Prospective relationships between appearance-based social media use and adolescent mental health

Abstract

Growing evidence links adolescent social media use to mental health difficulties, yet most research is cross-sectional, assumes all social media activities are equally harmful, and rarely examines multiple forms of psychopathology. This longitudinal study tested reciprocal relations between mental health symptoms (social anxiety, body dissatisfaction, depression) and appearance-based social media behaviours (upward appearance comparisons and appearance investment). Participants were 1942 adolescents (grades 7–11; W1 M_age = 13.97, SD = 1.24; W2 M_age = 14.51, SD = 1.19) from four Australian schools. Bidirectional prospective relationships emerged between upward comparisons and all three mental health symptoms. Furthermore, Wave 1 body dissatisfaction and social anxiety predicted greater Wave 2 appearance investment, and more Wave 1 appearance investment predicted increases in Wave 2 depressive symptoms. Findings highlight upward comparisons as both a predictor and consequence of poor mental health and suggest appearance investment may link body dissatisfaction and social anxiety to depression.

Keywords

Adolescent body image mental health social comparison social media

With a seemingly ubiquitous influence on the modern world, social media use offers an unprecedented opportunity for connecting with others and improving communal wellbeing (Allen et al., 2014; Valkenburg and Peter, 2011). Yet, given the speed of social media’s development and uptake, its possible effects on adolescents have been of increasing public and scientific concern (Bell et al., 2015; Haidt and Allen, 2020; Twenge and Campbell, 2019), particularly because this demographic pioneers adoption of new social technologies (Kemp, 2022). Despite the rapid expansion of research into this area, the field’s emphasis on cross-sectional studies has prevented clarifying the direction of many reported effects. In this way, there remains a dearth of longitudinal research examining the role of social media use in either bolstering or worsening adolescent mental health (Orben, 2020; Orben and Przybylski, 2019; Schønning et al., 2020). This study focused on potential mechanisms theorised to underpin worsened mental health outcomes in adolescent social media users – namely, their appearance-based social media behaviours.

During adolescence (i.e. 13–18 years), individuals experience crucial growth in their physical, cognitive, emotional, and social development (Guyer et al., 2016; Rapee et al., 2019; Somerville, 2013). Although this typically leads to positive outcomes, adolescence also confers increased risk of certain mental health difficulties or “social-emotional disorders” (Rapee et al., 2019). Notably, the prevalence and incidence of levels of depression (Costello et al., 2011; Lawrence et al., 2016; Maughan et al., 2013), social anxiety (de Lijster et al., 2017; Kessler et al., 2005), and body dissatisfaction (Gowers and Shore, 2001; McLaughlin et al., 2015) all increase during adolescence.

While there are innumerable processes that inform normative and aberrant psychosocial development during adolescence (for review, see Rapee et al., 2019), two are of particular relevance to understandings of social media use. Not only is adolescence characterised by the growing salience and intensity of peer relationships (Brown and Larson, 2009; Crone and Dahl, 2012; Nesi et al., 2018), but it also precipitates markedly increased efforts to inform one’s own self-concept (Sebastian et al., 2008). This can entail experience-seeking, social feedback, internalising sociocultural standards of an “ideal” self (Thompson and Heinberg, 1999), predicting how one is perceived by others, and engaging in social comparison with others (Festinger, 1954). Social comparison is intensified during adolescence, given the aforementioned increase in relationship salience (Rapee et al., 2019), and is demarcated directionally. For example, upward social comparisons entail perceiving others as “better” than oneself (e.g. based on appearance, intellect, popularity). The Tripartite Influence Model of body image proposes that individuals experience tension between internalised sociocultural beauty ideals (e.g. thinness, muscularity) and their actual appearance via social comparison, which subsequently triggers and exacerbates body dissatisfaction and poor mental wellbeing (e.g. Lawler and Nixon, 2011; Rodgers et al., 2015; Thompson and Heinberg, 1999). Conversely, multiple theories of psychopathology within the broader cognitive behavioural therapy (CBT) framework (Beck, 2002; Kircanski et al., 2012; LeMoult and Gotlib, 2019) stress the importance of negative cognitive biases (e.g. upward social comparisons) and avoidance of threat (even manifesting behaviourally via appearance investment; Shafran et al., 2004). These factors are thought to perpetuate and predispose individuals to developing negatively biased perceptions of the self, a hallmark feature of social emotional disorders (Rapee et al., 2019). Therefore, clinical theories suggest a reciprocal association between appearance-based activities and social-emotional disorders.

In line with communication theories (Valkenburg and Peter, 2013), social media facilitates greater opportunities for selective self-presentation (i.e. management of image, appearance, or online identity via curation and editing; Walther, 2007). Given that physical attractiveness and peer evaluation are particularly salient during adolescence (Rapee et al., 2019), many adolescents may feel motivated to present and maintain an idealised version of their appearance on social media. This may involve editing images or using filters to enhance appearance, seeking feedback from peers before posting, selectively sharing only the most flattering images, and removing content that is perceived as unattractive or receives insufficient positive feedback (e.g. likes or comments). Collectively, these behaviours reflect strategic management of one’s physical appearance in online contexts and may indicate a heightened level of appearance investment.

Such acts may also be intuitively informed by social comparison (i.e. one chooses to present themselves in a way that is relatively favourable). Indeed, social media may stand as the “perfect” environment for unhelpful appearance comparisons because users are exposed to the most flattering (and sometimes unrealistic or enhanced) depictions of their peers, family, strangers, celebrities, and social media influencers (Fardouly et al., 2017; Verduyn et al., 2017). Correspondingly, there is mounting evidence that the relationship between social media use and worsened mental health is mediated by the magnitude and direction of social comparisons (Jabłońska and Zajdel, 2020; Masciantonio et al., 2021). However, because findings in this area are inconsistent (e.g. Coyne et al., 2020; Sarmiento et al., 2020), debate persists over the nature of social media effects (and indeed, if such effects exist at all; Orben et al., 2019; Twenge and Campbell, 2018). These discrepancies may perhaps arise due to the considerable variability in how social media use and (poor) mental health is operationalised (Petropoulos Petalas et al., 2021). That said, some preliminary evidence has revealed significantly worsened mental health (especially body dissatisfaction) in users of social media compared to non-users. This occurs among pre-adolescents (Fardouly et al., 2020), adolescents (Tiggemann and Slater, 2014), as well as adults (Stronge et al., 2015), suggesting this effect endures across the lifespan.

