Does gender matter? The association between different digital media activities and adolescent well-being

Adolescents of today spend an increasing portion of their free time using different digital media, such as smartphones, computers, and tablets [1]. In Sweden, the proportion of 14-year old adolescents using the internet for more than 3 h/day increased from 30% in 2010 to 76% in 2020 [2], and in the US, 45% of teens report that they are online on a “near-constant” basis [3]. At the same time, the rate of depression, anxiety, and suicide among adolescents has risen to historically high levels over the past 10–15 years [4‐6]. For example, the level of depression in 10–17-old Swedish girls has surged from approximately 500 to more than 1.000 per 100.000 from 2008 to 2018 [7]. Although the reasons for this increase in depression rate are likely to be multifaceted, including changed diagnostic criteria, the use of digital media has attracted increasing attention as a potential causal factor [8].

The association between digital media use (e.g., playing games, social media use, watching TV) or social media use more specifically (e.g., different activities on Instagram, Facebook or Snapchat) and well-being has been examined (and debated) [9, 10] in several studies with mixed results [11‐13]. Whereas the majority of studies – many of which have been based on cross-sectional designs – have found a negative association [8, 11], the few available longitudinal studies have shown either that social media use predicts lower well-being in terms of depression and internalizing symptoms (e.g., depression, anxiety, hypersensitivity, headache, worry) [14‐17], no association [18‐20], or that depression predicts an increase in social media use [21, 22]. In addition, some investigators have found evidence of a curvilinear u-shaped association [23, 24]. Other research has examined potential mediators, and found that the association between social media use and well-being could be mediated by factors, such as cyberbullying, lack of sleep, and lower physical activity [15].

An intriguing possibility is that the inconsistent results in the existing literature are related to key methodological differences between studies, limiting the generalizability and conclusions that can be drawn from them [1]. To move beyond these limitations, we performed a large cross-sectional study and used in-depth measures of social media activities and internalizing symptoms, which constitute an important aspect of mental well-being. Our goal was to provide a deeper understanding of, not only where on the internet adolescents spend their time, but what kind of activities might contribute to changes in mental well-being. To clarify the specific contributions of our study, we begin by listing three central limitations in existing research.

A first limitation in past research that might contribute to the mixed findings is the lack of interest in specific digital media activities, and to what degree they are differently associated with well-being [18, 25, 26]. For example, studies that do not differentiate clearly between various digital media activities, have found both negative [27], and non-existent associations with well-being [18]. Internet use can include a wide range of activities that differ across gender in terms of agency, display mode, and social meaning. Therefore, it is important to examine different activities separately. A number of studies have found that playing games is associated with lower well-being in terms of internalizing problems and depressive symptoms [28‐31]. Importantly, social media use was found to be more strongly associated with well-being than playing games [1, 14]. One explanation of this difference could be that social media presents more opportunities for social comparison, which is known to decrease well-being, in comparison to gaming [1].

A second limitation in past research that might contribute to the mixed findings, is the varying measures used to assess social media use more specifically. For example, adolescents have been asked to define social media themselves [32], and self-report how active they have been on different platforms [20, 33], whether they chat or interact with their friends on different platforms [17, 34], and use a combined measure of instant messaging, photo-sharing or other social media activities on different platforms [22]. Different means of assessing social media might tap into distinct activities with varying relevance for the development and expression of well-being. We argue that it is important to go beyond existing approaches to examine both how and in what way different social media activities are associated with well-being in boys and girls separately.

To address these open questions, our study focuses on three different social media activities: (i) chatting in real-time with friends on platforms that are “private” (one-on one or private group chats); (ii) online sociability, i.e. communication taking place in public social media that can be viewed by others (not necessarily in real time); and (iii) self-presentation, such as selfies, film clips or other types of personal information that can be viewed by others. As far as we know, no previous study has examined how these three activities are associated with well-being for boys and girls. Self-presentation, which has been highlighted as a key factor underlying the association between social media use and adolescent well-being [35‐38], indicates an individual’s motivation to “brand” themselves and to reach digital social status through likes and new followers [39]. This search for social rewards lays the ground for social comparisons, known to affect well-being negatively [35, 40]. Indeed, recent meta-analyses [41, 42] demonstrate that social comparison in general, and upwards social comparison (i.e. comparison with superior other) in particular, predicts a decrease in well-being [35, 39, 43]. The lack of positive feedback (i.e. social rewards) and the presence of social punishment from the social media community has been shown to have a negative influence on the adolescent’s well-being [13, 44]. In light of these lines of research, we assume that a higher reliance on self-presentation renders the individual more vulnerable to negative feedback.

