Evidence on whether social media use is associated with poor mental health and stress remains mixed and controversial; whilst online connectivity can afford many advantages, such as increasing social capital and reducing loneliness (Wilson et al., 2012), prolonged use has also been associated with anxiety and depression (Seabrook et al., 2016). It is acknowledged that some if the inconsistency in reported effects may depend on differences in measurement approaches (e.g., cross-sectional vs. longitudinal tracking of effects), and in the operationalisation of social media ‘use’ (see e.g., Coyne, et al., 2020, Odgers & Jensen, 2020). When conceived as a stressor, researchers have tended to use either general measures of general stress symptoms to index the negative affective component of social media stress, such as ‘social media fatigue’ (Bright et al., 2015), or have inferred stress from use or dependency (van der Schuur et al., 2018).

Social media stress (SMS) (van der Schuur et al., 2018), offers a way to capture both excessive use and negative affect, that can stem from processing demands (e.g., coping with continuous information flow—‘push notifications’) and the key socio-emotional factors (e.g., seeking endorsement; fear of missing out on updates) that have been associated with anxiety and depression (e.g., Hunt et al., 2018). Indeed, it has been suggested that certain individual-level characteristics may act as risk factors for social media stress, including self-efficacy and sensitivity to social comparisons and influence (Laumer & Maier, 2021).

To better understand the role of individual differences in this relationship, our study seeks to examine whether emotional intelligence (EI)—our capacity to identify, understand and manage emotions—may be useful for managing social media stress to maintain mental health. EI can be measured in two different ways – as an ability EI (AEI), indexing maximal emotional knowledge and skill in perceiving, using, understanding and managing emotion, or a trait (TEI), capturing self-perceptions of emotional competencies. Therefore, whilst the former is a distinct form of intelligence for processing emotional information (Mayer et al., 2008), the latter is partially determined by personality and can be viewed as our emotional self-efficacy (Petrides, 2009). Early critics of EI suggested it was too ill-defined to be meaningful and lacked distinctiveness from existing measures of broadband personality or intelligence to offer new insights (Matthews et al., 2004), however, there is now evidence that EI can be measured reliably and can incrementally predict variance in adaptive outcomes (see e.g., MacCann et al., 2020). Thus, TEI and AEI are useful complementary approaches to understanding how emotionally knowledgeable a person is (AEI), and how readily they put this knowledge into practice in everyday life (TEI).

Both TEI and AEI are significantly associated with indicators of mental health, although effects are usually more robust for TEI than AEI (Martins et al., 2010; Sánchez-Álvarez et al., 2016) and each operates differently within stress-mental health pathways – protective effects are dependent upon stressor type and outcome examined (Davis & Humphrey, 2012a, b; Lea et al., 2019). Whilst evidence suggests that TEI relates to reduced levels of problematic social media use (Arrivillaga et al., 2022), and may be useful for managing negative affect arising from detecting ‘fake news’ on social media (Preston et al., 2021), researchers have yet to directly examine whether EI can modify the effects of social media stress. This is important since EI can be effectively trained through intervention (Hodzic et al., 2017).

Thus, the current study explored the relationships between EI (trait or ability) mental health (wellbeing and disorder) and social media stress to answer: Does EI moderate the effects of social media stress on mental health? Based on previous literature (e.g., Martins, et al., 2010), we predicted that measures of ability EI and trait EI would be negatively related to indicators of disorder (here measured as depression and anxiety symptoms), and positively related to wellbeing, although relationships would be more robust for trait rather than ability EI. In the Salovey and Mayer (1990), Mayer and Salovey (1997) four-branch model of ability EI, emotion management and understanding represent ‘strategic’ vs. ‘experiential EI skills (emotion perception and use), and these skills have been most consistently implicated as predictors of mental health outcomes hence their inclusion here (see e.g., e.g., Bastian et al., 2005; Goldenberg et al., 2006).

Given the absence of studies examining the relationship between EI and social media stress, we anticipated that both trait and ability EI would be related negatively to social media stress, following literature examining the relationship between EI and both acute (Lea et al., 2019) and chronic stressors (e.g., Davis & Humphrey, 2012a).


