Beyond structural inequality: a socio-technical approach to the digital divide in the platform environment

The paper examines how social media platforms, through human-technology interactions, shape contemporary forms of the digital divide beyond traditional structural explanations. Although participatory cultures and quantified attention (e.g., followers, retweets) seemingly enable broader engagement and mobility, technology diffusion patterns concentrate digital privilege among active platform participants, producing usage and outcomes gaps. Conventional digital divide research emphasizes structural reproduction (e.g., demographic and socioeconomic factors determining access, skills, and opportunities), but the platformed society adds dynamics such as algorithms, datafication, and digital capital that are not fully captured by structural accounts alone. In China’s rapidly expanding platform ecosystem—exemplified by Sina Weibo’s hundreds of millions of users—rapid digitization coincides with multifaceted stratification, highlighting the need to understand socio-technical mechanisms. The study proposes a platform affordance framework that links technological properties and human agency via two efficacy constructs: technology-efficacy (perceived accessibility/availability of functions) and self-efficacy (perceived capabilities and needs). It focuses on digital capital (followers) as a key outcome. The main hypotheses are: H1, technology-efficacy positively influences self-efficacy; H2, personal and positional characteristics moderate the technology-efficacy to self-efficacy link; and H3a/H3b, platform affordance contributes to stratified usage and, alongside usage patterns, to stratified digital capital.

Theoretical background covers the three levels of the digital divide—access (first level), skills/use (second level), and benefits/outcomes (third level)—noting their persistence in China (urban–rural, regional disparities, skill deficits, usage gaps, and unequal benefits). Social media platforms introduce new divide issues: a participation culture with user-generated content coexists with a pronounced gap in online content creation (OCC), distinguishing creators from consumers. OCC can be analyzed along decision to create, frequency, and diversity of creation. Research finds youth, males, and higher SES more engaged. Digital capital—externalized resources and internalized abilities—has become central; in narrow terms, platform metrics (e.g., followers) constitute portable digital identity and value, subject to Matthew effects. Traditional stratification mechanisms (gender, age, education, occupation, income, ethnicity) shape opportunities and outcomes, yet platform environments complicate this picture due to intermediary roles, opaque algorithms, and symbolic features. Prior work often treats affordances either as platform attributes (portability, replicability) or user-perceived outcomes; integrating both clarifies how socio-technical interactions influence inequalities. The authors argue for returning to a Gibsonian affordance perspective, emphasizing mutuality and perception, while acknowledging institutional and demographic influences on behavior in platform contexts.

Design and sample: An Internet-based survey was conducted in 2021 with Hanyi Big Data. Quotas aligned with Chinese netizen and Weibo user demographics (gender, age, education). From 1,737 valid adult questionnaires, 903 Sina Weibo users formed the analytic sample. For platform-relevant demographics, age was categorized (18–30, 31–45, >45) and education into four levels (≤senior high, some college, bachelor’s, master’s+). Measures: Platform functions were grouped by target users, button design, and physical properties into basic, simplified, multimedia, and topic functions. Technology-efficacy captured perceived accessibility and availability of functions via difficulty of finding icon locations and operation methods for the four function groups (two 1–5 scales per group; averaged across eight items; Cronbach’s α=0.93, M=4.05, SD=0.63). Self-efficacy measured perceived capabilities (proficiency) and needs/value for the four function groups using 5-point scales; averaged across eight items (Cronbach’s α=0.90, M=3.82, SD=0.62). Stratified uses operationalized OCC via: creation or not (74.75% engaged), creation frequency (1–7; M=4.01, SD=1.79), and creation diversity (count of creative functions used, 1–8; M=3.36, SD=1.66). Stratified outcomes used follower count (1–5 categorical; M=1.55, SD=0.73) as narrow digital capital. Demographics included gender, age, education, occupation, and income; internet experience variables controlled access and skill gaps (number of devices, digital skills). Analysis: Multiple linear regression tested H1 (technology-efficacy → self-efficacy) and marginal effects probed H2 (moderation by demographics). Logistic (creation yes/no), ordered logistic (creation frequency), and Poisson (creation diversity) nested models compared structural-only (Model 1) with socio-technical specifications (Models 2–3) to assess H3a. Ordered logistic models for outcomes assessed direct effects of socio-technical variables (Model 1) and added OCC behaviors sequentially (Models 2–4) to test H3b. Model fit was compared via AIC/BIC. Robustness checks are reported in appendices.

