Understanding protective and risk factors affecting adolescents’ well-being during the COVID-19 pandemic

COVID-19-related school suspensions in Hong Kong replaced face-to-face schooling with extensive online learning and socialization, potentially affecting adolescents' studies and daily life. The study examines how adolescents' self-regulation in behavior (digital activities), emotions (cognitive-emotional regulation), and cognition (online learning self-efficacy), along with family factors (parent-child relationship quality and parental monitoring), during school suspension relate to their well-being after school resumption. The research addresses gaps regarding links between participation in digital-social activities (including cyberbullying), self-regulatory processes, and post-resumption worries. It also investigates whether improved parent-child relationships and parental monitoring during lockdowns supported adolescents' self-regulation, and explores sex and age differences through multiple-group analysis.

Prior research indicates increased engagement in digital entertainment/social activities during lockdown is associated with psychological distress and anxiety in adolescents and may heighten exposure to cyberbullying, which impairs well-being. Greater time online can detract from study time, lowering academic self-efficacy and increasing worries about learning outcomes. Emotional regulation strategies are protective against emotional problems and can bolster well-being and academic self-efficacy; resilience and positivity mitigate adverse effects of cyberbullying. Family systems affect adolescent outcomes via spillover: improved parent-child relationships and appropriate parental monitoring are linked to better emotional regulation and social-emotional wellness. Different parental monitoring styles can influence both academic and non-academic digital activities and relate to cyberbullying involvement. Sex and age differences exist in digital participation, cyberbullying, academic self-efficacy, psychological status, emotion regulation strategies, and parental involvement, and may moderate interactions among these factors.

Design and context: Part of Hong Kong's 'eCitizen Education 360' project, aiming to understand and support students' study and daily life in the pandemic-driven 'new normal.' Data were collected via anonymous online surveys (Qualtrics) from June 8 to July 14, 2020 (secondary schools). Ethical approval was obtained from The University of Hong Kong; informed consent was collected. Sample: 932 secondary students (23 schools) and one parent each. 66% junior secondary; remaining senior secondary. Mean age: females 14.80 (SD=1.56), males 15.00 (SD=1.60). Measures (student): Online learning self-efficacy (OSE; 6 items; 5-point Likert; adapted from SEQ-C); Perceived worries for future study and life (WOR; 6 items; 5-point Likert); Cognitive-emotional regulation strategies (CER; 3 items; 5-point Likert; from CERQ); Digital activities for socialization and entertainment (DSE; 5 items; frequency scale); Cyberbullying involvement (Cyl; roles as perpetrator, victim, both, bystander, never; 6 events; adapted instrument). Measures (parent): Improvement in parent-child relationship during suspension (PCR; 4 items); Parental monitoring of online activities (PM; 3 items); both on 5-point Likert scales. Psychometrics: CFA with robust maximum likelihood for all latent constructs except Cyl. Model fit criteria: CFI > .90, RMSEA and SRMR < .08; reliabilities (McDonald's ω): CER=0.71, DSE=0.74, OSE=0.82, WOR=0.73, PCR=0.84, PM=0.74. Cyl measured using the Nominal Response Model (NRM), estimating latent trait θ via EAP; empirical reliability 0.79; item parameter estimates and residual plots supported model fit. Analysis: Multiple-group structural equation modeling (MG-SEM) by sex (Mplus v8.1) to estimate direct and indirect effects, with 10,000 bootstrapped bias-corrected CIs for indirect effects. Measurement invariance testing followed a stepwise procedure (configural, metric, scalar); partial invariance adopted as needed (one DSE item freely estimated). MG-SEM overall fit: RMSEA=0.048, CFI=0.903, SRMR=0.065.