Appearance-based behaviours such as upward appearance comparison and appearance investment (i.e. selective self-presentation) are suggested mechanisms linking social media use to youth mental health problems (de Valle et al., 2021). Compelling cross-sectional research demonstrates a reliable association between appearance-based behaviours and mental health difficulties (e.g. see Bonfanti et al., 2025 for a review). For example, a correlational study of Australian adolescents found that appearance investment behaviours on social media were associated with poorer body image and greater depressive symptoms, but not social anxiety (Fardouly et al., 2020). Similarly, a meta-analysis of cross-sectional studies reported consistent, moderate associations between appearance comparisons on social media and greater body image concerns (Bonfanti et al., 2025). However, the cross-sectional nature of this research limits conclusions about the directionality of these relationships.

Most longitudinal studies concerning appearance-based behaviours have focused on either body dissatisfaction in isolation or an aggregate score of internalising difficulties – obfuscating nuance that could inform clinical practice. Regarding the directionality of the association with social anxiety, Rapee et al. (2022) found evidence for a latent variable including social media-based appearance comparisons that precedes worsened social anxiety. Relatedly, depressed adults and adolescents appear more vulnerable to ruminating on upward social comparison, which inadvertently substantiates their relatively poorer negative self-image and consequently their symptoms (Ahrens and Alloy, 1997; Swallow and Kuiper, 1988). A reciprocal longitudinal association also exists between appearance comparisons and body dissatisfaction (see Ooi et al., 2025, for a review). Therefore, there exists a need to validate these conjectures with further empirical evidence that canvases multiple mental health problems simultaneously.

Present study

This study examined the reciprocal relations between adolescent body dissatisfaction, depression, and social anxiety symptoms and negative appearance-based social media behaviours –upward appearance comparison frequency and curation of online appearance (hereafter, referred to as “appearance investment”). To achieve this, a cross-lagged panel design was employed over two time points, one year apart. Given that negative cognitive biases predispose and perpetuate social anxiety, depression, and body dissatisfaction, it was anticipated that greater appearance-based behaviours would share a reciprocal relationship with worsened social anxiety, depressive, and body dissatisfaction symptoms over time.

Methods

Participants and procedure

This study drew on data from two mental health screeners (collected November 2019/August 2020) delivered to secondary school students from four schools on the Northern Beaches of Sydney, Australia. Inclusion criteria were enrolment in Years 7–11 at a selected school. All students were provided with the opportunity to participate, and only recruited if their parent/guardian provided informed consent and they assented. The current study comprised responses from 1942 secondary school students on at least one testing occasion. The final sample included: 1603 respondents at W1 (50.9% female; M_age = 13.97; SD = 1.24, range = 11–16 years), and 1232 at W2 (52.3% female; M_age = 14.51; SD = 1.19, range = 12–17 years). As this was a mental health screener, no additional demographic data was collected (e.g. ethnicity, language). At Wave 1, 33.4% of males and 24.0% of females, and at Wave 2, 31.2% of males and 26.6% of females, scored above the clinical cutoff for elevated depressive symptoms (SMFQ ⩾ 6 for males and ⩾12 for females; Jarbin et al., 2020). For social anxiety, 21.7% of males and 31.0% of females at Wave 1, and 19.6% of males and 31.0% of females at Wave 2, exceeded the clinical cutoff (SCAS-Social ⩾ 8 for males and ⩾10 for females; Spence, 1998).

A fixed order of survey presentation was employed. Both screeners were delivered to the students using an anonymous Qualtrics link that required registration via their student number, wherein they would report their demographic details (e.g. age, sex, school, grade). The screeners were administered during class using classroom computers under teacher supervision, taking approximately 30 minutes on each occasion. Students were notified both orally and in writing that they were not obliged to answer any questions and could leave questions blank if any made them uncomfortable. Upon review of the responses, students with elevated scores on measures of emotional health were identified as at-risk. Subsequently, school staff contacted the student’s carer and suggested referral options. No compensation or remuneration was provided for participation. The original mental health screener, and any study employing its data, received ethics approval from Macquarie University Human Research Ethics Committee (5569).

Measures

Social media use

Participants specified whether they ever used social media (yes/no). Respondents endorsing social media use (94% at W1, 96% at W2) reported their time spent on social media using a six-point descriptive scale (1 = a few minutes a day, 2 = up to an hour a day, 3 = up to 2 hours per day, 4 = 2 to 4 hours per day, 5 = 4 to 6 hours per day, 6 = more than 6 hours per day).

Appearance comparisons

Similar to Fardouly et al. (2020), adolescents were asked to rate their frequency of appearance comparisons to others on social media (“When you look at pictures of other people on social media, how often do you compare yourself to them on how good looking they are”) using a five-point Likert-type scale (1 = never, 5 = always). Tendency for upward appearance comparisons (i.e. increased tendency to perceive others as more attractive than oneself) was assessed with one item (“When you look at pictures of other people on social media, do you think they are . . .”), scored on a four-point scale (1 = not as good looking as you, 2 = about as good looking as you, 3 = a little better looking than you, 4 = much better looking than you). An interaction term between the frequency of appearance comparison and comparison direction (i.e. the tendency for upward appearance comparison) was created via the multiplication of these two items. Total scores ranged from 1 to 20, whereby higher scores reflected more frequent engagement in appearance comparisons that were upward in direction.

Appearance investment

Drawing on the appearance improvement items contained in Fardouly et al. (2020), this study asked adolescents four questions regarding their frequency of investing in behaviours to manage their appearance on social media. This was assessed using a five-point Likert-type scale (1 = never, 5 = always). These appearance investment activities included: “tak[ing] lots of selfies in a row and post[ing] only the most attractive one”; “edit[ing] pictures with filters or other apps before posting them”; “Send[ing]/show[ing] pictures to your close friends to make sure you look good in the pictures before posting them”; and “Tak[ing] down pictures if they do not get enough good comments or likes.” Responses were averaged across items to create an overall appearance investment score ranging from 1 to 5. Higher scores indicated greater appearance investment. Internal consistency of the appearance investment measure was good in this study (W1 α = .79, W2 α = .78), like past research (α = .79; Fardouly et al., 2020).

Body dissatisfaction

Body dissatisfaction symptoms were measured using the Appearance and Weight subscales of the Body Esteem Scale for Adolescents and Adults (BESAA; Mendelson et al., 2001). Adolescents rated, on a five-point Likert-type scale (1 = never, 5 = always), how often they experienced (dis)satisfaction with their appearance and their weight based on 18 statements (e.g. “My looks upset me”). Responses were summed for all statements, with reverse-coded positively valanced statements. Possible scores ranged from 1 to 78. Higher scores reflected higher body dissatisfaction. This scale has good psychometric properties and is sensitive to changes in body dissatisfaction (Cragun et al., 2013), and demonstrated excellent internal consistency in this study (W1 & W2 α = .95).