A third limitation in previous research is that previous studies have treated gender as a control variable, overlooking the possibility that the association between digital media use and well-being is different for boys and girls [14, 26]. For example, some researchers have analysed boys and girls together, founding that digital media use was not associated with well-being [20], whereas others have shown girls to have lower well-being as indicated by, for example, depressive symptoms [6, 37, 45], anxiety [6, 45] and internalizing problems [16]. Girls use social media [1, 2, 28, 46] and chat with friends [28] more than boys who in turn tend to spend more time gaming [1, 2, 28]. In addition, girls more frequently post different types of selfies, use filters, manipulate their photos, and delete post more frequently than boys, whereas boys update their profile more often on sport and technology [2, 37, 43, 47, 48]. Against this background we argue that it is important to examine the associations by gender.

The importance of including gender as a variable is supported by previous research finding social media use to be more strongly associated with well-being for girls than for boys [1, 17, 49, 50], whereas some did not find any gender differences [16]. Differences in the selfie culture may be one explanation that helps us to understand why the association between self-presentation and depressive symptoms is larger for girls [37]. Gaming has been found to be associated with well-being for both boys and girls [1]. These gender differences indicate that social media use, and different activities of social media may have a larger association with well-being for girls, and that the association between gaming and well-being might be rather similar for boys and girls.

To sum up, we identified three areas of limitations in previous research related to the importance of separately examining the role of (1) kinds of digital media activities, (2) different types of activities within social media use, and (3) gender. In our cross-sectional study, we aim to use various measures of digital media activities, well-developed assessments of social media activities, and examine how these activities are associated with internalizing symptoms as a measure of girls’ and boys’ well-being. In spite of the inherent limitations of cross-sectional designs, which are unsuitable to make strong causal claims, we believe that these methodological improvements will contribute to our understanding of the specific processes linking social media use and mental health problems and well-being in adolescents.

We test the following three hypotheses:

Table 2 presents differences between girls and boys in relation with internalizing symptoms, social media use, chatting, online sociability, self-presentation and playing games. We observe that girls report significantly higher on the internalizing symptoms than boys (t = 19.99, p = < .001). Moreover, girls score significantly higher on the measure of social media use (t = 11. 47, p = < .001), as well as on chatting (t = 5.14, p = < .001), online sociability (t = 9.49, p = < .001) and self-presentation (t = 11.01, p = < .001). Boys report significantly higher on the playing game measure (t = − 28.81, p = < .001). The effect sizes presented by the Cohen’s d range from .16 (for online sociability) to −.92 (for playing games).

In order to examine our key hypotheses, i.e. whether the different digital media activities are differently associated with internalizing symptoms for girls and boys we present a number of regression models in Table 3. The results presented in model 1 show that social media use and playing games are both positively associated with internalizing symptoms. According to the standardized regression coefficient, social media use is slightly more strongly associated with internalizing symptoms than playing games. The results also show that the squared terms of the two measures are significant, indicating a curvilinear association between both digital media activities and internalizing symptoms. Gender is negatively associated with internalizing symptoms, which means that girls report higher on internalizing symptoms, adjusting for the other predictors.

In the second model, where the interaction terms included, and the results show that the interaction between gender and social media use is strongly negatively associated with internalizing symptoms (B = −.191, p = < .001), meaning that the association between social media use is more pronounced for girls then for boys. The squared term of social media use is still associated with internalizing symptoms, indicating that the association is not linear it is curvilinear. Finally, the interaction between gender and playing games is not significantly associated with internalizing symptoms, i.e. there is no gender differences in the association between playing games in relation with internalizing symptoms.

To visualize the interaction, we plotted the predicted values of internalizing symptoms on social media use by gender, with their 95% CIs. Figure 1 show that the association between social media use and internalizing symptoms is only significant for girls but not for boys. The association is curvilinear indicating that the association between social media use and internalizing symptoms becomes more and more pronounced the more girls are using the social media.