- H1 supported: Technology-efficacy positively predicts self-efficacy. Users who perceive function icons and operation methods as more accessible/available report higher perceived capabilities and needs. - H2 partially supported: Gender, age, and education moderate the technology-efficacy → self-efficacy link. Lower technology-efficacy corresponds to heightened gender inequality; the largest generational gaps are between 18–30 and >45; educational differences narrow as technology-efficacy increases. No clear moderating effects were found for social position (occupation/income). - Stratified uses (H3a): • Creation (yes/no): Older (>45) and unstable occupation groups are less likely to create. Technology-efficacy is positively associated with being a creator (Model 2), but becomes insignificant after adding self-efficacy (Model 3); self-efficacy significantly increases the odds of being a creator. Structural differences shrink when socio-technical variables are included. • Creation frequency: Females post more frequently; higher education (master’s+) is associated with lower frequency in Model 1. Technology-efficacy is positive in Model 2 but loses significance once self-efficacy is included (Model 3); self-efficacy remains positive. • Creation diversity: SES-related variables matter; both technology-efficacy and self-efficacy are positively associated with using a wider range of creative tools. Technology-efficacy also indirectly predicts diversity through self-efficacy. • Across models, socio-technical variables provide better fit (lower AIC/BIC) than structural-only models, indicating platform affordance surpasses structural factors in explaining usage gaps. Self-efficacy is a key intermediary between technology-efficacy and online participation. - Stratified outcomes (H3b): • Occupation and income relate to follower count (e.g., private institution employment and average income show positive associations). Technology-efficacy does not directly predict followers. • Self-efficacy is positively associated with followers in models without usage controls, but its effect becomes non-significant after controlling for creation frequency and diversity. • Being a creator and, among creators, posting more frequently and using more diverse creation tools are strongly associated with accumulating more followers. • Models including platform affordances and usage patterns fit substantially better (e.g., AIC decreases from 1602.01 in structural-only to 1264.30 when adding frequency and diversity), evidencing that affordances shape outcomes primarily by mobilizing and structuring participation. Overall, active, frequent, and diverse creators accumulate more digital capital (followers).

The study elucidates how platform affordance operates as a sequential mechanism: technology-efficacy defines the boundaries of action possibilities by making function surfaces and layouts perceptible, while self-efficacy activates engagement based on perceived capabilities and needs. This socio-technical interplay explains persistent hierarchical distinctions in participatory cultures and in the distribution of digital capital. Motivation and attitudes emerge as central psychological gateways to inclusion, helping explain digital disconnection and dropouts. An unexpected pattern shows highly educated users report lower self-efficacy in an entertainment-oriented platform context, reflecting platform vernacular and user–function perceptions rather than lack of skills. Structural inequality, while less dominant in direct usage/outcomes models, moderates the translation of technology-efficacy into self-efficacy across gender, age, and education, indicating a reconfigured role of social structure via affordance. Affordances function in a graded fashion—request, allow, encourage, discourage, refuse—thus channeling different user groups into distinctive action levels and usage patterns. Network effects and widely perceived function properties shape dominant action possibilities, promoting competition and potential monopolization of platform resources. Narrow digital capital (followers) is partly decoupled from offline status and is primarily accumulated through active, frequent, and diverse creation; affordances guide but do not determine outcomes, consistent with Gibson’s view that affordances suggest or encourage actions rather than fix achievements.

The paper advances a socio-technical framework of the digital divide centered on platform affordance, demonstrating a sequential link from technology-efficacy to self-efficacy and, in turn, to stratified usage and outcomes on Sina Weibo. It shows that socio-technical factors often surpass traditional structural variables in explaining who creates, how frequently, with what diversity, and who accumulates followers as digital capital. The framework clarifies how social structures and technological features co-construct platform environments, revealing both bridging and generative dynamics of inequality. For future research, the authors propose cross-platform comparisons and panel designs to capture dynamic user–algorithm co-evolution, resource conversion over time, and the heterogeneous impact of different technological configurations on the digital divide.

- Cross-sectional design limits causal inference and the ability to trace resource conversion paths and long-term reinforcement of inequalities. Panel or longitudinal studies across cohorts/generations are recommended. - Self-reported measures of function accessibility/visibility and perceived capabilities/needs may introduce biases and may not perfectly reflect objective technological properties. - Opaque algorithmic recommendation systems and platform configurations may reinforce or create inequalities not fully captured by the survey, constraining the depiction of actual inequality mechanisms without more sophisticated methods or trace data.