Depression

The 13-item Short Mood and Feelings Questionnaire (SMFQ; Angold et al., 1995) was used to measure depressive symptoms. Adolescents rated how true 13 statements (e.g. “I felt lonely”) were for them in the past month, using a three-point scale (0 = not true, 2 = mostly true). Responses were summed to create a total score. Possible scores ranged from 0 to 26, with higher scores indicative of higher depressive symptom severity. Among student populations, the SMFQ is effective at detecting differences in depressive symptomatology (Thabrew et al., 2018). Internal consistency of the SMFQ was excellent in this study (W1 α = .92, W2 α = .93), similar to past research (α = .90–.92; Yap et al., 2018).

Social anxiety

The six-item social anxiety subscale of the Spence Children’s Anxiety Scale–Youth Report (SCASsoc; Spence, 1998) was used to measure social anxiety symptoms. Participants rated how often each statement (e.g. “I worry what other people think of me”) occurred to them on a four-point Likert-type scale (0 = never true, 3 = always true). Responses were summed to create a total score. Possible scores ranged from 0 to 18, with higher scores indicative of greater social anxiety symptom severity. This subscale has established validity for identifying paediatric social anxiety, and demonstrable psychometric properties (Reardon et al., 2019). The SCASsoc’s internal consistency was good in this study (W1 α = .80, W2 α = .82), similar to past research (α = .75–.78; Rapee et al., 2022).

Data analytic plan

IBM SPSS (Version 28) was used to complete all preliminary analyses, including score computation and descriptive, bivariate, reliability, and missing value analysis. While correlations provide information about the strength of associations between variables, they do not account for prior levels of the constructs or distinguish the direction of relationships. Thus, cross-lagged panel models (CLPM), with observed social media variables, social anxiety, body dissatisfaction, and depression variables, were conducted using Mplus version 8 (Muthén and Muthén, 1998–2017) to examine longitudinal reciprocal relations of interest. Missing data were handled using the Full Information Maximum Likelihood (FIML) procedure and all CLPM analyses used the maximum likelihood estimator with robust standard errors (MLR) to account for variables’ non-normality. The two observed social media variables (i.e. upward appearance comparison frequency and appearance investment) and three mental health variables (i.e. body dissatisfaction, depression, social anxiety) were each examined in a different model, resulting in six separate models. Autoregressive paths provided information about construct stability, reducing bias in parameter estimation, and permitting more tenable conclusions about changes in the predicted variables of interest (Cole and Maxwell, 2003). Cross-lagged paths were estimated to investigate the unique prospective effects of social media use on mental health symptoms over time, and vice versa, after accounting for their prior levels and concurrent associations. Residual variances per variable were correlated within time points in all models. We conducted multigroup analyses to test for moderation by adolescent sex by running models with and without autoregressive and cross-lagged parameters constrained to equality between male and female participants. Model comparisons were based on decrements in model fit, with decreases in the comparative fit index (CFI) greater than .01 denoting worse model fit (Cheung and Rensvold, 2002). Following Orth et al. (2024), we used .03 (small effect), .07 (medium effect), and .12 (large effect) as benchmarks for the effect size of cross-lagged paths. Given the large number of comparisons in the current paper, significance levels were adjusted to account for a 5% false discovery rate using the Benjamini–Hochberg procedure (Benjamini and Hochberg, 1995). The resulting adjusted p-value was .041.

Results

Attrition rate and treatment of missingness

Notably, 342 students were missing data from W1, 723 cases were missing W2 responses, and 877 secondary students had returned responses on both occasions. Reasons for drop out were a variety of practical factors including student relocation, absence on testing day (either W1 or W2), new enrolments, and potentially selective attrition. The attrition rate from W1 to W2 was 45.49%. However, the final sample size provided more than sufficient power to detect small to moderate effects, across two timepoints, with an intraclass correlation of 0.6 (Pan et al., 2018). There was approximately 17.8% and 37% missing data at W1 and W2 (respectively). The coverage covariance matrix indicated that available data across both waves ranged from 41% to 82%, which is understandable given the nature of dropout from the universal screener, and the decision to include all participants who had returned a response on at least one testing occasion.

A series of t-tests were performed to determine whether participants who completed both waves (n = 877) differed from those who dropped out of the study at W2 (n = 723) on all W1 outcome measures. Results indicated some significant differences between completers and non-completers on depressive symptoms and body dissatisfaction at W1. In W1, non-completers had significantly higher depressive symptoms (M_{non-completers} = 6.76, SD = 6.19; M_Completers = 5.85, SD = 5.68, t(1595) = -3.06, p = .002, d = -.15) and reported more body dissatisfaction (M_{non-completers} = 27.79, SD = 16.87; M_Completers = 25.33, SD = 16.00, t(1598) = −2.99, p = .003, d = −.15). Notwithstanding this, Little’s MCAR test (Little, 1988) was non-significant (χ² = 47,245.674, df = 47,578, p = .859), suggesting that the data were consistent with a missing at random assumption, and it was deemed appropriate to employ the FIML to manage missing data (Enders, 2011).

Preliminary analyses

As path analysis is concerned with the reproduction of variance-covariance matrices, particular attention was given to multivariate kurtosis (Byrne, 2013). Multivariate normality was examined using univariate and bivariate distributions as a necessary precondition (Johnson and Wichern, 1992). Multiple variables had skewness and kurtosis z-scores more than ± 1.96 (see Table 1), which was substantiated upon visual inspection of histograms and q-q plots (West et al., 1995). Normality indices were largely disregarded given their susceptibility to error when employed with large samples (Field, 2013). However, such findings are expected of community adolescent samples (e.g. severe positive skew on mental health outcomes). Consequently, Spearman’s Rho, Bootstrapping and the MLR robust maximum likelihood estimator were employed in descriptive and cross-lagged analyses, respectively. The MLR maximum likelihood estimator is robust against violations of normality (Maydeu-Olivares, 2017).

Table 1.

Full sample scores on all measures.

Variable	N	M	SD	Min	Max	Skewness	Kurtosis	Cronbach’s α
Appearance comparison
Wave 1	1351	7.35	5.73	1	20	0.89 (.07)	−0.35 (.13)	.70 (2 items)
Wave 2	1047	7.40	5.82	1	20	0.89 (.08)	−0.49 (.15)	.71 (2 items)
Appearance investment
Wave 1	1484	1.61	0.87	1	5	1.70 (.06)	2.45 (.13)	.79 (4 items)
Wave 2	1141	1.63	0.89	1	5	1.53 (.07)	1.56 (.15)	.78 (4 items)
Body dissatisfaction
Wave 1	1600	26.44	16.44	1	78	0.61 (.06)	-0.18 (.12)	.95 (18 items)
Wave 2	1219	26.99	16.90	1	78	0.63 (.07)	-0.04 (.14)	.95 (18 items)
Depression
Wave 1	1597	6.26	5.93	0	26	1.14 (.06)	0.72 (.12)	.92 (13 items)
Wave 2	1215	6.34	6.18	0	26	1.13 (.07)	0.58 (.14)	.93 (13 items)
Social anxiety
Wave 1	1603	6.41	3.85	0	18	0.49 (.06)	-0.33 (.12)	.80 (6 items)
Wave 2	1223	6.28	3.91	0	18	0.55 (.07)	-0.27 (.14)	.82 (6 items)

Histograms and box-and-whisker plots of standardised residuals were analysed to reduce outlier effects. In line with Kwak and Kim (2017), 26 values were identified which exceeded ±3 standard deviations. However, these outliers were ultimately retained as there were no significant differences between models that included/excluded identified outliers.

