Information Technology, Urban–Rural Health Disparities and Pathways to Sustainable Development: Evidence from the 2023 Chinese General Social Survey

Abstract

Objective: Although information technology’s impact on social inequality has attracted widespread attention, existing research remains divided on whether technology generates “digital dividends” or exacerbates the “digital divide”. Most studies assume uniform effects across all life domains; yet, the mechanisms through which technology affects different aspects of social stratification may vary substantially. Method: Using data from the 2023 Chinese General Social Survey (CGSS) covering 5332 respondents, we examine information technology’s differential effects on economic income and health outcomes between urban and rural residents through multiple regression, interaction analysis and Blinder–Oaxaca decomposition. Result: Our findings reveal that information technology significantly enhances both income and health status, validating the digital dividend hypothesis. However, technology effects exhibit domain-specific patterns: urban–rural differences are modest in economic domains but pronounced in health domains, where rural residents benefit significantly more than urban counterparts. Information technology contributes 10.8% to urban–rural income gaps but plays larger roles in health disparities. Conclusion: We propose a digital dividend differentiation theory, whereby technology effects tend toward homogenization in standardized economic domains governed by market logic while exhibiting compensatory functions in health domains dependent on geographic proximity. These findings challenge the assumptions of technological uniformity and provide foundations for differentiated digital development policies. Moreover, by revealing how digital technologies can reduce structural inequalities in income and health, this study advances the understanding of digital inclusion as a core driver of sustainable urban–rural development, offering both academic contributions to sustainability scholarship and practical guidance for achieving inclusive development goals.

1. Introduction

The relationship between information technology and social stratification in contemporary China presents us with a puzzle of considerable complexity. On the one hand, we witness remarkable success stories: rural live streamers achieving dramatic income increases through digital platforms, internet-based medical services opening new avenues for healthcare delivery in remote areas. On the other hand, we observe the persistent marginalization of digitally impoverished populations amid waves of technological innovation and the stubborn persistence of urban–rural digital divides. This paradoxical landscape suggests that the impact of information technology on social stratification is far more nuanced and multidimensional than conventional theoretical frameworks have been able to capture. Information technology functions neither as the automatic equalizer envisioned by technological optimists nor as the simple amplifier of existing inequalities feared by technological pessimists. Rather, it operates more like a complex system that produces differentiated effects across varying social contexts [1,2].

The existing literature offers contradictory answers to these questions, and this theoretical divergence itself warrants careful examination. Some studies document that information technology can significantly reduce urban–rural gaps, providing empirical support for the inclusive effects of digital dividends. Others confirm the tenacious persistence of the digital divide, arguing that technological progress may replicate or amplify existing structures of inequality. What is striking about this divergence is that it may stem from a commonly overlooked problem: the prevalence in existing research of what we might call “technological uniformity assumptions” and “impact mechanism homogenization assumptions”—that is, treating information technology as a uniformly acting exogenous variable and presupposing that it should exhibit consistent patterns of effects across all domains of social life. While such assumptions provide analytical convenience, they may obscure the true complexity of technology–society relationships [3,4].

This study attempts to break free from this theoretical impasse by proposing what we believe is a fundamentally different analytical strategy: treating economic income and health status as two parallel domains of information technology impact, rather than assuming simple linear causal relationships between them. Our central theoretical proposition is that the social effects of information technology exhibit what we term “domain-specific” characteristics—across different spheres of social life, technology may operate according to different logics and produce different patterns of stratification. More specifically, we advance a digital dividend differentiation theory: information technology effects in the relatively standardized economic domain are primarily governed by market logic, resulting in relatively modest urban–rural differences; meanwhile, in health domains, which depend more heavily on geographic proximity, technology’s compensatory functions are more pronounced in resource-scarce rural areas, producing significant urban–rural heterogeneous effects.

Through systematic empirical analysis based on the 2023 Chinese General Social Survey (CGSS), this study aims to contribute to scholarship in three key areas: theoretically, by providing a more nuanced analytical framework for digital inequality research that emphasizes the domain-specific nature of technology effects; methodologically, by developing a multi-level empirical testing framework through the integration of multiple statistical techniques; and in terms of policy implications, by providing empirical foundations for developing differentiated digitization strategies. We contend that only by acknowledging and understanding the complexity of information technology’s social effects can we provide more scientifically grounded and precise theoretical guidance for social development in the digital age.

