Rolf Degen summarizing... Large-scale, long-term study that allows tracking of individual developments gives the all-clear as regards the dangers of intensive social media use for adolescents

The complex association between social media use intensity and adolescent wellbeing: A longitudinal investigation of five factors that May affect the association. Maartje Boer et al. Computers in Human Behavior, October 28 2021, 107084. https://doi.org/10.1016/j.chb.2021.107084

Highlights

• On average, within-person changes in SMU intensity and wellbeing were not related.

• Within-person relations between SMU and wellbeing varied across adolescents.

• At the between-person level, more SMU was somewhat related to less wellbeing.

• Between-person relations between SMU and wellbeing were confounded by SMU problems.

• Active and passive SMU did not yield differential associations with wellbeing.

Abstract: The present study examined five possible explanations for the mixed findings on the association between adolescents' social media use (SMU) intensity and wellbeing. Particularly, it investigated whether the association between SMU intensity and life satisfaction depended on (1) the type of SMU activity the adolescent engaged in, (2) the (non)linearity of the association, (3) individual differences, (4) inclusion of SMU problems, and (5) the level of analysis. Data from four waves of longitudinal data among 1419 adolescents were used (Mage(T1) = 12.51 (0.60), 45.95% girl). Multilevel analyses showed that at the within-person level, on average, changes in different types of SMU activities were not associated with changes in life satisfaction. Within individuals, the associations ranged from negative to positive across adolescents. In general, this variation could not be explained by adolescents' engagement in upward social comparisons. At the between-person level, the higher adolescents' average intensity of certain SMU activities, the lower their average level of life satisfaction. However, these associations were confounded by adolescents’ SMU problems. No curvilinear associations were found. Overall, the findings underline that to enhance our understanding of the association between SMU and wellbeing in adolescence, it is important to acknowledge the heterogeneity of effects, distinguish between SMU intensity and SMU problems, and disentangle within-from between-person effects.

Keywords: Social media useWellbeingLife satisfactionAdolescentsLongitudinal study

4. Discussion

The present study investigated the extent to which the association between SMU intensity and wellbeing is dependent on (1) the SMU activity adolescents engage in, (2) the (non)linearity of the association, (3) individual differences, (4) whether SMU problems are considered, and (5) the level of analyses. In doing so, we distinguished SMU activities ranging from more active (i.e., SNS posting, IM sending, SNS responding, SNS liking) to more passive (i.e., SNS viewing, IM viewing). Wellbeing was indicated by life satisfaction. At the within-person level, there was no average association between any of the SMU activities and life satisfaction, regardless of whether we controlled for SMU problems. However, the associations at the within-person level varied: For some adolescents, increases in SMU activities were associated with decreases in life satisfaction, whereas for others, increases in SMU activities were associated with increases in life satisfaction. In general, this variation could not be explained by adolescents' tendency to engage in upward social comparisons. At the between-person level, higher average intensity of some more passive activities (i.e., SNS and IM viewing) and one more active (i.e., IM sending) activity were associated with lower average life satisfaction with a small effect size. However, these associations disappeared when controlling for adolescents’ average level of SMU problems. In addition, for none of the SMU activities, evidence was found that the association between SMU intensity and life satisfaction was curvilinear.

Our findings highlight the importance of three factors for understanding the association between SMU activities and wellbeing in adolescence. First, answering the question whether the association between SMU intensity and wellbeing differs across adolescents (RQ3a), our findings showed that within-person effects of SMU intensity ranged from positive to negative across adolescents. This result is in line with experience sampling studies showing that for some adolescents, momentary increases in the intensity of SMU activities were associated with momentary decreases in wellbeing, but for others with increases or no changes in wellbeing (Beyens, Pouwels, Valkenburg, & Van Driel, 2020; Beyens, Pouwels, Van Driel et al., 2020). This study extends these findings as it revealed that also with annual assessments, associations between adolescents’ intensity of SMU activities and life satisfaction varied across adolescents.

Second, answering the question whether a negative association between SMU intensity and wellbeing is driven by SMU problems (RQ4), our findings indicated that negative between-person associations between certain SMU activities and life satisfaction disappeared when controlling for SMU problems. These findings suggest that a negative association between SMU intensity and life satisfaction may be explained by the presence of SMU problems rather than by engagement in specific SMU activities. Therefore, negative associations between SMU intensity and wellbeing revealed in previous studies may have been driven by unobserved SMU problems (e.g., Kelly et al., 2018; Twenge et al., 2018). However, even after controlling for SMU problems, we found that the within-person associations between the SMU activities and life satisfaction ranged from negative to positive. Hence, for some adolescents, increases in SMU activities were associated with decreases in life satisfaction, which could not be attributed to increases in SMU problems.

