The data come from the BrainWaves study [19]. BrainWaves is a longitudinal cohort study that collects data on UK students aged 16 years and older. Three waves have been collected to date (wave 1: April 2024; wave 2: November 2024; wave 3: November 2025), with data collection ongoing. BrainWaves includes data on demographics and a broad collection of variables, including mental health, well-being, lifestyle, cognition, physical health, adversity, and relationships.

The sample used for this study was drawn from wave 3, when social media use motives were assessed, and included 11,498 participants. Participants with complete data on all social media use motive items were retained for the primary analyses. Analyses incorporating gender and deprivation were restricted to participants with available data on these covariates.

Ethics approval for the BrainWaves study was granted by the Central University Research Ethics Committee (R91383/RE001; amendment R91383/RE004). The study was conducted in accordance with the Declaration of Helsinki, and all participants provided written informed consent prior to participation. Participant data were stored and accessed within the BrainWaves Informatics Hub, which operates within the Dementias Platform UK Data Portal Trusted Research Environment, with controlled access procedures in place to protect participant privacy and confidentiality. Participants did not receive compensation for participation.

Exploratory factor analysis (EFA) was used to identify latent dimensions of social media use motives. We applied Geomin rotation with tetrachoric correlations and weighted least squares mean and variance adjusted (WLSMV) estimator to account for binary indicators. Model selection was guided by model fit indexes, parsimony, and factor interpretability [24]. Specifically, factors should be composed of at least 3 items, with loading thresholds of 0.40 or higher [25,26]. Items with substantial cross-loadings or low primary loadings were removed. The full sample size was used for the EFA, but the results (which remain unchanged) from a half-random split of the sample are also reported in Multimedia Appendix 1.

Confirmatory factor analysis (CFA) was used to estimate individual-level factor scores based on the selected factors from EFA. Again, the full sample size was used, and results from the other random split are reported in Multimedia Appendix 1. The models were estimated using the WLSMV estimator to accommodate the binary indicators [27]. Model fit was assessed using the comparative fit index (CFI; ≥0.95), root mean squared error of approximation (RMSEA; <0.06), and standardized root mean squared residual (SRMR; <0.08) [28].

To aid interpretability and visualize associations, simple correlations were visualized using linear plots between social media use motive factor scores and anxiety and depressive symptoms. Linear regression models were subsequently used to estimate the strength of these associations while adjusting for gender and deprivation. Interaction terms were included to examine whether associations between social media use motives and mental health outcomes varied by gender or deprivation.

SEM was used to simultaneously estimate the relationships between latent social media use constructs and multiple dimensions of mental health outcomes within a unified framework. In this model, the latent variables of social media use motives were predictors of anxiety, depression, well-being, and resilience. All error terms of the outcomes were allowed to covary, and the latent factors were also allowed to correlate. In addition to the fitted model, a second model including covariates (as predictors of social media use motives and anxiety, depression, well-being, and resilience) was also fitted and reported in Multimedia Appendix 1. Results from the split sample can be found in Multimedia Appendix 1.

The models were estimated using the WLSMV estimator to accommodate the mix of binary and continuous indicators [27]. The same model fit indexes as in the CFA were used (ie, CFI, RMSEA, and SRMR). Missing data on endogenous variables were accommodated using available data under WLSMV estimation. As Mplus treats observed exogenous covariates as fixed variables, participants with missing gender or deprivation data were excluded from models containing these covariates.

Of the 11,498 wave 3 participants, 6820 had complete data on all social media use motive items and were included in the primary analyses. Participants ranged from 16 to 20 (mean 16.72, SD 0.74) years, with only 4 participants aged 20 years; the sample therefore predominantly represented older adolescents. Analyses incorporating gender and deprivation included 5540 participants with available covariate data. Among participants with available deprivation data, 88.3% (5114/5795) reported none of the 3 deprivation indicators, signaling a highly skewed distribution (Table 1).

Initially, a 3-factor solution demonstrated the best fit but was not retained because the third factor was defined by only 2 items (“making money” and “seeking help”), limiting its interpretability and robustness (Table S1 in Multimedia Appendix 1). Accordingly, a 2-factor solution was selected as the most parsimonious and conceptually meaningful representation of the data (Table 2). The 2-factor solution distinguished between avoidant-escapist and social-recreational motives for social media use.