2. Theoretical Debates and Analytical Framework

2.1. The Economic Effects of Information Technology: Universal Dividends or Accelerated Differentiation

The impact of information technology on economic outcomes has long been a source of academic controversy, with two opposing theoretical camps emerging around this fundamental question. From the optimistic perspective of digital dividends, information technology is viewed as an inclusive instrument for enhancing economic welfare. The human capital theory provides crucial theoretical support for this view, conceptualizing information technology as a novel form of production input that can substantially enhance worker productivity and market competitiveness [5]. Through theoretical modeling and empirical analysis, Cheng and Zhang [6] demonstrate that internet penetration promotes economic growth through multiple pathways, including enhancing human capital accumulation and optimizing resource allocation efficiency, thereby generating broadly shared income improvements. The market expansion theory further elucidates the economic promotion mechanisms of information technology, emphasizing that technological progress can transcend traditional geographical constraints and create new economic opportunities and channels for value realization. Xu et al. [7] provide micro-level empirical evidence that information technology reduces farmers’ search costs and enables them to secure better prices in market transactions, demonstrating the inclusive effects of what might be termed “information dividends”. Yang and Liu’s [8] enterprise-level analysis similarly supports this perspective, finding that companies implementing “Internet Plus” strategies significantly outperform traditional enterprises in both earnings per share and return on assets.

International research provides additional empirical support for the digital dividend theory. Siegel and Indjikian’s [9] cross-national comparative study documents significant positive effects of information technology investment on economic performance, with these effects being particularly pronounced in developing countries. DiMaggio and Bonikowski’s [10] analysis of U.S. household data confirms a significant positive correlation between internet use and income growth, validating the continued relevance of the human capital theory in the information age. Cheng et al.’s [11] panel data analysis of 72 countries further demonstrates that the diffusion of information and communication technologies promotes national economic growth, with mobile device penetration having particularly important implications for economic development in middle- and low-income countries.

However, the critical perspective embodied in the digital divide theory presents a formidable challenge to such optimistic assessments. This theoretical framework emphasizes that information technology may exacerbate rather than ameliorate existing economic inequalities. Analyzing it from the vantage point of internet capital, Qiu et al. [12] argue that the essence of the digital divide lies in differential access to and utilization of internet capital across social groups, leading to “dividend differentiation” rather than “dividend universalization”. The Matthew effect theory suggests that groups with superior educational backgrounds, economic foundations and social capital are better positioned to capture technological dividends, thereby further widening the gaps with disadvantaged populations [13]. Through theoretical analysis, Parayil [14] demonstrates that income and wealth distribution in the information age exhibits a pronounced tilt toward those who control technology, while other groups face risks of exclusion.

The digital exclusion theory deepens our understanding of this phenomenon from institutional and structural perspectives. Bauer’s [15] research on American society documents particularly acute disparities in information technology resource distribution, with residents of different community types unable to participate equally in digital economic transformation. Fairlie’s [16] research further reveals the racialized characteristics of the digital divide, documenting significant income gaps and digital resource disparities between different ethnic groups. Chinese scholars Fang et al. [17] and Zhang et al. [18] provide evidence for similar concerns, documenting instances where information technology development has indeed widened urban–rural income gaps.

We recognize that economic income and information technology use exhibit bidirectional relationships, with economic resources facilitating technology adoption and technology use simultaneously enhancing earning potential. As Van Dijk [19] notes in his landmark review, the digital divide phenomenon involves multiple interconnected dimensions, including technology access, usage skills and socioeconomic status, and this complexity makes simple unidirectional causal analysis challenging. However, the policy-oriented nature of digital divide research means that researchers often need to strategically focus on specific causal pathways that address pressing theoretical and policy questions. DiMaggio and his colleagues [20] argue that research should be guided by sociological relevance rather than methodological comprehensiveness, particularly when investigating phenomena with immediate policy implications. Given the current importance of understanding how technology diffusion affects inequality outcomes, our analytical focus on technology’s impact represents a theoretically informed research choice. This approach aligns with established practices in empirical social research, where investigators routinely examine specific causal pathways within complex systems while acknowledging the broader network of reciprocal influences.

Drawing on this theoretical review, we propose Hypothesis 1.

Hypothesis 1.

Information technology access significantly enhances residents’ economic income, demonstrating “digital dividend” effects.

2.2. Health Effects of Information Technology: Multiple Pathways of Promotion

The mechanisms through which information technology affects health outcomes are considerably more complex and multifaceted than those governing economic effects. Comparative evidence from European countries provides robust empirical support for information technology’s health-promoting effects across diverse institutional contexts. Alvarez-Galvez et al. [21] analyzed 28 European countries using Flash Eurobarometer data, documenting significant health benefits from internet health information use despite persistent digital divides. Sørensen et al.’s [22] European Health Literacy Survey across eight countries similarly confirmed positive associations between digital health tool usage and health literacy enhancement. Notably, Kuźmar and Piątek’s [23] analysis of Central and Eastern European countries demonstrates stronger health improvement effects in regions with lower baseline digital development, providing cross-national evidence for diminishing marginal utility mechanisms that parallel our theoretical expectations regarding urban–rural differences.