Univariate statistics (i.e. means, standard deviations, skewness, kurtosis, and internal consistency) are presented in Table 1. Across both waves, 94% to 96% of adolescent respondents confirmed having used social media. During W1, the largest proportion of respondents reported social media use for up to 2 hours a day (26.1%), followed by 2 to 4 hours (25.9%); whereas at W2, respondents most frequently endorsed social media use for between 2 to 4 hours per day (29.7%), followed by up to 2 hours a day (24.9%).

Table 2 contains the bivariate correlations between all outcome measures within and across time (bootstrapping was performed to obtain robust bias-corrected confidence intervals). More frequent upward appearance comparisons were strongly positively associated with depressive symptoms, social anxiety, and body dissatisfaction within and across waves. Appearance investment shared moderate significant positive associations with depressive symptoms, social anxiety, and body dissatisfaction within and between waves.

Table 2.

Spearman’s rho correlations among study variables.

	1	2	3	4	5	6	7	8	9	10	11	12
1. Dep W1	1
2. Soc W1	.62**	1
3. Bod W1	.62**	.56**	1
4. AppC W1	.55**	.55**	.67**	1
5. AppI W1	.30**	.32**	.37**	.56**	1
6. Dep W2	.58**	.49**	.48**	.44**	.30**	1
7. Soc W2	.44**	.64**	.45**	.42**	.25**	.66**	1
8. Bod W2	.46**	.45**	.70**	.54**	.36**	.64**	.57**	1
9. AppC W2	.39**	.45**	.52**	.63**	.48**	.60**	.58**	.70**	1
10. AppI W2	.20**	.27**	.30**	.46**	.69**	.39**	.37**	.42**	.60**	1
11. Age	.06*	.05	.12**	.13**	.07*	.09*	.02	.07*	.07*	.05	1
12. Sex	.23**	.33**	.28**	.43**	.53**	.27**	.34**	.34**	.48**	.57**	−.01	1

W1: data collected at Wave 1; W2: data collected at Wave 2; Dep: Short Mood and Feelings Questionnaire; Soc: Social Anxiety Subscale of Spence Children’s Anxiety Scale; Bod: Body Esteem Scale for Adolescents and Adults; AppC: Appearance Comparisons; AppI: Appearance investment. All measures are self-reported.

p < .041. **p < .001.

Age positively correlated with depression symptom severity, body dissatisfaction, tendency to make upward appearance comparisons, and appearance investment (see Table 2), but not social anxiety, within and across waves. Therefore, age was controlled for in subsequent cross-lagged analyses. As expected, females reported significantly higher mental health difficulties than males, with a greater tendency to make frequent upward appearance comparisons and to engage in appearance investment across both waves.

Path model analysis

Appearance comparisons

Appearance comparisons and body dissatisfaction

Constraining autoregressive and cross-lagged parameters to equality between males and females did not worsen model fit (∆CFI = .001), suggesting the paths were not moderated by sex. The final model resulted in good model fit (CFI = .997, Tucker–Lewis index [TLI] = .984, root mean square error of approximation [RMSEA] = .041 and standardised root mean square residual [SRMR] = .009; Hu and Bentler, 1999). As seen in Figure 1 (Model 1A), autoregressive paths were positive and significant, suggesting stability in the variables over time. The cross-lagged paths were also significant and positive, suggesting bidirectional relationships between body dissatisfaction and appearance comparisons over time. W1 upward comparison tendency predicted medium increases in body dissatisfaction at W2, and W1 body dissatisfaction predicted strong increases in upward comparison tendency at W2.

Figure 1.

Cross-lagged models testing bidirectional relationships between upward appearance comparison frequency on social media and mental health symptoms. Age was controlled for in the models and correlations between variables at each wave were included but not reported in the figure for simplicity. Autoregressive paths are grey and cross-lagged paths are black. Significant paths are indicated with solid lines. Standardised beta coefficients are reported; these parameters are interpreted as the predicted change in the outcome variable based on a one standard deviation increase in the independent variable.

Appearance comparisons and depression

Sex did not moderate the autoregressive and cross-lagged paths, as constraining those parameters to equality between males and females did not worsen model fit (∆CFI = .004). The final model had good model fit (CFI = .998, TLI = .988, RMSEA = .034, SRMR = .007; Hu and Bentler, 1999). As seen in Figure 1 (Model 1B), all autoregressive and cross-lagged paths were positive and significant. A bidirectional relationship emerged between depressive symptom severity and upward appearance comparison tendency, with both being strong predictors of increases in the other over time.

Appearance comparisons and social anxiety

Constraining autoregressive and cross-lagged parameters to equality between males and females did not worsen model fit (∆CFI = .001), suggesting the paths were not moderated by sex. The final model resulted in good model fit (CFI = .998, TLI = .988, RMSEA = .034, SRMR = .006; Hu and Bentler, 1999). As seen in Figure 1 (Model 1C), all autoregressive and cross-lagged paths were positive and significant. W1 upward appearance comparison tendency predicted medium increases in social anxiety at W2, and W1 social anxiety predicted strong increases in upward comparison tendency at W2.

Appearance investment

Appearance investment and body dissatisfaction

Constraining autoregressive and cross-lagged parameters to equality between males and females did not worsen model fit (∆CFI = .006), suggesting the paths were not moderated by sex. The final model had good model fit (CFI = .997, TLI = .982, RMSEA = .039, SRMR = .008; Hu and Bentler, 1999). As seen in Figure 2 (Model 2A), autoregressive paths were significant and positive. W1 appearance investment did not significantly predict body dissatisfaction at W2, but W1 body dissatisfaction significantly predicted medium increases in appearance investment at W2.

Figure 2.

Cross-lagged models testing bidirectional relationships between appearance investment on social media and mental health symptoms. Age was controlled for in the models and correlations between variables at each wave were included but not reported in the figure for simplicity. Autoregressive paths are grey and cross-lagged paths are black. Significant paths are indicated with solid lines. Standardised beta coefficients are reported; these parameters are interpreted as the predicted change in the outcome variable based on a one standard deviation increase in the independent variable.