The boredom item loaded primarily on the avoidant-escapist factor (loading=0.388) but fell marginally below our prespecified loading threshold of 0.40. We retained the item because it was conceptually consistent with the avoidant-escapist construct and showed minimal cross-loading on the social-recreational factor (loading=0.062). As this represented a deviation from our prespecified criterion, sensitivity analyses were conducted excluding boredom; the 2-factor structure and CFA model fit remained substantively unchanged (CFI=0.978; RMSEA=0.046; SRMR=0.039; Table S5 and Figure S4 in Multimedia Appendix 1).

The final avoidant-escapist factor comprised 4 items reflecting disengagement and avoidance processes (“boredom,” “addictive use,” “distraction from problems,” and “procrastination”), while the social-recreational factor comprised 4 items reflecting social engagement and recreational use (“entertainment,” “socialising,” “communication with friends and family,” and “observing others”). All retained items loaded onto their respective factors (Table 2).

Results from CFA based on the 2 selected factors from EFA (ie, avoidant-escapist and social-recreational) also showed good fit (N=6820; CFI=0.977; RMSEA=0.041; SRMR=0.038). The correlation between the 2 latent factors was r=0.519 (P<.001; Figure S1 in Multimedia Appendix 1). Factor scores derived from CFA were used in subsequent visualization and regression analyses.

Figure 1 illustrates the unadjusted associations between social media use motive factor scores and mental health outcomes. Avoidant-escapist social media use showed clear positive associations with both depressive and anxiety symptoms. In contrast, associations between social-recreational use and mental health outcomes were positive but comparatively weaker in the unadjusted plots.

Linear regression analyses adjusting for gender, deprivation, and shared variance between dimensions demonstrated that avoidant-escapist social media use was significantly associated with greater anxiety and depressive symptoms, whereas social-recreational use was associated with lower levels of both outcomes (Figure 1 and Table 3). Associations were stronger for avoidant-escapist compared to social-recreational use. Being male was associated with lower anxiety and depressive symptoms relative to being female, and higher levels of deprivation were also associated with higher anxiety and depressive symptoms. No evidence of interaction between gender or deprivation and social media use motives was observed (P>.05 in all cases).

Our study used data from a large-scale cohort of older adolescents in the United Kingdom to examine associations between social media use and mental health. By applying EFA, CFA, and SEM, we identified underlying latent dimensions of social media use motives, accounting for measurement error, and estimated multiple relationships simultaneously within a unified framework. This approach goes beyond traditional methods that rely on individual items or summed scores, allowing for a more nuanced understanding of how distinct patterns of social media engagement relate to mental health outcomes.

On the basis of our EFA, we identified 2 underlying dimensions of social media use across 8 items: avoidant-escapist and social-recreational motives. These dimensions broadly align with findings from prior attempts to classify motivations of social media use [29,30]. For example, previous reviews have identified categories of motives related to relationship maintenance, entertainment, information seeking, and emotional regulation [31]. However, our findings suggest that motives that are grouped together under broader categories (eg, entertainment and escapism considered as positive sensation seeking) [9] may instead reflect psychologically distinct patterns of engagement with different associations with mental health. As such, applying a data-driven approach to categorizing social media use motives may provide a more informative framework to understanding its relationship with adolescent well-being.

Across both linear and structural equation models, we found that avoidant-escapist motives were associated with poorer mental health (ie, higher anxiety and depressive symptoms and lower well-being and resilience). Similar patterns were found in a study where escapist motivations for social media use were linked with higher self-reported stress [32]. Similarly, nonsocial motivations of social media use were associated with internalizing psychopathological symptoms in a Spanish population [12]. Furthermore, the literature has shown that digital use driven by procrastination and addiction is associated with poorer mental health [33]. These trends are likely explained by unfavorable social comparisons [34], worsening symptoms for those with existing mental health conditions [35], and disruptions in daily activities or social life due to excessive or uncontrolled engagement with social media [36]. Social media platforms are inherently rewarding environments that provide entertainment, distraction, and social reinforcement, which may help explain why adolescents continue engaging with them despite potential negative consequences [37].

In contrast, social-recreational motives were associated with better mental health outcomes (ie, lower anxiety and depressive symptoms and higher well-being and resilience). These results are consistent with literature reporting positive benefits of social media use in social contexts [38]. Often, social media is used to share personal experiences, keep up with friends, and give or receive support to others [39]. As such, it appears that social connections developed and fostered through social media play an important role in promoting young people’s well-being [40]. For example, time spent playing video games has been shown to be one of the most effective ways for adolescents to generate positive feelings due to its social and emotional benefits [41]. However, it is important to consider that such motives are not inherently positive; for example, entertainment could be linked to problematic smartphone use [42]. Specifically, social media provides positive reinforcement for continued use, which increases engagement and may lead to less controlled use and addiction.