Scholars have identified multiple parallel pathways through which these health benefits are realized. The enhancement of health information accessibility is widely recognized as the most direct mechanism through which information technology promotes health. Brodie et al. [24] emphasize that internet-based health information delivery has tremendous potential for improving public health knowledge levels across broad populations. Zhao and Liu’s [13] analysis of Chinese data demonstrates that internet use enables elderly individuals to acquire health-related knowledge, improve health literacy and thereby enhance their overall health status. Cohall et al.’s [25] research further documents the close relationship between computer use, internet access and online health information seeking behaviors, providing individuals with new channels for health information acquisition.

The improvement of medical service accessibility constitutes a second crucial mechanism of information technology’s health effects. Andreassen et al.’s [26] survey of internet health service utilization in Norway demonstrates that online medical services effectively complement traditional healthcare delivery and enhance healthcare accessibility. Reinfeld-Kirkman et al.’s [27] research documents a positive correlation between individual health status and the likelihood of seeking internet-based health information, indicating that information technology indeed provides new instruments for health management. Chinese scholars Lu and Wang’s [28] analysis based on China Family Panel Studies data confirms that residents’ internet use exerts significant positive effects on self-rated health, with these effects operating primarily through information acquisition and service accessibility channels.

Social support mechanisms constitute a third important pathway through which information technology affects health outcomes. Wang’s [29] research demonstrates that internet use affects elderly individuals’ physical and mental health primarily through social connections and emotional support. Song et al.’s [30] empirical analysis based on CHARLS data confirms that internet use significantly alleviates loneliness among elderly individuals, with effects operating primarily through social network expansion and enhanced social support. Cotten et al.’s [31] longitudinal analysis further establishes a causal relationship between internet use and reduced depression levels among retired elderly individuals.

Mental health promotion mechanisms have also attracted considerable scholarly attention. Jin and Zhao’s [32] analysis of China Longitudinal Aging Social Survey data demonstrates that internet use promotes active aging among elderly individuals, primarily through enhanced self-efficacy and life satisfaction. Hong et al.’s [33] research similarly confirms the positive effects of internet use on elderly individuals’ mental health. International research provides comparable evidence: Cresci and Jarosz [34] document through community-based intervention projects that helping urban elderly individuals bridge the digital divide significantly improves their health status.

Nevertheless, some scholars have raised concerns about information technology’s health effects. Zhu et al.’s [35] research suggests that internet media use may negatively affect perceptions of social fairness, potentially creating indirect negative effects on mental health. Li’s [36] analysis of trends in social fairness perceptions similarly suggests that rapid information technology development may, under certain circumstances, exacerbate perceptions of social injustice, thereby negatively affecting mental health.

Despite these concerns, the preponderance of research supports positive effects of information technology on health. Rains [37] argues through theoretical analysis that broadband internet access enhances health communication quality and helps reduce health inequalities in the context of digital divides. Zhao and Li’s [38] analysis based on China Family Panel Studies data similarly confirms the positive effects of internet use on individual health.

Building on the theoretical foundations of health sociology, we propose Hypothesis 2.

Hypothesis 2.

Information technology access significantly improves residents’ physical and mental health status.

2.3. Urban–Rural Heterogeneous Effects: Theoretical Mechanisms of Differentiated Impact

Urban–rural differences, as a fundamental characteristic of China’s social structure, may play crucial moderating roles in the processes through which information technology effects are realized. However, the direction and magnitude of these moderating effects exhibit considerable variation across different studies.

The diminishing marginal utility theory provides a key theoretical framework for understanding urban–rural heterogeneity. This theory suggests that in resource-constrained environments, additional resources generate greater marginal utility. Analyzing this from a spatial sociology perspective, Lin and Liu [39] argue that China’s urban–rural development inequality is fundamentally a form of spatial inequality, with urban areas enjoying spatial advantages that enable resource concentration, while rural areas remain in relatively disadvantaged positions. Against this backdrop, the introduction of information technology may generate substantially greater marginal improvement effects in rural areas. Wen et al.’s [40] research on urban–rural differences in internet use suggests that while urban–rural gaps exist in internet adoption rates, rural residents may derive higher income returns from internet use.

The compensatory resource theory further elucidates this mechanism. This theoretical framework suggests that information technology functions primarily as compensation in areas characterized by traditional resource scarcity. Luo et al.’s [41] research documents that the internet’s impact on urban–rural income gaps exhibits a pattern of initial widening followed by narrowing, suggesting that equalizing effects gradually emerge as technology penetration increases. Analyzing this from the perspective of digital village construction, Lü [42] demonstrates that information technology can alleviate rural areas’ resource disadvantages by enhancing information production possibilities, information access capabilities and information utilization competencies.

However, the digital divide theory challenges such optimistic assessments. Lu and Wei [43] develop an analytical framework for elderly digital divide governance, arguing that the persistence of digital divides may further exacerbate urban–rural differences. Cheng et al.’s [44] case study of Beijing documents significant associations between digital divides and elderly individuals’ self-rated health, although these associations exhibit different patterns in urban versus rural contexts. Korupp and Szydlik’s [45] international comparative research similarly confirms that the causes and trends of digital divides vary significantly across regions.