Finally, we estimated the regression models using the three different dimensions of social media use – chatting, online sociability and self-presentation in the association with internalizing symptoms in Table 4. Of the three social media activities, were chatting and self-presentation positively associated with internalizing symptoms, whereas online sociability is negatively associated with internalizing symptoms, adjusting for the other factors presented in model 1. According to the standardized regression coefficients, the results shows that self-presentation is the activity that is most strongly associated with internalizing symptoms of the different digital activities. Further, the results in model 2, show that self-presentation interact with gender in the association with internalizing symptoms (B = −.459, p = < .001), indicating that self-presentation is significantly stronger associated with internalizing symptoms for girls than for boys. The results also show that the interaction between gender and online sociability is near significant (B = −.215, p = .057). This interaction indicates that online sociability is negatively associated with internalizing symptoms for boys.

To visualize the interaction between self-presentation and gender, we plotted the predicted values of internalizing symptoms. Figure 2 show that the association between self-presentation and internalizing symptoms is only significant for girls but not for boys.

The current study examined how different digital activities (social media use and playing games), and different types of social media use (chatting, online sociability and self-presentation) are associated with internalizing symptoms as a measure of well-being in adolescent boys and girls. Our primary finding was that social media use was strongly and positively associated with internalizing symptoms in girls. This finding dovetail with previous research highlighting the negative health effects of social media use that offers an arena for self-presentation and self-comparison. Interestingly, the association in our study showed a curvilinear association, suggesting that the association will become stronger as the use of social media increases. The different activities of social media use were differently associated with internalizing symptoms. On the one hand, we found both chatting and self-presentation to be positively associated with internalizing symptoms and demonstrated that self-presentation was more strongly associated with internalizing symptoms for girls. On the other hand, online sociability was negatively associated with internalizing symptoms and this pattern was more pronounced for boys. Finally, our results also indicated that playing games was positively associated with internalizing symptoms.

Our results are consistent with previous cross-sectional [28] and longitudinal [15] studies that found social media use to be more strongly associated with measures of well-being for girls than for boys. At the same time, our results contradict findings from other longitudinal studies that found no association between social media use and well-being [18‐20]. A limitation in some of these longitudinal studies is that they used gender as a control variable and not examined whether the association could be different for boys and girls [20, 22]. Another limitation of previous studies is that varying measures were used to assess social media use. For example, several researchers [22] combined a number of different social media activities within a combined measure of social media use, and others [18] used a measure of social media asking how much time the adolescents spend on social networking (e.g., Facebook) on a typical day.

A key finding in our study is that self-presentation is the social media activity associated with internalizing symptoms for girls. This result is in line with the argument that adolescents that present information about themselves, often in search of likes and followers, will be more susceptible to self-comparisons [35, 39‐42], which is known to lead to decrease in well-being [37, 41, 42]. Additionally, not receiving the social rewards one hoped for (e.g., likes and followers), and the presence of social punishers, might be linked to an experienced loss of self-worth, stress and social isolation, all of which are known features of internalizing symptoms [35, 39, 43]. Finally, very few studies have gone beyond establishing an association to trying to explain why there is an association between social media use and well-being [35].

Our study improved on several methodological shortcomings in previous research studying the relationship between social media use and well-being. In particular, we measured (1) both social media use and playing games, (2) different forms of social media use, including chatting, online sociability and self-presentation, which has not been used before in the study of adolescent’s well-being, and (3) examined the role of gender.

A few limitations also need to be acknowledged. First, our cross-sectional design constrained our ability to draw any causal conclusions. For example, social media might cause internalizing symptoms, but it is also possible that internalizing symptoms influence social media use and finally, the association can be bi-directional. Our results do, however, closely align with the growing recognition of the role of social media in negatively affecting psychological well-being [35, 41, 42]. A second weakness of our study is that the response options in our survey did not allow for the same resolution in terms of time (number of hours) spent with social media as several previous studies [1, 14], although a few previous studies have used similar response alternatives as ours [15, 33]. This indicates that we cannot determine the exact number of hours that the respondents spend on social medial on a daily basis. That being said, our results nonetheless indicate the importance for future studies to employ more detailed measures of time spend online. Finally, our gaming measure includes both gaming together with other people and gaming alone, making it impossible to draw conclusions about the role of socialization during gaming.

In conclusion and in line with our hypotheses, our study shows that different social media activities are differently associated with internalizing symptoms as a measure of adolescent’s well-being, and that activities entailing self-presentation stand out because of their negative association with well-being for girls, but not boys. Our results suggest that policy makers should be aware of that not all social media activities are created equal in terms of their implications for the health of adolescents, and that these implications might be gender-specific. Future research needs to continue to use increasingly differentiated measures of social media activities, and include both longitudinal and experimental designs to provide a better understanding of how these activities are causally linked to the health of the young.