The effect of social media engagement on social integration of elderly migrants in China: the mechanism of perceived social support and psychological resilience

China’s rapid urbanization and internal migration have led many older parents (trailing parents) to relocate to cities to support adult children or retire, while retaining their rural hukou. These older migrants face barriers to integrating into host communities due to institutional constraints (hukou), reduced access to services, declining functional capacity with age, cultural and linguistic differences, and disrupted social networks. Social media usage among older adults has grown markedly in China and may support information access, social connection, and leisure. Yet, the linkage between social media use and social integration for trailing parents, and its psychological mechanisms, remains underexplored. This study asks what utilitarian goals motivate older migrants’ social media use, and how social media engagement promotes social integration via perceived social support (external buffer) and psychological resilience (internal resource). The study applies MSD theory and the Stress-buffering hypothesis, modeling relations among social media dependencies (understanding, orientation, play), social media engagement versus intensity of use, perceived social support, psychological resilience, and social integration.

Theoretical background: Media System Dependency (MSD) theory conceptualizes how individuals depend on media to achieve goals; in digital contexts, dependency extends to social media. Life changes (e.g., migration) increase environmental ambiguity and motivate information seeking through media. This study operationalizes three dependency relations—understanding, orientation, and play—and posits each fosters social media engagement (exposure condition), distinct from mere intensity of use (selective exposure). Stress-buffering hypothesis emphasizes perceived social support as a protective mediator/moderator under stress, more predictive than received support. Prior work links social media use to perceived social support and migrants’ integration. Literature and hypotheses: Social media can fulfill understanding (information/news, identity), orientation (decision-making, skills, services), and play (entertainment, distraction) needs; thus H1–H3 posit positive effects of these dependencies on engagement. Prior studies suggest intensity of use may not enhance integration; excessive passive use may even hinder it, whereas active, interactive engagement improves integration—leading to H4 (negative effect of intensity on integration) and H5 (positive effect of engagement on integration). Social media engagement is expected to raise perceived social support (H6), which should increase integration (H7) and mediate the engagement–integration link (H8). Social media engagement may also enhance psychological resilience (H9), which predicts integration (H10) and could mediate engagement–integration (H11). Perceived social support is expected to bolster resilience (H12), implying a sequential mediation of engagement → perceived support → resilience → integration (H13).

Design and sampling: Cross-sectional online survey using Tencent Questionnaire panel. Target population: older migrants (age 55+) in mainland China who followed adult children to urban areas for 6+ months for caregiving or retirement, retained original hukou, and used social media for 2+ months. Cities and data collection: Beijing, Xi’an, Nanjing, and Shenzhen; June 20–July 15, 2023. Screening ensured eligibility; responses with duration >1 minute retained. Sample: N=1001 valid responses (Beijing 297; Xi’an 144; Shenzhen 398; Nanjing 162). Demographics: 55.8% female; ages 55–64 majority (81.8%); 88.5% married; education largely middle-to-high school; ~90% migrated >1 year; 71.5% receive a pension. Measures: - Social media dependencies (understanding, orientation, play): adapted 5-point Likert subscales (Carillo et al., 2017); each 6 items; Cronbach’s alpha: 0.817, 0.805, 0.805. - Intensity of social media use: single item (7-point; Zhao et al., 2021b). - Social media engagement (HOC, reflective–reflective): 3 dimensions from Wei & Gao (2017): information production (4 items, α=0.792), information retrieval (4 items; one item deleted; α=0.700), social activities (5 items, α=0.739); 5-point frequency scale. - Perceived social support (HOC, reflective–formative): MSPSS (Zimet et al., 1990): significant other, family, friends (each 4 items; 7-point scale), plus one global single item for redundancy analysis; α: 0.895, 0.887, 0.877. - Psychological resilience: CD-RISC-10 (Campbell-Sills & Stein, 2007), 7-point scale; α=0.929. - Social integration: 4 items (Wei & Gao, 2017), 5-point Likert; α=0.819. Translation used back-translation; pre-test with six participants informed questionnaire refinement. Analysis: SPSS 27 for descriptives; PLS-SEM (SmartPLS) for measurement and structural modeling given theory development focus, higher-order constructs, and non-normal data. Common method bias: Harman’s single factor 30.74% (<50%); full collinearity VIFs 1.106–1.949 (<3.3). Measurement model: Adequate reliability (CR >0.8), convergent validity (outer loadings mostly >0.7; AVE >0.5), acceptable discriminant validity (HTMT mostly <0.85/0.9). One IR item deleted; SA3 retained with loading <0.7 but acceptable overall construct reliability. HOCs assessed using disjoint two-stage approach. PSS HOC validated via redundancy analysis (β=0.755 to global item), VIFs <3.3, significant outer weights. Structural model: Inner VIFs 1–2.417; bootstrapping 5000 subsamples for hypothesis tests. Predictive accuracy: R^2—SME 0.300, PSS 0.169, PR 0.478, SI 0.390; Q^2—SME 0.198, PSS 0.123, PR 0.288, SI 0.245.