- Model fit and reliability were satisfactory across constructs; NRM for cyberbullying showed good fit and empirical reliability of 0.79. - Digital activities (DSE) during suspension were not directly associated with worries (WOR) after resumption for either girls or boys. - For girls: DSE negatively predicted OSE (B = -0.22, p < 0.01) and positively predicted cyberbullying involvement (Cyl; β = 0.15, p < 0.01). No direct Cyl→WOR effect (β = 0.06, p = 0.17). - For boys: DSE was not significantly related to OSE, Cyl, or WOR. Cyl positively predicted WOR (β = 0.20, p < 0.05). - Cognitive-emotional regulation (CER) positively predicted OSE for both sexes (girls: β = 0.47, p < 0.01; boys: β = 0.43, p < 0.01). For boys, CER negatively related to Cyl (β = -0.20, p < 0.05); not significant for girls (β = -0.04, p = 0.43). - OSE strongly and negatively predicted WOR (girls: β = -0.64, p < 0.01; boys: β = -0.42, p < 0.01). The OSE→WOR association was significantly stronger for girls than boys (difference = 0.35, p < 0.05), controlling for age. - Age effects: For boys, age was not significantly related to DSE, Cyl, CER, or OSE; the four predictors explained 24.2% of WOR variance. For girls, age positively related to DSE (β = 0.13, p < 0.01) and WOR (B = 0.11, p < 0.01), and negatively to Cyl (β = -0.16, p < 0.01); the four predictors explained 46.5% of WOR variance. - Family factors: For girls, improved parent-child relationship (PCR) during suspension was associated with higher CER (B ≈ 0.11, p < 0.05) and OSE (β = 0.18, p < 0.01), and lower DSE (β = -0.15, p < 0.01); PCR did not relate to Cyl (β = -0.05, p = 0.29). Parental monitoring (PM) showed no significant associations with CER, OSE, Cyl, or DSE. For boys, neither PCR nor PM significantly related to these variables. - Mediation: For girls, CER and PCR reduced WOR via OSE (CER: β = -0.35, p < 0.01; PCR: β = -0.11, p < 0.01). Girls’ DSE increased WOR via reduced OSE (β = 0.13, p < 0.01). PCR further reduced WOR via CER→OSE (β = -0.03, p < 0.05) and via DSE→OSE (β = -0.02, p < 0.05). For boys, CER reduced WOR via OSE (β = -0.19, p < 0.01). Group differences in indirect effects favored females for CER→OSE→WOR (difference = 0.16, p < 0.05), DSE→OSE→WOR (difference = -0.11, p < 0.05), and PCR→DSE→OSE→WOR (difference = 0.02, p < 0.05).

The study shows adolescents' post-resumption well-being is shaped by both individual and family-level processes during school suspension. Online learning self-efficacy (OSE) emerged as a central protective factor, strongly reducing worries (WOR), especially in girls, potentially reflecting developmental and cultural tendencies toward greater self-regulatory engagement and internal attributions. Leisure-oriented digital activities (DSE) did not directly increase worries but indirectly undermined girls’ well-being by reducing OSE and increasing cyberbullying involvement (Cyl). In boys, cyberbullying had a direct adverse link to worries, while DSE was not a significant predictor of OSE or Cyl. Cognitive-emotional regulation (CER) supported higher OSE in both sexes and directly protected boys by reducing Cyl. Family spillover effects were evident: improved parent-child relationships (PCR) during suspension benefited girls’ behavior (lower DSE), emotion (higher CER), and cognition (higher OSE), yielding indirect reductions in worries. These findings highlight the importance of strengthening adolescents’ self-regulatory capacities and fostering supportive family relationships to buffer the impacts of disruptions like pandemic-related school closures, with attention to sex- and age-specific patterns.

Adolescents’ well-being after COVID-19 school resumption was most strongly supported by higher online learning self-efficacy developed during suspension, with cognitive-emotional regulation and improved parent-child relationships serving as key upstream protective factors. Girls appeared more sensitive to the benefits of OSE and the risks of leisure-oriented digital engagement, while boys benefitted from CER in reducing cyberbullying and its downstream worries. Interventions should prioritize building students’ self-efficacy for online/self-directed learning, teaching positive emotion regulation strategies, guiding healthy digital habits, and nurturing emotionally secure parent-child relationships. Future work should examine causal pathways longitudinally, incorporate both parent and child perspectives on parenting, and explore contextual moderators (e.g., cultural norms, school supports) and developmental differences.

Parent-reported measures of parent-child relationship and parental monitoring may not align with adolescents’ perceptions; future studies should compare multi-informant perspectives. The cross-sectional design limits causal inference; longitudinal designs are needed to establish temporal ordering and causality. Generalizability may be limited to Hong Kong secondary students during the specific pandemic period.