens expressed a need to communicate with others for emotional support and building online relationships when mentioning Snapchat, Whatsapp, and Facebook. On the other hand, teens also recognized and appreciated its positive aspects. For instance, they highlighted how social media served as a coping mechanism during difficult times and how it provided them with inspiration to strive for greater success, particularly citing Snapchat, Facebook, Instagram, and Twitter. Lastly, teens also sought information and guidance on how to effectively navigate the different social media platforms, mostly pointing to YouTube, Twitter, and Instagram. In the next section, we discuss the implications of these findings.

Our qualitative findings for RQ1 unpacked negative experiences that were indexed more frequently than the positive ones, as shown in Table 2. Yet, these seemingly overwhelming number of negative experiences might stem from teens’ inclination to share during challenging times, especially since the context of the digital trace data analyzed was intended to garner peer-support. Therefore, we stress for the readers to avoid marking teens’ social media usage using a false dichotomy of negative or positive. Instead, our focus should be on understanding how teens feel vulnerable to which online harms and for what underlying reasons so that we can provide appropriate support. This recommendation is well-aligned with prior research that emphasized studying the factors that make some teens at higher risk than others online (Orben, 2020). The impact of social media platforms on teens can vary from individual to individual, impacted by factors such as personality traits (Rahardjo and Mulyani, 2020), family environment (Vanwesenbeeck et al., 2018), and the overall digital environment (McCreanor et al., 2013).

Our research also highlights positive aspects of social media use among teens, such as coping with stress by sharing their content, connecting with others online, watching inspirational content to boost their mood and self-esteem, and seeking information during the pandemic. Zooming out from the immediate findings, positive aspects of teens’ social media usage were well-documented in the surveys and/or empirical studies (e.g., increased social connection, identity development, and positive emotions through social support) (Anderson et al., 2022a; Haddock et al., 2022). In addition, the meta-analyses confirmed that the association between social media use and teens’ negative psychological well-being is still unclear, as effects have been found to exist in both (positive and negative) directions, and there has been little work done to rule out potential confounders (Orben, 2020). In a recent Pew Research survey, the most common way teens describe the impact of social media was neither positive nor negative (Anderson et al., 2022a). Nevertheless, heightened attention to the negatives of teens’ social media usage led U.S. Senators Blumenthal and Blackburn to propose the “Kids Online Safety Act (KOSA)”  (https://www.blumenthal.senate.gov/, 2022) to protect teens from online risks. The legislation requires social media platforms to proactively mitigate risks to minors. To do so, it requires independent audits and public scrutiny from experts to ensure that social media platforms are taking meaningful steps to address risks to kids (https://www.blumenthal.senate.gov/, 2022). Some embrace the idea of KOSA given multiple claims that social media companies fail to protect teens (Fung, 2023), while others fear that this legislation would incentivize social media sites to collect even more information about minors to prevent potential risks for them.

Given the positive stories teens shared in our study and the trends found in existing research, we need to move away from over-emphasizing the harms that social media usage can bring to teens and be overly paternalistic about protecting them from potential harm. In the same Pew Research survey above, 22% teens perceived that their parents are extremely worried about their social media life, while 39% teens share that their online experiences are better than parents think (Anderson et al., 2022a). Therefore, we once again highlight that social media usage is not necessarily positive or negative; rather, the two aspects co-exist without voiding one another, and net effects (positive or negative) can vary across different teens in various contexts. Therefore, further research is needed to understand differing teens’ online experiences and how to amplify positive aspects at the same time tackle negative aspects to better support teens’ digital well-being.