The infrastructure constraint theory emphasizes the crucial importance of technological access conditions. Song and Liu’s [46] analysis of spatiotemporal patterns in China’s informatization development demonstrates significant regional differences in informatization levels, which may affect the realization of information technology effects. Kilenthong and Odton’s [47] research on information technology access in urban and rural Thailand confirms similar patterns, demonstrating that infrastructure conditions significantly affect technology utilization effectiveness.

International research provides additional empirical evidence for urban–rural heterogeneous effects. Jun et al.’s [48] comparative analysis of urban and rural elderly health status in Korea documents different patterns of urban–rural differences across various health indicators. Park’s [49] research further confirms that urban elderly individuals score significantly higher than rural elderly individuals on measures of physical activity, interpersonal relationships and stress management, although these differences may be ameliorated through technological interventions. Gloria et al.’s [50] research on very elderly individuals similarly documents complex age-moderated effects of urban–rural residential environments on health outcomes.

It is particularly noteworthy that urban–rural heterogeneous effects may exhibit different patterns across different domains of social life. In the economic sphere, the universality of market logic may promote effect homogenization, as returns to technical skills in labor markets are relatively standardized, and the diffusion of digital economy opportunities is relatively universal. In the health domain, urban–rural differences may be more pronounced, as urban–rural disparities in health resource allocation are more substantial, and the complexity of health needs is greater.

Building on the marginal utility and compensatory resource theories, we propose this study’s Hypothesis 3.

Hypothesis 3.

Information technology’s beneficial effects are greater for rural residents than for urban residents, but these effects may exhibit different patterns across economic and health domains.

2.4. Income–Health Linkage Mechanisms: Fundamental Causes or Parallel Processes

The relationship between income and health has long constituted a core concern in health sociology and health economics research. In the context of rapid information technology development, the nature and mechanisms of this relationship may be undergoing important transformations.

The fundamental cause theory of health inequality provides a classical theoretical framework for understanding income–health relationships. This theory treats socioeconomic status as the fundamental determinant of health differences, emphasizing that income affects health status through its influence on health resource access capabilities, health behavior choice spaces and health environment quality. Xie’s [51] research on income-related health inequality in China confirms the applicability of this theoretical framework in the Chinese context. Wang and Peng [52] further elucidate the mechanisms through which elderly individuals’ socioeconomic status affects health, documenting that income levels influence health outcomes through three mediating pathways: lifestyle, public service access and sociopsychological states.

However, the advent of the information technology era may be altering traditional patterns of income–health relationships. On the one hand, information technology may affect health status through direct health promotion pathways that operate relatively independently of individuals’ economic status. Lu and Wang’s [28] research demonstrates that internet use affects self-rated health primarily through information acquisition and social connection mechanisms, rather than through simple economic mediation. On the other hand, information technology may also indirectly affect health through economic pathways—that is, by enhancing income levels and thereby improving health investments and health environments.

More complexly, the relative importance of these two pathways may vary according to urban–rural differences. In rural areas characterized by relatively scarce medical resources, information technology’s direct health effects may be more pronounced, as it directly improves the accessibility of health information and services. Lou et al.’s [53] research documents that social capital’s effects on elderly individuals’ health and well-being vary across regions, suggesting the moderating role of regional conditions in health impact mechanisms. In urban areas with higher degrees of medical service marketization, economic mediation mechanisms may assume greater importance.

Some research has begun to focus on the complexity and dynamism of income–health relationships. Li and Li’s [54] analysis of Chinese elderly health differences from a gender perspective documents that socioeconomic factor effects on health exhibit gender differences, suggesting the diversity of impact mechanisms. Lu et al.’s [55] urban–rural decomposition analysis of health opportunity inequality among middle-aged and elderly individuals further confirms that income’s impact mechanisms on health indeed differ between urban and rural contexts.

International research also provides important insights into this question. Jeong’s [56] comparative research on life satisfaction and depression among urban and rural elderly individuals in Korea documents that the factors affecting urban and rural elderly individuals’ mental health differ, with economic factors exhibiting different levels of importance across environments. These findings underscore the need for more nuanced understanding of income–health relationship patterns across different social contexts.

In the context of rapid information technology development, it becomes necessary to re-examine the patterns of income–health relationships. While the traditional fundamental cause theory assumes that income serves as the fundamental cause of health differences, information technology may provide new pathways for health improvement that operate relatively independently of economic status. At the same time, information technology may also indirectly affect health through economic mechanisms, creating multiple parallel pathways of influence.

Building on recognition of the complexity of income–health relationships in the information technology era, we treat income and health as two parallel domains of information technology impact, rather than assuming simple linear mediating relationships, in order to explore information technology’s differentiated mechanisms of action across different domains. Figure 1 operationalizes this theoretical approach by specifying direct effects of information technology on both economic income (H1) and health status (H2) while incorporating urban–rural status as a systematic moderator (H3) to test whether technology benefits vary by residential context. This analytical framework enables systematic testing of competing theoretical predictions: whereas the traditional digital divide theory anticipates uniform rural disadvantage across domains, our digital dividend differentiation theory predicts domain-specific patterns with modest urban–rural differences in market-standardized economic outcomes but pronounced compensatory effects in geographically dependent health domains.