Direct effects (standardized coefficients, p-values): - Understanding → Social media engagement (SME): β=0.282, p<0.001 (small effect size f^2=0.061). - Orientation → SME: β=0.192, p<0.001 (small f^2=0.022). - Play → SME: β=0.150, p<0.001 (f^2=0.015, negligible). - SME → Perceived social support (PSS): β=0.412, p<0.001 (moderate f^2=0.204). - SME → Psychological resilience (PR): β=0.088, p=0.001 (f^2=0.012, negligible). - PSS → PR: β=0.651, p<0.001 (substantial f^2=0.674). - SME → Social integration (SI): β=0.143, p<0.001 (small f^2=0.026). - PSS → SI: β=0.295, p<0.001 (small f^2=0.070). - PR → SI: β=0.315, p<0.001 (small f^2=0.085). - Intensity of use → SI: β=−0.043, p=0.052 (ns; f^2=0.003). Mediation: - SME → PSS → SI: significant partial mediation; indirect effect β=0.118 (p<0.001), variance accounted for (VAF)=47% (supports H8). - SME → PR → SI: indirect β=0.028 (p=0.002) but VAF=17.4% (<20% threshold), thus not a valid mediation (does not support H11). - Chain SME → PSS → PR → SI: significant partial mediation; indirect β=0.085 (p<0.001), VAF=39% (supports H13). Model fit/predictive power: R^2—SME 0.300; PSS 0.169; PR 0.478; SI 0.390. Key substantive insights: - Understanding and orientation are primary motives driving SME; play/entertainment has minor influence. - Engagement quality (active, diverse SME) predicts integration; mere intensity (time/frequency) does not and may even be counterproductive. - SME meaningfully elevates perceived social support; perceived support and resilience both directly enhance integration. - SME’s effect on resilience is small; resilience itself strongly linked to integration. - Perceived social support strongly boosts resilience, enabling a significant chain pathway from SME to SI.

Findings reinforce that for older migrants, social media is used primarily to reduce uncertainty (understanding) and guide behavior and interactions (orientation); entertainment (play) is secondary. The quality of engagement—active information production, retrieval, and social activities—matters for social integration, whereas time/frequency alone does not. Social media facilitates perceived social support by maintaining strong ties and expanding weak ties, overcoming dialect and mobility barriers; this supports the Stress-buffering hypothesis in a digital migration context. The limited effect of SME on resilience suggests that typical use among trailing parents is more passive and less interactive, which may not translate into substantial inner strength gains; nevertheless, resilience is positively associated with integration. The strong link from perceived social support to resilience highlights that external supportive contexts catalyze internal adaptive resources, producing a sequential pathway from SME to integration via support and resilience. The results extend MSD theory to an older migrant population by demonstrating that dependency relations differentially drive engagement and that engagement (exposure condition) rather than selective exposure predicts integration. Practically, enhancing older migrants’ active, purposeful engagement and strengthening online-to-offline support networks can improve belonging and participation in host communities.

The study validates MSD and Stress-buffering frameworks in the intersection of urban migration and digital communication among older migrants in China. Understanding and orientation dependencies drive social media engagement, which, in turn, promotes social integration; intensity of use does not. Perceived social support is the principal psychological mechanism linking engagement to integration, and together with resilience forms a sequential pathway: disadvantage → perceived social support → resilience → desirable integration. App designers should simplify interfaces and promote features that enable older users to produce content and interact meaningfully; communities should leverage social media to connect and mobilize older migrants; adult children should assist parents in diversifying and deepening their online engagements. The findings provide guidance for fostering older migrant-friendly digital ecosystems that translate into improved belonging and adaptation.

- Cross-sectional design limits causal inference; longitudinal multi-wave designs are recommended to establish temporal ordering (e.g., motivations → engagement → psychological mechanisms → integration). - Online convenience sampling may bias representativeness; probability sampling and face-to-face surveys could improve generalizability. - Social media engagement and social integration were treated as broad constructs; future work should parse specific activity types and distinct integration dimensions (cultural-social, economic, psychological). - The three dependency relations explain limited variance in engagement; additional drivers (psychological traits, family dynamics, socio-environmental factors) should be explored. - While perceived social support showed partial mediation and a sequential effect with resilience, other mediators should be investigated to refine the psychological process linking engagement to integration.