Appearance investment and depression

Sex did not moderate the autoregressive or cross-lagged paths, as constraining those parameters to equality between males and females did not worsen model fit (∆CFI = .007). The final model had good model fit (CFI = 1.00, TLI = 1.00, RMSEA ⩽ .001, SRMR = .004; Hu and Bentler, 1999). As seen in Figure 2 (Model 2B), autoregressive paths were significant and positive, suggesting stability in the variables over time. W1 appearance investment predicted medium increases in depressive symptom severity at W2, but W1 depression did not significantly predict W2 appearance investment.

Appearance investment and social anxiety

Constraining autoregressive and cross-lagged parameters to equality between males and females did not worsen model fit (∆CFI = .005), suggesting the paths were not moderated by sex. The final model resulted in good model fit (CFI = 1.00, TLI = 1.00, RMSEA ⩽ .001, SRMR = .003; Hu and Bentler, 1999). All autoregressive paths were significant (see Figure 2, Model 2C). W1 appearance investment did not significantly predict social anxiety at W2, but W1 social anxiety significantly predicted strong increases in appearance investment at W2.

Discussion

Given that the extant literature linking social media use with adolescent mental health is largely cross-sectional, this study adopted a longitudinal approach to elucidate the directionality between appearance-based social media behaviours (i.e. upward appearance comparison and appearance investment) and mental health over two time points. As anticipated, within each wave, appearance-based social media behaviours were significantly associated with each mental health symptom. In support of our hypothesis, bidirectional prospective relationships were found between upward appearance comparisons on social media and each of the mental health symptoms. A different pattern emerged for social media appearance investment. Body dissatisfaction and social anxiety predicted increases in appearance investment over time, and appearance investment predicted increases in depressive symptoms. Importantly, these findings highlight that mental health is a predictor, not just an outcome, of appearance-based social media behaviours. Adolescents’ sex did not moderate any of the relationships found in the study. These findings suggest that although females report more appearance-based social media activities and more social-emotional disorder symptoms than males, the relationships between those variables are similar for both sexes.

This study replicated the correlational pattern between appearance-based social media use and mental health symptoms. That is, body dissatisfaction, social anxiety, and depressive symptoms were found to have a strong relationship with appearance comparisons, but a small to moderate positive relationship with appearance investment, within both waves. This result is consistent with theoretical conjectures (Festinger, 1954; Rapee et al., 2019) and aligns with previous findings (e.g. Bonfanti et al., 2025; McLean et al., 2015). Furthermore, correlations indicated that the two appearance-based social media activities were similar yet sufficiently distinct, within and across waves, which permitted modelling each process separately.

It was anticipated that appearance-based activities would bidirectionally predict increased mental health symptoms over time. Indeed, bidirectional relationships were found between upward appearance comparisons and each of the mental health symptoms with medium to strong effect sizes, which is consistent with a growing body of empirical evidence (de Valle et al., 2021; Ooi et al., 2025; Schønning et al., 2020). These findings also provide further support for sociocultural models (Thompson et al., 1999), social comparison theories (Festinger, 1954), and cognitive theories of the aetiology of social-emotional disorders (Beck, 2002; LeMoult and Gotlib, 2019; Teasdale, 1983). These findings highlight the importance and negative role of upward appearance comparisons on social media for adolescent mental health. The consistent pattern between each distinct mental health concern aligns with previous demonstrations of their strong inter-correspondence, validating their treatment as a cluster of “social-emotional difficulties.” These findings also suggest that appearance-based comparisons have implications for broader psychopathology, extending beyond concerns about appearance.

It is worth noting that the paths leading from body dissatisfaction and social anxiety to upward appearance comparisons were twice as strong as the inverse relationship (see Figure 1). Much of the literature focuses on the negative impact of upward appearance comparisons on the body image and mental health of users. Although our findings support this literature, they suggest that mental health concerns are an important driver of adolescents’ tendency to make upward appearance comparisons, and this direction warrants greater emphasis in future research. Adolescents who are more dissatisfied with their appearance, more socially anxious, and more depressed are often more uncertain of themselves (i.e. have low self-concept clarity), which may lead them to define themselves based on their external characteristics, such as their physical appearance (Vartanian et al., 2018). As such, adolescents who experience more social-emotional disorder symptoms may seek out upward appearance comparisons on social media as a source of inspiration to improve their physical attractiveness and to determine their standing relative to others, which in turn may result in stronger social-emotional disorder symptoms when they frequently judge themselves to be less attractive than others. Future longitudinal research should investigate how appearance comparisons and symptoms of social-emotional disorders develop across adolescence to clarify which tend to emerge first.

Unexpectedly, appearance investment on social media did not precede the development of social anxiety and body dissatisfaction a year later. Rather, social anxiety and body dissatisfaction preceded increased appearance investment (i.e. selective self-presentation), which accords with clinical models of anxiety (Rapee and Heimberg, 1997; Spence and Rapee, 2016). From this, online appearance investment might be deemed a safety behaviour or form of subtle avoidance to prevent public scrutiny (Piccirillo et al., 2016), which inadvertently exacerbates appearance-based insecurities (and the risk of psychopathology) in the long term. Appearance investment was found to precede worsened depressive symptoms, but not the inverse. Individuals may be motivated to engage in appearance investment as a means of coping with low body dissatisfaction and social anxiety (Tiggemann et al., 2020). This, in turn, may contribute to the formation/consolidation of negative self-beliefs (e.g. concerning one’s own perceived inferiority or unattractiveness), thereby perpetuating an already-poor self-concept and intensifying depressogenic thinking. Alternatively, these findings could be attributed to the nature of appearance investment as an activity (following behavioural theories of depression; Lejuez et al., 2011; Lewinsohn, 1974; Teychenne et al., 2008). That is, appearance investment (especially of high frequency/duration) would oblige users to spend more time sedentarily and passively engaging with their social networks, rather than being positively reinforced by face-to-face interaction. In either case, the lack of relationship from depression to appearance investment is understandable, as depressed individuals may be less motivated to actively curate an online presence compared to non-depressed individuals. When combined, these findings suggest a possible mediational relationship whereby social anxiety and body dissatisfaction predict engagement in appearance investment, which then predicts future depression. This hypothesis will require longitudinal research across at least three waves to be tested.