Figure 1. Theoretical framework of information technology effects on urban–rural inequality.

3. Data and Methods

3.1. Data

This study utilizes data from the 2023 Chinese General Social Survey (CGSS). As China’s longest-running national, comprehensive social survey project, CGSS employs a multi-stage stratified probability sampling methodology, encompassing 125 counties (districts) across 31 provinces, autonomous regions and municipalities. The 2023 survey obtained 11,326 valid responses, providing a nationally representative data foundation for this analysis.

In accordance with the theoretical requirements of our research questions and the technical demands of our analytical approach, we conducted systematic data cleaning procedures. First, we eliminated observations with missing values on key variables to ensure analytical sample integrity; second, we addressed outliers in income variables by treating negative incomes, extreme values exceeding CNY 5 million and obvious coding errors as missing values; finally, we removed cases with anomalous values on basic demographic variables, such as age. Following these data cleaning procedures, our final analytical sample comprises 5332 valid observations, including 2019 rural residents (37.9%) and 3313 urban residents (62.1%).

This sample composition reflects the fundamental patterns of China’s contemporary urbanization process while providing sufficient cases for robust urban–rural comparative analysis. It is worth noting that although urban respondents constitute a numerical majority, the rural sample size remains adequate for supporting reliable statistical inference—a crucial consideration for testing our core hypotheses regarding differential impacts of information technology.

3.2. Measurement

3.2.1. Dependent Variables

This study constructs a comprehensive dependent variable framework encompassing both economic income and health status dimensions, reflecting our theoretical understanding of the multidimensional nature of social stratification.

The economic income dimension employs the natural logarithm of personal annual income. The original income variable derives from the survey item “What was your personal total occupational/labor income last year (2022)?” Given the right-skewed distribution characteristics of income data and the technical requirements of regression analysis, we applied logarithmic transformation to income data—a procedure that not only improves the statistical distribution properties of the variable but also facilitates economic interpretation of coefficients.

The health status dimension encompasses two sub-dimensions: physical health and mental health, reflecting the World Health Organization’s comprehensive definition of health. Physical health is measured through the following item: “In the past four weeks, how frequently have health problems affected your work or other daily activities?” Mental health is assessed through the following item: “In the past four weeks, how much have you felt depressed or discouraged?” Both variables employ 5-point Likert scales. For interpretive clarity, we reverse-coded the original measures, so that higher values indicate greater negative impact of health problems on daily life—that is, poorer health status.

3.2.2. Explanatory Variables

Information technology access serves as our primary explanatory variable, measured through the item “How frequently do you use the internet?” encompassing five levels from “never” to “very frequently”. This measurement strategy captures both the accessibility and intensity of information technology use, providing more precise analytical tools for understanding digital dividend realization mechanisms.

Urban–rural status is operationalized based on residence type, with urban residents coded as 1 and rural residents as 0. While this binary classification necessarily simplifies the complex urban–rural continuum, it captures the fundamental characteristics of China’s social spatial structure and provides a clear analytical framework for examining urban–rural heterogeneity in information technology effects.

3.2.3. Control Variables

Drawing on established research in the social stratification theory and health sociology, we construct a multi-level control variable framework encompassing demographic characteristics, socioeconomic status and political identity. Demographic characteristics include gender (male = 1), age and its squared term (capturing non-linear age effects) and ethnicity (Han = 1); socioeconomic status includes years of education and marital status (using single as the reference category, with dummy variables for married and divorced/widowed); political identity is measured through Communist Party membership (member = 1).

3.3. Analytical Strategy

This study employs a progressively sophisticated analytical strategy, advancing from basic effect estimation to complex moderation and mediation mechanism analysis, a design that reflects both full recognition of social phenomena complexity and systematic analytical requirements.

The initial stage of our analysis focuses on direct effects of information technology on economic income and health status. We employ standard multiple linear regression models:

$$Y_i={\beta }_0+{\beta }_1{\mathrm{Internet}}_i+{\beta }_2{\mathrm{Urban}}_i+X_i^{\prime }\gamma +{\epsilon }_i$$

In this baseline model, ${\mathrm{Y}}_{\mathrm{i}}$ represents individual $\mathrm{i}$’s outcome variables (log income, physical health impact or mental health impact); ${\mathrm{Internet}}_{\mathrm{i}}$ represents information technology usage frequency; ${\mathrm{Urban}}_{\mathrm{i}}$ represents urban–rural status; ${\mathrm{X}}_{\mathrm{i}}^{\prime }$ represents the control variable vector; and ${\mathsf{\epsilon }}_{\mathrm{i}}$ represents the random error term. The core logic underlying this model specification is that it enables us to isolate the net effects of information technology under conditions of controlling for other important factors, thereby establishing a solid foundation for subsequent complex mechanism analysis.