Third, related to our question at which level a negative association between SMU intensity and wellbeing occurs (RQ5), we found no average associations at the within-person level, while there were negative associations at the between-person level (although only when not controlling for SMU problems). This finding demonstrates that between-level associations do not necessarily reflect within-person dynamics, which was also found in earlier longitudinal studies (Beeres et al., 2020; Coyne et al., 2020; Orben et al., 2019). Conceptually, this finding suggests that the observed between-person association between higher SMU intensity and lower wellbeing was not a causal relation, as changes in SMU intensity were not related to changes in wellbeing within an adolescent.

Above all, some of the factors affecting the association between SMU intensity and life satisfaction need to be considered in concert when understanding this association. As noted above, SMU problems confound the association between certain SMU activities and life satisfaction, but only with regards to between-person associations.

We also examined which type of SMU activity could be detrimental to wellbeing (RQ1). At the within-person level, we found no average associations between any of the SMU activities and life satisfaction, which aligns with findings from experience sampling studies (Beyens, Pouwels, Van Driel et al., 2020; Jensen et al., 2019). At the between-person level, the observed negative associations between adolescents' intensity of engaging in SMU activities and life satisfaction were not specific to passive SMU activities, as proposed by researchers (Liu et al., 2019; Verduyn et al., 2017). In line with our findings, other studies also showed that adolescents’ active as well as passive SMU activities were negatively correlated with their wellbeing at the between-person level (Beyens, Pouwels, Van Driel et al., 2020). Passive and active SMU activities are possibly difficult to disentangle, because adolescents often engage in such SMU activities simultaneously (Valkenburg, Van Driel, & Beyens, 2021). For example, responding to a message on an IM requires viewing that message first. Accordingly, our study showed very high correlations between IM sending and IM viewing at the between-person level. As such, their differential associations with wellbeing may be difficult to grasp, which may explain why in our study IM sending and IM viewing were both negatively related to life satisfaction. However, we stress that these negative associations disappeared when we controlled for SMU problems.

Based on the Goldilocks hypothesis (Przybylski & Weinstein, 2017), we also investigated whether the association between SMU intensity and wellbeing was nonlinear (RQ2), which was not confirmed in our study. Findings of the present study are thereby consistent with other longitudinal studies that did not find curvilinear associations (Houghton et al., 2018; Jensen et al., 2019). Curvilinear associations were mainly found in cross-sectional studies (Przybylski & Weinstein, 2017; Twenge et al., 2018), which could imply that the Goldilocks hypothesis applies to associations at the between-person level at one particular timepoint. Alternatively, earlier found curvilinear associations may have been country-specific. International research shows that the association between adolescents’ SMU and wellbeing are susceptible to country-level factors, for example the extent to which social media are adopted among youth within society (Boer et al., 2020).

Further, we examined whether the association between adolescents’ SMU intensity and wellbeing would depend on the tendency to engage in upward social comparisons (RQ3b). We found no evidence for this moderating effect, with one exception: Among adolescents reporting high levels of upward social comparison, increases in SNS liking were associated with decreases in life satisfaction, which supports the social comparison perspective (De Vries et al., 2018). Among adolescents reporting low levels of upward social comparison, increases in SNS liking were associated with increases in life satisfaction, which corresponds to the emotional contagion perspective (De Vries et al., 2018). However, the individual differences in the associations between SNS liking and life satisfaction were not reduced when upward social comparisons were considered. Also, this was the only moderating effect found out of the six SMU activities that were examined. Therefore, future studies are necessary to replicate our findings.

Our findings provide several implications for future research on the association between SMU intensity and adolescent wellbeing. Specifically, future longitudinal studies that acknowledge heterogeneity in effects, consider SMU problems, and distinguish between within-person and between-person effects would be promising. Research considering these three factors seems more informative than research aiming to disentangle the effects of different SMU activities or examining curvilinear associations. Furthermore, our findings illuminate why earlier studies on the link between SMU intensity and adolescent wellbeing are so inconsistent: Depending on whether researchers investigate specific groups of adolescents, control for SMU problems when examining SMU intensity, or study within-person or between-person associations, the link can range from positive to negative.