Participants and Procedure

201 opportunity-sampled social media users (83.6% female; mean age 26.12, SD = 8.86) consented to complete an online battery of measures (20 min duration). Participants were recruited via social media networks. To capture time spent on social media, participants were asked: “In the last 30 days, on an average day, how many hours did you use the internet for social networking?”. Following Bányai et al., (2017), participants selected an appropriate time category with 40.8% of the sample spent 2–3 h on social networking sites per day (30 min = 3.5%; 1 h = 12.5%; 4–5 h = 21.4%; 6 + hours = 21.9%).


Social Media Stress

The 10-item Social Media Stress Scale (van der Schuur et al., 2018) indexes emotional responses and excessive social media use via a 5-point scale, 0 (never) to 5 (completely true) [total score: 0–50]. Example items include “Have you felt tense or restless when you could not use social media?’ ‘Neglected other activities to use social media?’

Emotional Intelligence

The Trait Emotional Intelligence Questionnaire–Short Form (Petrides, 2009) requires participants to use a 7-point scale, 1 (completely disagree) to 7 (completely agree) to rate 30-items indicating sociability, emotionality, self-control, well-being, [global TEI score: 30–210], e.g., “Many times, I can’t figure out what emotion I'm feeling.” Strategic ability emotional intelligence was indicated by the Situational Test of Emotion Management-Brief (Allen et al., 2015), where participants choose the most effective course of action for managing emotion (anger, sadness, fear) across 18 scenarios [total score: 0–18]. For example, “Jumah has been working at a new job part-time while he studies. His shift times for the week are changed at the last minute, without consulting him. What action would be the most effective for Jumah? (a) Refuse to work the new shifts. (b) Find out if there is some reasonable explanation for the shift changes. (c) Tell the manager in charge of shifts that he is not happy about it. (d) Grumpily accept the changes and do the shifts.” The Situational Test of Emotion Understanding-Brief (Allen et al., 2014) assesses knowledge relating to 14 different emotions through 19 multiple-choice scenarios scored as correct/incorrect [0–19], e.g., “Clara receives a gift. Clara is most likely to feel? (a) happy (b) angry (c) frightened (d) bored (e) hungry”.


The 14-item Warwick-Edinburgh Mental Wellbeing Scale (Tennant et al., 2007) requires participants rate how closely a statement matches their experience over the past 2 weeks using a 5-point scale, 1 (none of the time) to 5 (all of the time) [14–70], e.g., “I’ve been feeling cheerful”.

Anxiety and Depression Symptoms

The Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983) requires participants to rate 14 statements using a 4-point scale, 0 (never) to 3 (most of the time) with higher scores indicating higher levels of symptoms [depression 0–21, e.g., “I look forward with enjoyment to things”; anxiety 0–21, e.g., “worrying thoughts go through my mind”].


Screening revealed one univariate outlier (z-scores > 3.29 SD from the mean) but no multivariate outliers (Mahalanobis distances > X2 (7), 24.32, p < 0.001) thus all cases were retained. Nonparametric re-sampling was employed to adjust for non-normality (positive skew depression/anxiety data) in main analyses (Preacher & Hayes, 2008).

Table 1 displays descriptive statistics and significant associations between Social Media Stress (SMS), wellbeing/symptoms and Emotional Intelligence (Trait emotional intelligence TEI; ability emotional intelligence AEI-EM/AEI-EU) in the expected directions (Martins et al., 2010; Sánchez-Álvarez et al., 2016). TEI related robustly to greater wellbeing and fewer anxiety/depression symptoms, whilst AEI related moderately to fewer depressive symptoms only. Males reported higher levels of depression than females (M = 7.30, SD = 4.07, t (199) = 2.44, p < 0.05). Age was negatively associated with SMS (r = -0.21, p < 0.01) and AEI (understanding: r = -0.19, p < 0.01; management: r = -0.15, p < 0.05. Social media use (5-point scale from 0 [< 30 min] through to 5 [6 + hours]) was significantly associated with SMS (rs = 0.21, p < 0.01). These patterns were clarified through OLS regression analyses which tested combined predictive models for each of the three mental health outcomes (with age, sex, social media use, EI and social media stress as predictors—see supplementary materials, Table 1). Wellbeing, depression and anxiety were predicted by social media stress and TEI, whilst AEI emotion management and emotion understanding predicted fewer depression symptoms only. All other variables were non-significant.