To examine urban–rural moderating effects, we introduce interaction terms between information technology and urban–rural status into our baseline model:

$$Y_i={\beta }_0+{\beta }_1{\mathrm{Internet}}_i+{\beta }_2{\mathrm{Urban}}_i+{\beta }_3\left({\mathrm{Internet}}_i\times {\mathrm{Urban}}_i\right)+X_i^{\prime }\gamma +{\epsilon }_i$$

The sign and significance of the interaction term coefficient ${\mathsf{\beta }}_3$ reveal the differential nature of information technology effects between urban and rural contexts.

We employ the three-step procedure proposed by Baron and Kenny [57] to test income’s mediating role in information technology’s effects on health.

Step 1: Estimate information technology’s total effect on health (c pathway); Step 2: Estimate information technology’s effect on income (a pathway); Step 3: Estimate information technology’s direct effect on health controlling for income (c′ pathway).

The magnitude of the mediation effect is calculated as $\mathrm{a}\times \mathrm{b}$, where $\mathrm{b}$ represents income’s effect on health.

We utilize Blinder–Oaxaca decomposition techniques to quantify information technology factors’ contributions to urban–rural gaps. The Blinder–Oaxaca decomposition, independently developed by Blinder [58] and Oaxaca [59], decomposes mean outcome differences between two groups into endowment effects (characteristic differences) and coefficient effects (return differences). The decomposition formula is

$${\overline{Y}}^U-{\overline{Y}}^R={\left({\overline{X}}^U-{\overline{X}}^R\right)}^{\prime }{\widehat{\beta }}^R+{\overline{X}}^U\left({\widehat{\beta }}^U-{\widehat{\beta }}^R\right)$$

Here, the superscripts U and R denote urban and rural, respectively, with the first term representing endowment effects (characteristic differences) and the second term representing coefficient effects (return differences). All analyses were conducted using Stata 17.0 with the “oaxaca” command [60]. We selected this method over alternatives (e.g., Shapley value decomposition) for three reasons: (1) theoretical appropriateness for binary urban–rural comparisons; (2) direct policy interpretability—endowment effects suggest resource redistribution, while coefficient effects indicate institutional reform needs; and (3) methodological precedent in urban–rural inequality research [61]. While Shapley methods offer advantages in precise variable contribution quantification, they represent promising directions for future research with more granular data.

4. Empirical Results

4.1. Descriptive Analysis: The Contours of Urban–Rural Digital Divide

The descriptive statistics provide essential foundational insights into the contemporary configuration of urban–rural digital divides and inequality patterns in China. As Table 1 demonstrates, substantial disparities exist between urban and rural areas across economic income, health status and information technology access—disparities that constitute the empirical foundation for our subsequent analysis.

Table 1. Descriptive statistics.

Variable	Full Sample (N = 5332)	Rural (N = 2019)	Urban (N = 3313)
Core Variables
Personal Annual Income (CNY)	33,997.31 (71,127.98)	18,810.07 (59,600.60)	44,694.31 (76,453.28)
Log Personal Annual Income	8.121 (3.974)	7.135 (4.051)	8.815 (3.767)
Health Impact on Work	2.281 (1.267)	2.608 (1.364)	2.050 (1.139)
Depression Level	2.174 (1.129)	2.411 (1.223)	2.007 (1.026)
Internet Usage Frequency	3.539 (1.545)	3.206 (1.606)	3.773 (1.455)
Demographic Variables
Male	0.408 (0.491)	0.460 (0.498)	0.371 (0.483)
Age	54.889 (16.078)	56.444 (15.830)	53.794 (16.168)
Han Ethnicity	0.895 (0.306)	0.814 (0.389)	0.952 (0.214)
Years of Education	8.739 (4.565)	7.073 (4.150)	9.913 (4.547)
Marital Status
Single	0.116 (0.321)	0.100 (0.300)	0.128 (0.334)
Married	0.735 (0.441)	0.771 (0.420)	0.710 (0.454)
Divorced/Widowed	0.149 (0.356)	0.129 (0.335)	0.163 (0.369)
Other Variables
CCP Member	0.112 (0.316)	0.088 (0.283)	0.129 (0.335)
Urban Residence	0.587 (0.492)	—	—

Notes: (1) Standard deviations are reported in parentheses; (2) Health Impact on Work and Depression Level are measured on a 5-point scale (1 = lowest, 5 = highest); (3) Internet Usage Frequency is measured on a 5-point scale (1 = never, 5 = very frequently); (4) Dummy variables: Male (1 = male, 0 = female), Han Ethnicity (1 = Han, 0 = other), Single/Married/Divorced–Widowed (1 = yes, 0 = no), CCP Member (1 = member, 0 = non-member), Urban Residence (1 = urban, 0 = rural).