When teens discussed their negative experiences using social media, they often mentioned specific social media platforms. Specifically, our statistical analysis (RQ2) illustrated how teens mostly mentioned Instagram whenever they discussed body-shaming and its harmful effect on their self-esteem, Facebook when they talked about interpersonal privacy violations, and YouTube when they discussed their struggles with time management (Figure 1 and Figure 2). To some extent, these statistical differences can be explained by the social media affordances, features, and social norms of these platforms that facilitate how users communicate, share information, and engage with others (Zhao et al., 2013). For instance, Instagram is a photo-focused platform, hence, a unique culture (e.g., users get more attention by posting physically appearing photos) was created on this platform that has strongly emphasized the significance of physical appearances (Tiggemann and Barbato, 2018), which may affect teens as viewing some idealized pictures of others diminishes their self-esteem (Casale et al., 2021). Moreover, we saw posts in which teens described how following individuals whom they perceive as having a desirable lifestyle or attractive look on Instagram led them to social comparison and feel lower self-esteem, which has also been documented in prior research (Jiang and Ngien, 2020). Meanwhile, Facebook was launched over 20 years ago as one of the first platforms that allow users to use their real names (Lee, 2021). Given this high identity affordances, family relationships are emphasized on this platform in a way that parents like to post about their teens (Burke et al., 2013). These “sharenting”  (Blum-Ross and Livingstone, 2020; Steinberg, 2016) posts were found in our qualitative analysis as a source of frustration for these teens. On the other hand, YouTube offers a sheer volume of videos available for teens and employs mechanisms, such as autoplay, that make them feel less in control of the time they spend on the platform (Lukoff et al., 2021), which may result in their discussions about the long time they spent on this platform. While prior research has mainly focused on studying these affordances to understand the needs that would motivate people to use specific social media sites (Karahanna et al., 2018), our statistical analyses comparing the platforms based on teens’ negative experiences discussions highlight the importance of adopting an affordances-based approach to address teens’ negative experiences. Given the fast and evolving landscape of social media platforms, adopting this approach become of great importance for future researchers to provide a more nuanced understanding of how specific affordances contribute to the negative experiences of teens and/or other populations. Below, we abstract some of the key social media affordances that shaped teens’ social media experiences, as described in our results:

• •

  Visibility of Status: When status visibility was exposed, this often led to teens feeling cheated, overlooked, and excluded, especially when someone’s words (e.g., “I am going to bed.”) did not align with their online activity (e.g., Green - “Active Online”)

• •

  Public Content Sharing: Social media platforms that facilitated the public sharing of content, particularly image-based content (e.g., Instagram) amplified opportunities for cyberbullying, which was often perpetrated through private channels.

• •

  Image-based Sharing: Platforms that emphasized image-based sharing (Instagram) and selfies are more susceptible to negative experiences, such as body shaming.

• •

  Attention-seeking or Status-signalling Affordances: Attention or status-seeking affordances, such as follows, views, and likes, made teens feel undervalued or unpopular when their counts were lower than their peers or than their expectations.

• •

  Self-Promotion versus Interpersonal Connection Affordances: The affordances of some social media platforms (e.g., Instagram, YourTube, Twitter) were better suited for teen content creators who wanted to self-promote and create their own personal brands, while other platforms were better suited for forming interpersonal connections (e.g., Snapchat, WhatsApp). Differing privacy features afforded users to manage the audiences that they would like to interact with on each platform.

• •

  Connecting Publicly with Like-minded Peers: Social media platforms that afford social features (e.g., “Follow” and/or “Hashtag” on Twitter) facilitated users with building social connections by following each other and forming groups/communities of users with similar interests.

• •

  Connecting Privately with Strangers: When social media platforms allowed teens to connect with strangers privately (e.g., WhatsApp), this often led to the formation of regrettable relationships.

• •

  Content Filters: While most platforms had filters for inappropriate content, teens expected them to work better and expressed frustration when the content in their feeds (e.g., nudity) broke with this expectation. In contrast, when social media platforms afforded teens to create personalized content filters for what they wanted to consume versus avoid (e.g., sensitivity filter on Twitter), this contributed to more positive experiences for teens by lowering the chances of being exposed to unwanted content.