Table 1 Bivariate correlations, reliabilities and descriptive statistics for predictor variables

The Moderating Effect of EI on Social Media Stress and Mental Health

Nine path models estimated the effects of social media stress on mental health outcomes (wellbeing; anxiety; depression symptoms), with EI as moderator (TEI; AEI emotion understanding; AEI emotion management) using PROCESS for SPSS version 3.5.3 (Hayes, 2018). Variables were mean centred prior to analyses. Control variables (age; sex; social media use), EI, social media stress (SMS) and the interaction term (EI x SMS) were regressed on each mental health indicator. Neither measure of AEI interacted with SMS to predict mental health (see supplementary materials, Tables 2 and 3). However, TEI significantly modified the effect SMS on depression (F (6, 194) = 23.46, p < 0.001; R2 = 0.421, with the interaction term contributing to 1.3% of the change in variance, ΔR2 = 0.013, F (1, 194) 4.34, p = 0.03 (see Table 2). Figure 1 shows the interaction probed at conditional values of trait EI (+ 1 SD; -1 SD mean). In those with lower levels of trait EI, the effect of social media stress on depression is amplified (β = 0.13, p < 0.001 [0.06, 0.20]) relative to those higher in trait EI (β = 0.04, p = 0.30 [-0.03, 0.10]). Interactive TEIxSMS models predicting wellbeing and anxiety symptoms were n.s., see supplementary material).

Table 2 Moderating effect of trait EI on the relationship between social media stress and depression symptoms
Fig. 1
figure 1

Data plot of the simple slope interaction for trait emotional intelligence (Trait EI) x social media stress on depression


Our preliminary data show that emotionally intelligent individuals experience reduced levels of social media stress. However, on the potential usefulness of EI for mitigating the effects of social media stress (SMS), findings are outcome and EI specific; only TEI directly moderated the effects of social media stress, and for depression symptoms only. Whilst strategic AEI (emotion management; understanding) could directly predict depression symptoms (in line with Martins et al., 2010), neither skill significantly interacted with SMS to predict mental health outcomes, which accords with previous findings of the specificity of interactive effects for chronic stressors (e.g., Davis & Humphrey, 2012a).

This suggests that emotional self-confidence, rather than emotional skill, is important for maintaining mental health when experiencing SMS. This is in line with findings that TEI, not AEI, is effective for managing mood deterioration in socially stressful situations (Lea et al., 2019) and for managing negative affect online (Preston et al., 2021). ‘Offline’ stress research shows that having a clear and stable sense of self is important for maintaining manageable levels of negative affect through cognitive appraisals (e.g., Lee-Flynn et al., 2011), and that higher levels of emotional self-confidence (as TEI) may support effective implementation of adaptive coping techniques to reduce the impact of stressors on mental health (Davis & Humphrey, 2012b). More research is needed to unpack these mechanisms in relation to online stress. Through experimental investigation of social media stress in real-time, with physiological markers to enhance self-report data, we may also find AEI plays a role in mood recovery in situ (see e.g., Lea et al., 2019). Since we explored strategic AEI only, other experiential components – perceiving and using emotion—could now also be examined using this fine-grained approach. It will also be important to explore the potential interaction of EI skills and self-confidence to predict lower levels of depression over time.

We offer the first, preliminary analysis of relationships between EI and social media stress—construed as a composite of dependency and negative affect arising from engagement with social media. However, this work should be considered exploratory owing to the modest sample size and the potential role that common method variance (i.e., multiple self-report; single-respondent; single-occasion) may play in artificially inflating relationships. Harman’s single-factor test (Podsakoff & Organ, 1986) identified multiple factors underlying our criterion and predictor variables (eigenvalues > 1 = 32; first factor accounting for only 13.73% cumulative variance), indicating that our data were not unduly affected by this source of bias. Moreover, TEI-mental health associations do appear robust to socially desirable responding (Choi et al., 2011) and criterion contamination (Williams et al., 2010). However, these questions should now be definitively addressed in a larger data set utilising additional, multimodal measures of EI, and objective social media use alongside self-report (e.g., Johannes et al., 2021).

Whilst preliminary, our findings offer tentative support to the rationale for training TEI, and we suggest benefits may extend to managing virtual as well as in-person contexts.