Limitations and future directions

Several limitations temper this study’s findings. First, an important consideration is to contextualise Wave 2 (August 2020) in the COVID-19 lockdown period from which NSW students had just emerged, which likely contributed to overall elevations in mental health symptoms at Wave 2. Students were naturally more dependent on social media use to fulfil their social needs during lockdowns (Schreurs and Vandenbosch, 2022). Lockdown stresses may also have contributed to the observed drop-out rates, although they were nonetheless representative of selective attrition in other community samples outside of the COVID-19 pandemic (Dupuis et al., 2019). Although the cross-lagged panel model adjusts for individuals’ pre-pandemic baseline levels of both social media use and mental health, pandemic-related stressors represent an unmeasured time-varying influence that cannot be fully disentangled from other sources of change. As replication (post-pandemic) may be warranted, investigators could study whether these trends emerge in clinical samples (e.g. regarding the severity of symptoms and dose-response; Liu et al., 2022) and over differing time scales (e.g. multiple years; within 1 year; Hamaker et al., 2015). Second, mental health symptoms were examined in separate models and the analyses do not allow us to determine whether the observed associations reflect variance shared across symptoms (e.g. a general internalising or socio-emotional distress factor) or variance unique to any specific symptom. Rather, the findings indicate that each mental health construct is longitudinally associated with appearance-based social media behaviours when examined separately. Third, we assessed appearance-based social media behaviours across platforms without distinguishing among them. Although a process-focused approach allows examination of mechanisms that generalise across social media environments, platform-specific features and norms may shape these experiences. Future research should examine whether associations differ across platforms such as Instagram and TikTok. Fourth, future studies could also explore the influence of potential moderators (e.g. socioeconomic status, initial age of social media use) and/or mediators (e.g. internalisation of beauty ideals) on the observed effects. Finally, qualitative exploration of adolescents’ experiences of online spaces, especially those with heightened social-emotional disorder symptoms, would enrich theoretical understandings of the meaning and implications of social media use for teens and provide specific mechanism-tied hypotheses for quantitative research.

Conclusion

Given the predominant cross-sectional literature base, a key strength of this study was the longitudinal modelling of specific types of social media use with specific mental health difficulties. Creditably, this study adopted an affordances approach (i.e. conceptualising social media platforms generally in terms of what they offer; Valkenburg and Peter, 2013) rather than restricting analyses to specific platforms that may fall out of popularity with teenagers (e.g. Facebook). As such, this study extends previous literature by examining adolescents’ engagement in specific appearance-based social media behaviours, providing insight into how particular forms of use may relate to teen wellbeing. The bidirectional relationships observed between upward appearance comparisons and mental health symptoms are particularly concerning, given the high prevalence of upward comparison targets within highly visual social media environments, such as Instagram and TikTok, which are widely used by adolescents. These findings also suggest that appearance-based social media behaviours may not only contribute to poorer mental health, but may also be reinforced by existing psychological distress. Together, this reciprocal pattern indicates that adolescents with existing vulnerabilities may become caught in a reinforcing cycle of appearance-focused engagement and declining wellbeing. Accordingly, the present findings provide further support for efforts to reduce the prominence of idealised and digitally enhanced appearances on social media, particularly to protect adolescents who may already be at risk for mental health difficulties.

Footnotes

Acknowledgements

The authors gratefully acknowledge the dedication and work contributed by Ian Bowsher and the principals and school staff from the Peninsula Community of Schools, without whose efforts this research could not have been possible.

Author contributions

Cameron J Ragg: Conceptualisation; Formal analysis; Writing – original draft.

Ronald M Rapee: Conceptualisation; Methodology; Project administration; Supervision, Writing – review & editing.

Jasmine Fardouly: Conceptualisation; Formal analysis; Writing – original draft.

Ethical considerations

The original mental health screener, and any study employing its data, received ethics approval from Macquarie University Human Research Ethics Committee (number 5569).

Consent to participate

Written informed consent was obtained from the parent/guardian, and the adolescents provided their assent.

Consent for publication

Not applicable.

Data availability statement

Data is available from the authors on request.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Ronald M Rapee

Jasmine Fardouly

Author biographies

Cameron J Ragg, MPsych(Clin), is an early career clinical psychology registrar based in Sydney, Australia. He currently works across public health settings, including inpatient psycho-oncology and mental health rehabilitation. His broad clinical and research interests include anxiety, severe mental health, psychological wellbeing, psychological treatment, aetiology, and body image.

Ronald M Rapee, PhD, is a Distinguished Professor of Psychology at Macquarie University, with a background in clinical psychology. His research interests include body image, mental health screening, anxiety, and depression across the lifespan.

Jasmine Fardouly, PhD, is an Australian Research Council Discovery Early Career Fellow and Horizon Fellow at the School of Psychology, University of Sydney, Australia. Her research interests include social influences on body image and mental health, particularly the influence of social media use.

References

Ahrens

Alloy

(1997) Social comparison processes in depression. In: Buunk

Gibbons

(eds) Health, Coping, and Well-being: Perspectives From Social Comparison Theory. Hillsdale, NJ: Lawrence Erlbaum Associates, pp. 389–410.

Allen

Ryan

Gray

, et al. (2014) Social media use and social connectedness in adolescents: The positives and the potential pitfalls. The Australian Educational and Developmental Psychologist 31(1): 18–31.

Angold

Costello

Messer

, et al. (1995) Development of a short questionnaire for use in epidemiological studies of depression in children and adolescents. International Journal of Methods in Psychiatric Research 5: 237–249.

Beck

(2002) Cognitive models of depression. In: Leahy

Dowd

(eds) Clinical Advances in Cognitive Psychotherapy: Theory and Application. Cham: Springer, pp. 29–61.

Bell

Bishop

DVM

Przybylski

(2015) The debate over digital technology and young people. BMJ 351: h3064.

Benjamini

Hochberg

(1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B (Methodological) 57(1): 289–300.

Bonfanti

Melchiori

Teti

, et al. (2025) The association between social comparison in social media, body image concerns and eating disorder symptoms: A systematic review and meta-analysis. Body Image 52: 101841.

Brown

Larson

(2009) Peer relationships in adolescence. In: Lerner

Steinberg

(eds) Handbook of Adolescent Psychology. New York: John Wiley & Sons, pp. 74–103. DOI: 10.1002/9780470479193.adlpsy002004.

Byrne

(2013) Structural Equation Modeling with Mplus: Basic Concepts, Applications, and Programming. London: Routledge. DOI: 10.4324/9780203807644.

10.

Cheung

Rensvold

(2002) Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling: A Multidisciplinary Journal 9(2): 233–255.

11.

Cole

Maxwell

(2003) Testing mediational models with longitudinal data: Questions and tips in the use of structural equation modeling. Journal of Abnormal Psychology 112(4): 558–577.

12.

Costello

Copeland

Angold

(2011) Trends in psychopathology across the adolescent years: What changes when children become adolescents, and when adolescents become adults? Journal of Child Psychology and Psychiatry 52(10): 1015–1025.

13.

Coyne

Rogers

Zurcher

, et al. (2020) Does time spent using social media impact mental health?: An eight year longitudinal study. Computers in Human Behavior 104: 106160.

14.

Cragun

DeBate

Ata

, et al. (2013) Psychometric properties of the Body Esteem Scale for Adolescents and Adults in an early adolescent sample. Eating and Weight Disorders - Studies on Anorexia, Bulimia and Obesity 18(3): 275–282.