In economic terms, urban residents’ average annual income of CNY 44,694 substantially exceeds rural residents’ CNY 18,811, yielding an urban–rural income ratio of 2.38:1, which speaks to both the depth and breadth of urban–rural economic inequality. What is particularly striking is that this income gap manifests not merely in average levels but also in the degree of internal differentiation (standard deviations: 76,453 versus 59,601), further revealing the more complex stratification structures within urban society. This finding provides crucial contextual information for understanding the potentially differentiated impacts of information technology across varying social contexts.

The urban–rural divide in health dimensions proves equally revealing. Rural residents report significantly higher levels of physical health problems affecting daily work activities (2.608) compared to urban residents (2.050), with the urban–rural gap in mental health being similarly pronounced (2.411 versus 2.007). This phenomenon—what we might characterize as a “dual health disadvantage”—reflects profound structural shortcomings in rural areas across multiple dimensions, including medical resource allocation, environmental quality and social support networks. At the same time, it provides important empirical grounds for expectations regarding information technology’s potential role as a health equalizer.

Regarding information technology access, urban–rural differences appear relatively modest (3.773 versus 3.206); yet, we must recognize that this surface-level “convergence” may mask deeper dimensions of digital divide, including crucial aspects, such as usage quality, functional development depth and ultimate benefit conversion efficiency. When we juxtapose this relatively narrow technology access gap with the substantial income and health disparities, a fundamental theoretical question emerges: does information technology function as an “amplifier” of existing inequality structures, or does it serve as a “social equalizer” with genuine equalizing potential?

The distribution patterns of control variables further confirm the deep-seated characteristics of urban–rural dual structure. Urban residents’ average educational attainment (9.913 years) significantly exceeds that of rural residents (7.073 years), while Party membership distribution also exhibits clear urban bias (12.9% versus 8.8%). These differences constitute not merely important components of urban–rural inequality patterns but may also influence individuals’ information technology usage capabilities and benefit realization capacities, thereby moderating the ultimate realization of information technology effects.

4.2. Basic Effects of Information Technology: Validating the “Digital Dividend” Hypothesis

The regression results presented in Table 2 provide compelling empirical evidence for information technology’s digital dividend effects. Controlling for other variables, information technology usage frequency exerts a significant positive effect on personal income (β = 0.286, p < 0.001), indicating that each unit increase in information technology usage frequency corresponds to approximately 28.6% income growth. This finding provides strong support for Hypothesis 1—namely that information technology generates substantial economic dividends.

Table 2. Effects of internet usage on income and health.

Variable	Model 1	Model 2	Model 3
	Log Personal Annual Income	Health Impact on Work	Depression Level
Internet Usage Frequency	0.286 *** (0.045)	−0.103 *** (0.016)	−0.068 *** (0.015)
Male	1.064 *** (0.111)	−0.116 *** (0.037)	−0.134 *** (0.035)
Age	0.277 *** (0.026)	0.020 *** (0.007)	0.016 ** (0.007)
Age Squared	−0.002 *** (0.000)	−0.000 (0.000)	−0.000 ** (0.000)
Han Ethnicity	0.479 ** (0.212)	0.044 (0.066)	0.008 (0.066)
Years of Education	0.156 *** (0.016)	−0.026 *** (0.005)	−0.027 *** (0.005)
Married	0.129 (0.203)	−0.089 (0.064)	−0.057 (0.061)
Divorced/Widowed	0.470 * (0.245)	0.014 (0.079)	0.091 (0.077)
CCP Member	0.692 *** (0.153)	−0.148 *** (0.057)	−0.127 ** (0.054)
Urban Residence	1.218 *** (0.128)	−0.397 *** (0.041)	−0.298 *** (0.039)
Constant	−3.446 *** (0.724)	2.192 *** (0.179)	2.563 *** (0.185)
Observations	5332	5332	5332
R-Squared	0.146	0.179	0.071

Notes: (1) Robust standard errors are reported in parentheses; (2) *** p < 0.01, ** p < 0.05, * p < 0.10; (3) Reference categories: Female (for Male), Non-Han Ethnicity (for Han Ethnicity), Single (for Married and Divorced/Widowed), Non-CCP Member (for CCP Member), Rural Residence (for Urban Residence).

From a mechanistic perspective, information technology’s income-enhancing effects likely operate through multiple channels. First, information technology enhances labor productivity, enabling workers who master relevant skills to command additional skill premiums in employment markets; second, information technology expands market boundaries, providing individuals with enhanced economic opportunities—particularly in rural areas, where e-commerce platforms, online marketing and other emerging business models create new channels for value realization in traditional agriculture and handicrafts; third, information technology reduces information search costs, improves market transaction efficiency and thereby indirectly promotes income growth.

