Mental wellbeing effects of disclosing life events on social media

Life events—significant changes in personal or social circumstances—can alter behavior, emotions, and wellbeing. With social media’s rise, many individuals disclose these experiences online (e.g., Facebook), yet the mental wellbeing consequences of such disclosures remain underexplored. Prior work offers mixed evidence on social media’s impact: associations with mental health risks (e.g., cyberbullying, addictive use) versus benefits such as social support, decreased loneliness, and increased life satisfaction. Much existing research relies on aggregate use metrics (e.g., screen time) and cross-sectional self-reports, obscuring what people actually do online and the psychological salience of specific behaviors. Life event disclosures are psychologically meaningful, involving vulnerability and identity. The central research question is: How does disclosing a life event on social media impact an individual’s mental wellbeing? Addressing this helps reduce bias in digital behavioral health assessments by accounting for offline life events and informs platforms and interventions about potential benefits or risks of disclosure.

The paper surveys contrasting evidence on social media and wellbeing. Detrimental aspects include links to mood/anxiety disorders, cyberbullying, and addictive use, though some longitudinal work finds no worsening effects from increased use. Benefits include therapeutic potential, reduced loneliness, and increased life satisfaction, particularly via self-disclosure, peer support, and self-reflection. Research on sensitive life event disclosures (e.g., gender transition, bereavement, childbirth, job loss, pregnancy loss) highlights complex decisions around disclosure and non-disclosure. Theoretical lenses invoked include Uses and Gratifications (media use to fulfill psychosocial needs), Social Capital (bonding and bridging support), and the Stress Buffering Hypothesis (social support mitigating adverse effects of stress). Methodological critiques of prior work emphasize overreliance on cross-sectional data, self-reports, and coarse use metrics, motivating analysis of specific online behaviors like life event disclosures. The paper also notes positivity bias in social platforms and risks of social comparison, which may dampen wellbeing effects of positive disclosures.

Study context and participants: Data were sourced from the Tesserae project, a longitudinal passive sensing initiative of U.S. information workers, with 236 participants providing year-long Facebook activity and responding to periodic wellbeing surveys (positive/negative affect via PANAS-Short, stress and anxiety single-item 1–5 scales, sleep hours via MITRE scale) and an exit PERI life events survey. Participants also provided baseline demographics, cognitive ability (Shipley Abstraction and Vocabulary), personality (BFI-2), trait affect (PANAS-X), trait anxiety (STAI), and trait sleep quality (PSQI). Ethics: IRB-approved, informed consent for each modality, de-identified data stored securely. Life event identification and attributes: Life events were identified and coded in prior work using a PERI-derived codebook and multi-phase annotation with substantial reliability (Fleiss’ K≈0.71). Attributes included type (School, Health, Personal, Financial, Work, Local), valence (negative/neutral/positive scaled −1 to +1), intimacy (Low/Medium/High), anticipation (anticipated vs unanticipated), temporal status (continuous vs discrete), scope (low/medium/high directness), and significance (scaled 0–1). Design and causal inference approach: A quasi-experimental design with potential outcomes and Difference-in-Differences (DiD) was adopted. Treatment was the occurrence of a life event; outcomes were changes in self-reported wellbeing. For each treated instance, wellbeing was measured in the two-week window before and after the event, and compared to a synthetic control constructed via placebo dates (50 randomized dates per individual outside the treatment windows). DiD was computed as (after−before in treated) minus (after−before in controls), averaged to yield the Average Treatment Effect (ATE). Statistical tests included Cohen’s d, paired t-tests, and KS-tests. Parallel trends were visually assessed and appeared plausible. Analytic models: 1) Regression models linked DiD changes in wellbeing (positive affect, negative affect, stress, anxiety, sleep) to individual differences (demographics, cognitive ability, personality, trait affect/anxiety/sleep) and life event attributes, including an indicator for Facebook disclosure. 2) A post-hoc analysis modeled the relationship between social media engagement (number of reactions and comments on life event posts) and wellbeing outcomes, controlling for the same covariates and event attributes. 3) Engagement differences were examined between life event posts and other posts (Cohen’s d, paired t-tests, KS-tests), and across valence groups (Kruskal–Wallis tests). Missing data in daily wellbeing surveys were imputed using individual mean scores. The study focused on short-term (two-week) effects.