15.

Crone

Dahl

(2012) Understanding adolescence as a period of social–affective engagement and goal flexibility. Nature Reviews Neuroscience 13(9): 636–650.

16.

de Lijster

Dierckx

Utens

EMWJ

, et al. (2017) The age of onset of anxiety disorders. The Canadian Journal of Psychiatry 62(4): 237–246.

17.

de Valle

Gallego-García

Williamson

, et al. (2021) Social media, body image, and the question of causation: Meta-analyses of experimental and longitudinal evidence. Body Image 39: 276–292.

18.

Dupuis

Strippoli

Gholam-Rezaee

, et al. (2019) Mental disorders, attrition at follow-up, and questionnaire non-completion in epidemiologic research. Illustrations from the CoLaus|PsyCoLaus study. International Journal of Methods in Psychiatric Research 28: e1805.

19.

Enders

(2011) Analyzing longitudinal data with missing values. Rehabilitation Psychology 56(4): 267–288.

20.

Fardouly

Magson

Rapee

, et al. (2020) The use of social media by Australian preadolescents and its links with mental health. Journal of Clinical Psychology 76(7): 1304–1326.

21.

Fardouly

Pinkus

Vartanian

(2017) The impact of appearance comparisons made through social media, traditional media, and in person in women’s everyday lives. Body Image 20: 31–39.

22.

Festinger

(1954) A theory of social comparison processes. Human Relations 7(2): 117–140.

23.

Field

(2013) The beast of bias. In: Field

(ed.) Discovering Statistics Using IBM SPSS Statistics. 4th ed. London: SAGE, pp. 184.

24.

Gowers

Shore

(2001) Development of weight and shape concerns in the aetiology of eating disorders. British Journal of Psychiatry 179(3): 236–242.

25.

Guyer

Silk

Nelson

(2016) The neurobiology of the emotional adolescent: From the inside out. Neuroscience & Biobehavioral Reviews 70: 74–85.

26.

Haidt

Allen

(2020) Scrutinizing the effects of digital technology on mental health. Nature 578(7794): 226–227.

27.

Hamaker

Kuiper

Grasman

RPPP

(2015) A critique of the cross-lagged panel model. Psychological Methods 20(1): 102–116.

28.

Bentler

(1999) Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal 6(1): 1–55.

29.

Jabłońska

Zajdel

(2020) Artificial neural networks for predicting social comparison effects among female Instagram users. PLoS ONE 15(2): e0229354.

30.

Jarbin

Ivarsson

Andersson

, et al. (2020) Screening efficiency of the Mood and Feelings Questionnaire (MFQ) and Short Mood and Feelings Questionnaire (SMFQ) in Swedish help seeking outpatients. PLoS ONE 15(3): e0230623.

31.

Johnson

Wichern

(1992) Applied Multivariate Statistical Analysis (ed Wichern

). 3rd ed. Englewood Cliffs, NJ: Prentice Hall.

32.

Kemp

(2022) Digital 2022: Australia. Available at: https://datareportal.com/reports/digital-2022-australia

33.

Kessler

Berglund

Demler

, et al. (2005) Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the national comorbidity survey replication. Archives of General Psychiatry 62(6): 593.

34.

Kircanski

Joormann

Gotlib

(2012) Cognitive aspects of depression. Wiley Interdisciplinary Reviews. Cognitive Science 3(3): 301–313.

35.

Kwak

Kim

(2017) Statistical data preparation: Management of missing values and outliers. Korean Journal of Anesthesiology 70(4): 407.

36.

Lawler

Nixon

(2011) Body dissatisfaction among adolescent boys and girls: The effects of body mass, peer appearance culture and internalization of appearance ideals. Journal of Youth and Adolescence 40(1): 59–71.

37.

Lawrence

Hafekost

Johnson

, et al. (2016) Key findings from the second Australian child and adolescent survey of mental health and wellbeing. Australian & New Zealand Journal of Psychiatry 50(9): 876–886.

38.

Lejuez

Hopko

Acierno

, et al. (2011) Ten year revision of the brief behavioral activation treatment for depression: Revised treatment manual. Behavior Modification 35(2): 111–161.

39.

LeMoult

Gotlib

(2019) Depression: A cognitive perspective. Clinical Psychology Review 69: 51–66.

40.

Lewinsohn

(1974) A behavioral approach to depression. In: Coyle

(ed.) Essential Papers on Depression. New York: New York University Press, pp. 150–172.

41.

Little

RJA

(1988) A test of missing completely at random for multivariate data with missing values. Journal of the American Statistical Association 83(404): 1198–1202.

42.

Liu

Kamper-DeMarco

Zhang

, et al. (2022) Time spent on social media and risk of depression in adolescents: A dose–response meta-analysis. International Journal of Environmental Research and Public Health 19(9): 5164.

43.

McLaughlin

Belon

Smith

, et al. (2015) Mothers’ and daughters’ beliefs about factors affecting preadolescent girls’ body satisfaction. Body Image 13: 9–17.

44.

McLean

Paxton

Wertheim

, et al. (2015) Photoshopping the selfie: Self photo editing and photo investment are associated with body dissatisfaction in adolescent girls. International Journal of Eating Disorders 48(8): 1132–1140.

45.

Masciantonio

Bourguignon

Bouchat

, et al. (2021) Don’t put all social network sites in one basket: Facebook, Instagram, Twitter, TikTok, and their relations with well-being during the COVID-19 pandemic. PLoS ONE 16(3): e0248384.

46.

Maughan

Collishaw

Stringaris

(2013) Depression in childhood and adolescence. Journal of the Canadian Academy of Child and Adolescent Psychiatry 22(1): 35–40.

47.

Maydeu-Olivares

(2017) Maximum likelihood estimation of structural equation models for continuous data: Standard errors and goodness of fit. Structural Equation Modeling: A Multidisciplinary Journal 24(3): 383–394.

48.

Mendelson

White

(2001) Body-esteem scale for adolescents and adults. Journal of Personality Assessment 76(1): 90–106.

49.

Muthén

(1998-2017) Mplus User’s Guide. Muthén & Muthén. Available at: https://www.statmodel.com/download/usersguide/MplusUserGuideVer_8.pdf

50.

Nesi

Choukas-Bradley

Prinstein

(2018) Transformation of adolescent peer relations in the social media context: Part 1—A theoretical framework and application to dyadic peer relationships. Clinical Child and Family Psychology Review 21(3): 267–294.

51.

Ooi

Nasser

Simmons

, et al. (2025) A systematic review and meta-analysis on the temporal relationship between appearance comparisons and body dissatisfaction. Body Image 53: 101885.

52.

Orben

(2020) Teenagers, screens and social media: A narrative review of reviews and key studies. Social Psychiatry and Psychiatric Epidemiology 55(4): 407–414.