• •

  Sharenting: Some platforms (e.g., Facebook), likely due to real name affordances and the user audience, facilitated practices such as “sharenting,” parental sharing of content related to their teens  (Kumar, 2021; Moser et al., 2017), which led to interpersonal privacy violations.

• •

  Ephemerality of Content: Social media platforms that provided features where users could set how long their networks could view the content (e.g., Snapchat) had the potential of lowering teens’ pressure to self-present and impression management, which has also been suggested by prior works  (Piwek and Joinson, 2016; Bayer et al., 2016).

• •

  Curated Memories and Flashbacks: Similar to the findings of prior work, social media platforms that curated past memories (e.g., Snapchat) had the risk of retraumatizing youth based on negative past experiences (Scott et al., 2023), particularly for those who experience major changes in their lives (Corvite et al., 2022).

• •

  Autoplay Content and Algorithmic Recommendations: Social media platforms that used algorithms to recommend content and autoplay content (e.g., YouTube) had the potential of facilitating addictive patterns of consumption and time management struggles, which has been confirmed by prior research (Lukoff et al., 2021).

• •

  Other Engagement Features: Features that promoted engagement (e.g., Snapchat streaks) served as a double-edged sword of promoting motivation and distraction from life’s problems but also addictive behaviors.

In summary, we urge future research to adopt our affordances perspective when attempting to understand the differential patterns and consequences of social media use on youth. By taking this broader perspective, we can make a marked shift in our public discourse, research, design practices, as well as policies aimed at safeguarding youth from the detrimental effects of social media by moving towards a “safety by design” (des Hommes, 2022; Agha et al., 2023a; Perrino, 2022) perspective that proactively considers how to shape the design-based affordances of social media in a way that is protective of youth without restricting or over-policing their online experience.

Our study provides insight into the ways how teens discuss different social media platforms when interacting with other teens on online peer support tools, suggesting features and mechanisms for adolescents’ online safety tools that would be needed to help them stay safe on social media.

There are several limitations of our work that inform future research directions. First, we were not the first to uncover many of the themes that we identified from our analyses, as issues like cyberbullying, body-shaming, and privacy violations have been well-documented by prior research (Kwan et al., 2020; Reddy et al., 2022; De Wolf, 2020). However, our research is one of the first to take an affordances perspective to better understand youths’ differentiated social media experiences across different platforms. As such, our work serves to both confirm many of the findings from previous work and build upon and add nuance to those findings, as well as shift the conversation towards a more practical design-based approach to proactively address some of the problems youth encounter when using different social media platforms. Second, our sample was skewed toward older adolescents (ages 15-17) who identified as males, females, or who did not specify their gender. Furthermore, these teens were active users on an online platform dedicated to peer and mental health support. Therefore, our results may not be generalizable to the whole teen population of different sexual orientations, ages, and those not actively seeking peer support pseudo-anonymously online. Future work needs to triangulate our results across broader and more diverse demographics of youth to ensure that our results hold. Additionally, we caution readers to not assume that a higher percentage of negative posts provides evidence that the drawbacks of social media use outweigh the benefits. Prior research has confirmed negative bias in online reviews (Qahri-Saremi and Montazemi, 2022). Therefore, we assume that youth were more likely to share their negative experiences than the good ones.

Another key limitation of our work is that we did not analyze the support and advice teens received from others based on comments made on their posts. Such analyses would be helpful in understanding potential recommendations for coping the negative experiences, amplifying positive ones, and generally the helpfulness of such advice. Since 10% of posts were removed before coding because they were duplicates, this suggests that these teens may have received fewer responses than they sought. Finally, our analyses were constrained by the time period available for analyses (2019 and 2020) as the social media landscape has since changed. Yet, a key strength of our work is in demonstrating for the purpose of future research the importance of examining the lived experience of teens across different social media platforms and doing so based on digital trace data in places where they are actively seeking support based on these experiences.