In addition, our findings can also inform those concerned with the wellbeing of adolescents, including parents and teachers. They suggest that most adolescents engaging in higher SMU intensity are not at risk for impairments in wellbeing, regardless of whether this concerns engaging in more active or more passive SMU activities. Higher SMU intensity may be considered as normative adolescent behavior that contributes to adolescents’ individual development and daily interaction with peers (Granic, Morita, & Scholten, 2020; Valkenburg & Peter, 2011). Nevertheless, our findings imply that risks to wellbeing could arise when adolescents report SMU problems, indicated by symptoms of addiction (e.g., loss of control over SMU). Therefore, investing in the prevention, early detection, and treatment of problematic SMU may be warranted. Yet, our findings also showed that for a particular group of adolescents, increases in SMU intensity are indicative of decreased wellbeing. Research focusing on identifying the individual characteristics that make adolescents vulnerable to negative SMU effects could provide directions for targeted prevention or intervention programs.

Although we tested many ways in which adolescents’ SMU and their wellbeing could be related, the association may be dependent on more factors that were not addressed in this study. First, it may depend on whom adolescents have contact with on social media. For example, longitudinal research on adults showed that receiving Facebook messages from close friends increased wellbeing, whereas receiving such messages from acquaintances did not change wellbeing (Burke & Kraut, 2016). Other research showed that adolescents who reported more Instagram use with close friends reported more friendship closeness than adolescents who showed less Instagram use with close friends (Pouwels, Valkenburg, Beyens, Van Driel, & Keijsers, 2021). This association was not observed with regards to Instagram use without close friends (Pouwels et al., 2021). Second, the association may depend on the wellbeing outcome being studied. Meta-analytic findings indicate that SMU intensity has different associations with self-esteem and social capital than with life satisfaction (Meier & Reinecke, 2020). Furthermore, research suggests that the association is different for positive indicators of wellbeing than for negative indicators, for example depression and negative affect (Huang, 2017; Wirtz, Tucker, Briggs, & Schoemann, 2020). Third, the association may be contingent on the social media platform used. More specifically, the use of highly visual social media, such as Instagram and Snapchat, may induce more impact than less visual social media, such as Facebook and Twitter. Highly visual social media are mainly focused on uploading visual content, including photos and videos, and allow users to edit this content in more appealing ways using filters (McCrory, Best, & Maddock, 2020). Exposure to modified idealized online portrayals may reinforce a negative body image, which, in turn, could undermine wellbeing (Marengo, Longobardi, Fabris, & Settanni, 2018).

4.1. Strengths and limitations

Using four waves of longitudinal data among secondary school adolescents and a systematic multilevel analytical approach, the present study examined five factors that possibly affect the association between SMU intensity and wellbeing. However, results of this study should be interpreted while considering several limitations. The yearly time intervals of the data used in the present study only allowed for estimating long-term associations. Consequently, potential short-term effects of the intensity of SMU activities could not be captured. Yet, findings from studies on the association between different SMU activities and wellbeing using (multiple) daily assessments showed some comparable results. Often, these studies also observed no average within-person relation between passive and active SMU activities and wellbeing. Also, they showed that these within-person associations ranged from negative to positive across adolescents (Beyens, Pouwels, Valkenburg, & Van Driel, 2020; Beyens, Pouwels, Van Driel et al., 2020; Jensen et al., 2019). Additionally, self-report measures of adolescents’ SMU intensity may not accurately represent actual use, because adolescents may over- or underestimate their use. Indeed, research showed a moderate correlation between self-report and tracked SMU (Parry et al., 2020). Research replicating our study using tracked data of SMU activities is warranted. In addition, the present analyses did not explore the direction of the associations between the intensity of SMU activities and life satisfaction. Studying directionality would require a different analytical approach (e.g., random intercept cross-lagged panel modelling), which cannot be adopted within the present multilevel framework. Although we examined life satisfaction as an outcome of higher SMU intensity, a reverse order may be plausible as well. A meta-analysis on the direction of the association supports our assumption, although it investigated the direction of the relation between screen time in general and depression symptoms (Tang, Werner-Seidler, Torok, Mackinnon, & Christensen, 2021). Finally, the data included considerable dropout of adolescents, which may have affected the quality of the data, especially in the final wave. However, this dropout was mostly not due to individual refusal (i.e., not due to selective dropout), but to classes and schools dropping out. Also, we aimed to limit any potential bias by imputing missing data based on available data at all waves (Madley-Dowd et al., 2019).