53.

Orben

Przybylski

(2019) The association between adolescent well-being and digital technology use. Nature Human Behaviour 3(2): 173–182.

54.

Orben

Dienlin

Przybylski

(2019) Social media’s enduring effect on adolescent life satisfaction. Proceedings of the National Academy of Sciences of the United States of America 116(21): 10226–10228.

55.

Orth

Meier

Bühler

, et al. (2024) Effect size guidelines for cross-lagged effects. Psychological Methods 29(2): 421–433.

56.

Pan

Liu

Miao

, et al. (2018) Sample size determination for mediation analysis of longitudinal data. BMC Medical Research Methodology 18(1): 32.

57.

Petropoulos Petalas

Konijn

Johnson

, et al. (2021) Plurality in the measurement of social media use and mental health: An exploratory study among adolescents and young adults. Social Media + Society 7(3): 205630512110353.

58.

Piccirillo

Taylor Dryman

Heimberg

(2016) Safety behaviors in adults with social anxiety: Review and future directions. Behavior Therapy 47(5): 675–687.

59.

Rapee

Heimberg

(1997) A cognitive-behavioral model of anxiety in social phobia. Behaviour Research and Therapy 35(8): 741–756.

60.

Rapee

Magson

Forbes

, et al. (2022) Risk for social anxiety in early adolescence: Longitudinal impact of pubertal development, appearance comparisons, and peer connections. Behaviour Research and Therapy 154: 104126.

61.

Rapee

Oar

Johnco

, et al. (2019) Adolescent development and risk for the onset of social-emotional disorders: A review and conceptual model. Behaviour Research and Therapy 123: 103501.

62.

Reardon

Creswell

Lester

, et al. (2019) The utility of the SCAS-C/P to detect specific anxiety disorders among clinically anxious children. Psychological Assessment 31(8): 1006–1018.

63.

Rodgers

McLean

Paxton

(2015) Longitudinal relationships among internalization of the media ideal, peer social comparison, and body dissatisfaction: Implications for the tripartite influence model. Developmental Psychology 51(5): 706–713.

64.

Sarmiento

Olson

Yeo

, et al. (2020) How does social media use relate to adolescents’ internalizing symptoms? Conclusions from a systematic narrative review. Adolescent Research Review 5(4): 381–404.

65.

Schønning

Hjetland

Aarø

, et al. (2020) Social media use and mental health and well-being among adolescents—A scoping review. Frontiers in Psychology 11: 1949.

66.

Schreurs

Vandenbosch

(2022) Different interactions with appearance-focused social media content and adolescents’ body dissatisfaction: A within-person perspective. Computers in Human Behavior 135: 107364.

67.

Sebastian

Burnett

Blakemore

S-J

(2008) Development of the self-concept during adolescence. Trends in Cognitive Sciences 12(11): 441–446.

68.

Shafran

Fairburn

Robinson

, et al. (2004) Body checking and its avoidance in eating disorders. International Journal of Eating Disorders 35(1): 93–101.

69.

Somerville

(2013) The teenage brain: Sensitivity to social evaluation. Current Directions in Psychological Science 22(2): 121–127.

70.

Spence

(1998) A measure of anxiety symptoms among children. Behaviour Research and Therapy 36(5): 545–566.

71.

Spence

Rapee

(2016) The etiology of social anxiety disorder: An evidence-based model. Behaviour Research and Therapy 86: 50–67.

72.

Stronge

Greaves

Milojev

, et al. (2015) Facebook is linked to body dissatisfaction: Comparing users and non-users. Sex Roles 73(5–6): 200–213.

73.

Swallow

Kuiper

(1988) Social comparison and negative self-evaluations: An application to depression. Clinical Psychology Review 8(1): 55–76.

74.

Teasdale

(1983) Negative thinking in depression: Cause, effect, or reciprocal relationship? Advances in Behaviour Research and Therapy 5(1): 3–25.

75.

Teychenne

Ball

Salmon

(2008) Physical activity and likelihood of depression in adults: A review. Preventive Medicine 46(5): 397–411.

76.

Thabrew

Stasiak

Bavin

, et al. (2018) Validation of the Mood and Feelings Questionnaire (MFQ) and Short Mood and Feelings Questionnaire (SMFQ) in New Zealand help-seeking adolescents. International Journal of Methods in Psychiatric Research 27: e1610.

77.

Thompson

Heinberg

(1999) The media’s influence on body image disturbance and eating disorders: We’ve reviled them, now can we rehabilitate them? Journal of Social Issues 55(2): 339–353.

78.

Thompson

Heinberg

Altabe

, et al. (1999) Exacting Beauty: Theory, Assessment, and Treatment of Body Image Disturbance. Washington, DC: American Psychological Association. Available at: https://doi.org/10.1037/10312-000

79.

Tiggemann

Slater

(2014) NetTweens. The Journal of Early Adolescence 34(5): 606–620.

80.

Tiggemann

Anderberg

Brown

(2020) Uploading your best self: Selfie editing and body dissatisfaction. Body Image 33: 175–182.

81.

Twenge

Campbell

(2018) Associations between screen time and lower psychological well-being among children and adolescents: Evidence from a population-based study. Preventive Medicine Reports 12: 271–283.

82.

Twenge

Campbell

(2019) Media use is linked to lower psychological well-being: Evidence from three datasets. Psychiatric Quarterly 90(2): 311–331.

83.

Valkenburg

Peter

(2011) Online communication among adolescents: An integrated model of its attraction, opportunities, and risks. Journal of Adolescent Health 48(2): 121–127.

84.

Valkenburg

Peter

(2013) The differential susceptibility to media effects model. Journal of Communication 63(2): 221–243.

85.

Vartanian

Hayward

Smyth

, et al. (2018) Risk and resiliency factors related to body dissatisfaction and disordered eating: The identity disruption model. International Journal of Eating Disorders 51(4): 322–330.

86.

Verduyn

Ybarra

Résibois

, et al. (2017) Do social network sites enhance or undermine subjective well-being? A critical review. Social Issues and Policy Review 11(1): 274–302.

87.

Walther

(2007) Selective self-presentation in computer-mediated communication: Hyperpersonal dimensions of technology, language, and cognition. Computers in Human Behavior 23(5): 2538–2557.

88.

West

Finch

Curran

(1995) Structural equation models with nonnormal variables: Problems and remedies. In: Hoyle

(ed.) Structural Equation Modeling: Concepts, Issues and Applications. London: SAGE, pp. 56–75.

89.

Yap

MBH

Mahtani

Rapee

, et al. (2018) A tailored web-based intervention to improve parenting risk and protective factors for adolescent depression and anxiety problems: Postintervention findings from a randomized controlled trial. Journal of Medical Internet Research 20(1): e17.