Primary wellbeing effects of disclosure (regression, DiD): Disclosing life events on Facebook was associated with improved wellbeing. Regression coefficients for the Facebook disclosure indicator: positive affect +0.12 (p<0.05), negative affect −0.32 (p<0.01), stress −0.25 (p<0.001), anxiety −0.18 (p<0.001), sleep +0.20 (p<0.05). Models had significant fit (R²≈0.18–0.30 for affect/stress/anxiety/sleep). Event attributes showed expected relationships (e.g., negative valence increased negative affect and stress; continuous status had mixed associations). Average treatment effects by valence and disclosure (two-week windows):

• Negative events: If not disclosed, negative affect tended to increase (ATE=+0.36, d=0.19). If disclosed, wellbeing improved: negative affect ATE=−1.53 (d=−0.73, t=−4.93***), stress ATE=−0.42 (d=−0.62, t=−4.21***), anxiety ATE=−0.40 (d=−0.74, t=−5.04***), sleep ATE=+0.57 (d=0.40, t=2.66**).
• Neutral events: No significant changes when not disclosed. When disclosed: improvements in negative affect (ATE=−0.59, d=−0.42, t=−2.09***), stress (ATE=−0.20, d=−0.37, t=−1.98**), anxiety (ATE=−0.23, d=−0.57, t=−3.04***), sleep (ATE=+0.45, d=0.41, t=2.39**).
• Positive events: No significant changes whether disclosed or not (p>0.05 across measures). Engagement and wellbeing (regression within life event posts): Higher engagement related to better wellbeing. Number of reactions positively associated with positive affect (+0.01, p<0.05) and negatively with negative affect (−1.7E−3, p<0.01), stress (−1.7E−3, p<0.05), and anxiety (−1.0E−4, p<0.05). Number of comments positively associated with positive affect (+3.4E−3, p<0.05) and sleep (+4.2E−3, p<0.05), and negatively with anxiety (−1.0E−4, p<0.05). Models had significant fit (R²≈0.26–0.60). Engagement differences: Life event posts received substantially more engagement than other posts from the same users: reactions +131.96% (means 21.44 vs 9.24; d=0.60; t=30.81***; KS=0.34***), comments +134.86% (means 4.12 vs 1.75; d=0.36; t=21.89***; KS=0.19***). By valence, negative event posts had fewer reactions (mean 11.25) than neutral (25.93) and positive (22.53) but more comments (mean 9.18) than neutral (3.52) and positive (3.38) (Kruskal–Wallis H-tests p<0.001).

Disclosing life events on Facebook is associated with short-term improvements in mental wellbeing, particularly for negative events, addressing the core question of whether platforms are conducive to sharing personally meaningful experiences. The findings support theories that disclosure fulfills psychosocial needs (Uses and Gratifications) and activates social capital, while engagement (reactions/comments) may provide stress-buffering support that mitigates adverse emotional impacts. The pronounced benefits for negative and neutral events suggest protective effects: comments on negative disclosures may function as empathic, informational, or validating support, improving affect, reducing stress/anxiety, and aiding sleep. The work argues for a nuanced view of digital wellbeing beyond coarse use metrics, highlighting the importance of behavior-specific analyses and the integration of offline event context in interpreting online data. These insights are relevant to mental health practitioners, researchers, and platform designers seeking to harness social media’s supportive potential while acknowledging its risks.

This study advances understanding of the mental wellbeing consequences of life event disclosures on social media. Using a quasi-experimental DiD design in a longitudinal cohort, it shows that disclosing life events—especially negative ones—on Facebook is linked to increased positive affect and sleep, and decreased negative affect, stress, and anxiety. Life event posts elicit more engagement than other posts, and negative events draw relatively more comments, consistent with supportive interactions underpinning protective effects. Contributions include: empirical evidence of short-term protective effects of disclosure, characterization of how event valence and engagement relate to wellbeing, and a methodological perspective centering psychologically meaningful online behaviors integrated with offline events. Future research should test longer-term trajectories, incorporate clinical assessments and measures of offline social support, examine content quality and reaction types, and extend analyses to other platforms and modalities (e.g., short-form video), using designs that strengthen causal inference.

• Quasi-experimental design cannot establish true causality; lack of true counterfactuals and reliance on placebo dates and DiD with visually assessed parallel trends.
• Potential endogeneity and unobserved confounding (e.g., offline social support networks, number of Facebook friends) may bias estimates; time-varying confounders can affect DiD validity.
• Short-term (two-week) assessment may not capture longer-term effects of disclosure.
• Engagement measured as quantity; content/quality of comments not analyzed; reaction types (e.g., like, sad, angry) unavailable.
• Possible misalignment between event occurrence and disclosure dates; recall bias in self-reported life events; missing daily surveys imputed by individual means.
• Limited to one platform (Facebook) and a pre-COVID 2018–2019 dataset; platform norms, affordances, and user behaviors may have changed; generalizability to other platforms uncertain.
• Self-selection bias in participants consenting to share data; sample of information workers may limit representativeness.
• Social desirability, platform norms, and selective self-presentation may influence disclosures; positivity bias may dampen effects for positive events.
• Lack of baseline clinical mental health diagnoses; anticipation duration effects not disentangled; interaction terms explored but not significant, favoring simpler models.