In terms of gender differences, we found that female individuals reported more of both social-recreational and avoidant-escapist social media use and also exhibited higher anxiety and depressive symptoms relative to male individuals. In contrast, male individuals reported higher overall well-being and resilience relative to female individuals. These results may be explained by gender differences in the coping strategies used that influence digital use and lead to different outcomes in problematic social media use [43]. Our results are consistent with those of previous studies that report gender differences in social media use [44,45]. For example, prior work has shown that girls primarily use social media to reinforce preexisting friendships, while boys mainly use these sites to flirt and make new friends [4,15]. Interestingly, we found no significant interactions between gender and social media use motives on mental health. This suggests that the associations between social media use motives and mental health outcomes are broadly consistent across genders despite differences in overall levels of use and symptom severity.

We found that higher levels of deprivation were associated with higher levels of avoidant-escapist use and greater anxiety and depressive symptoms. These results are consistent with those of a study that found greater negative impacts of social media use on adolescents with low socioeconomic status compared to adolescents with high socioeconomic status [46]. We also observed that low levels of deprivation were associated with more social-recreational motives and higher well-being and resilience, consistent with the findings of Kurten et al [47]. However, Kurten et al [47] also found that materially deprived adolescents (ie, those who lack resources for material participation in mainstream society) tend to spend more time on social media overall compared to those less deprived. Our findings therefore suggest that preexisting vulnerabilities such as deprivation may be related to motivations for social media use and their associations with mental health. However, we note that these covariate-related findings should be interpreted in light of the reduced analytic sample as participants excluded due to missing covariate data had poorer mental health on average than those retained (Table S7 in Multimedia Appendix 1).

Several limitations should be considered when interpreting these findings. First, the cross-sectional design precludes causal inference, and the observed associations may be bidirectional. For example, adolescents with preexisting psychological vulnerabilities may be more likely to engage with social media for avoidant or escapist reasons [35], while repeated engagement in these patterns could reinforce avoidance-based coping over time [48]. Longitudinal and experimental research is therefore needed to establish temporal relationships between social media use motives and mental health.

Second, the sample predominantly comprised older adolescents aged 16 to 19 years, with only 4 participants aged 20 years and no participants younger than 16 years. The findings should therefore not be assumed to generalize to younger adolescents, for whom patterns and functions of social media use may differ.

Third, inclusion of gender and deprivation in the covariate-adjusted structural equation model reduced the analytic sample from 6820 to 5540 because participants with missing predictor data were excluded. Participants excluded from this model reported higher anxiety and depressive symptoms and lower well-being and resilience on average than those retained. This pattern indicates potential selection bias, and estimates from the covariate-adjusted model may therefore underrepresent individuals experiencing poorer mental health. In addition, deprivation was highly zero-inflated, with 88.3% (5114/5795) of participants with available data reporting none of the 3 deprivation indicators. As such, associations involving deprivation should be interpreted cautiously.

Finally, social media use was assessed using self-report and may therefore be subject to reporting bias. Previous research suggests that discrepancies between self-reported and objectively measured social media use may vary according to social media platform and respondent gender [49].

Longitudinal and experimental studies are needed to investigate temporal directionality and causal mechanisms. Furthermore, it is important to examine underrepresented groups (nonbinary and transgender individuals and ethnic minority groups) to better understand and serve these populations. Understanding these associations and their dynamics is necessary to make the most of the benefits of social media while preventing risks and harms.

The results of our study suggest that social media use motives among older adolescents can be characterized by 2 broad dimensions (avoidant-escapist and social-recreational), which are differently associated with mental health. Our findings emphasize the importance of considering how and why young people use digital tools to better understand their relationship with mental health. Notably, social media use in youth is not uniformly harmful or beneficial, and its associations with mental health depend on underlying motives. Further research in this area is critical, especially as many institutions and governments worldwide are considering outright social media bans in schools and for young people [50]. Although often linked with adverse outcomes, social media use is nuanced and often offers benefits to its users. As such, total bans may not achieve the desired goals and may even lead to iatrogenic outcomes [51]. Interventions and guidelines should therefore move beyond screen time and consider patterns of engagement when assessing risk or benefit. Longitudinal and experimental research, particularly in diverse and underrepresented groups, is needed to better understand associations between digital use and mental health in younger populations.