Regarding health impacts, information technology similarly demonstrates significant beneficial effects. For physical health, increased information technology usage frequency significantly reduces the negative impact of health problems on work activities (β = −0.103, p < 0.001); for mental health, information technology use similarly reduces the occurrence of depressive symptoms (β = −0.068, p < 0.001). These findings validate Hypothesis 2—namely that information technology exerts significant beneficial effects on health status.

Information technology’s health-promoting effects can be understood across three dimensions. In terms of information access, the internet provides individuals with rich health knowledge and medical information, thereby improving health literacy and self-care capabilities; regarding social connections, online social interactions alleviate loneliness and strengthen social support, positively affecting mental health; in the domain of health management, digital health tools enable individuals to better monitor and manage their health status.

The impact patterns of control variables merit attention as well. The directional effects and significance levels of variables such as gender, age, education and Party membership generally conform to theoretical expectations, validating both the reasonableness of our model specifications and providing empirical evidence for understanding the multidimensional characteristics of social stratification. Particularly noteworthy is the fact that the urban–rural variable exhibits strong significance across all models, underscoring the structural and persistent nature of urban–rural differences.

4.3. Urban–Rural Moderating Effects: Heterogeneous Mechanisms of Information Technology Impact

The interaction effect analysis results presented in Table 3 reveal important heterogeneous characteristics of information technology impacts, providing crucial insights into the realization mechanisms of digital dividends across different social contexts.

Table 3. Interaction effects models.

Variable	Model 1	Model 2	Model 3
	Log Personal Annual Income	Health Impact on Work	Depression Level
Internet Usage Frequency	0.346 *** (0.063)	−0.144 *** (0.022)	−0.096 *** (0.021)
Urban Residence	1.613 *** (0.295)	−0.670 *** (0.101)	−0.486 *** (0.094)
Internet Usage Frequency × Urban Residence	−0.114 (0.077)	0.079 *** (0.025)	0.054 ** (0.024)
Male	1.064 *** (0.111)	−0.116 *** (0.037)	−0.134 *** (0.035)
Age	0.277 *** (0.026)	0.019 *** (0.007)	0.016 ** (0.007)
Age Squared	−0.002 *** (0.000)	−0.000 (0.000)	−0.000 ** (0.000)
Han Ethnicity	0.481 ** (0.212)	0.043 (0.066)	0.007 (0.066)
Years of Education	0.157 *** (0.016)	−0.026 *** (0.005)	−0.027 *** (0.005)
Married	0.125 (0.203)	−0.086 (0.064)	−0.055 (0.061)
Divorced/Widowed	0.466 * (0.245)	0.016 (0.079)	0.093 (0.077)
CCP Member	0.691 *** (0.153)	−0.147 *** (0.057)	−0.127 ** (0.054)
Constant	−3.647 *** (0.734)	2.331 *** (0.188)	2.658 *** (0.193)
Observations	5332	5332	5332
R-Squared	0.146	0.181	0.073

Notes: (1) Robust standard errors are reported in parentheses; (2) *** p < 0.01, ** p < 0.05, * p < 0.10; (3) The interaction term Internet Usage × Urban Residence captures the differential effect of internet usage between urban and rural residents; (4) Reference categories: Female (for Male), Non-Han Ethnicity (for Han Ethnicity), Single (for Married and Divorced/Widowed), Non-CCP Member (for CCP Member), Rural Residence (for Urban Residence).

Regarding income effects, while the interaction term coefficient lacks statistical significance (β = −0.114, p > 0.10), its negative sign suggests that rural areas may derive greater income dividends from information technology. The theoretical significance of this finding lies in its support for characterizing information technology as an “equalizer” rather than an “amplifier”, meaning that information technology diffusion helps narrow rather than widen urban–rural income gaps.

In terms of health effects, however, we observe more definitive patterns of urban–rural heterogeneity. Interaction term coefficients prove significantly positive in both physical health (β = 0.079, p < 0.001) and mental health (β = 0.054, p < 0.05) models, indicating that information technology’s health-promoting effects for rural residents significantly exceed those for urban residents. This finding strongly supports the health dimension of Hypothesis 3—namely that information technology’s health-promoting effects are greater for rural residents.

The underlying mechanism generating this urban–rural heterogeneity can be understood through the lens of diminishing marginal utility. Urban residents typically already enjoy relatively adequate medical resources, health information and social support; while information technology introduction proves beneficial, its marginal improvement effects remain limited. Rural residents, by contrast, face more severe health resource constraints, and information technology introduction can substantially improve their capacity to access health information, receive medical services and obtain social support, thereby generating greater marginal health utility.

4.4. Mediation Mechanism Analysis: The Income–Health Nexus

The stepwise regression results in Table 4 and mediation effect decomposition in Table 5 provide crucial evidence for understanding the internal mechanisms through which information technology affects health. The three-step regression results clearly demonstrate income’s mediating role in information technology’s health effects.

Table 4. Stepwise regression results for mediation analysis.

Variable	Step 1: X→Y	Step 2: X→M	Step 3: X + M→Y
Panel A: Health Impact on Work as Dependent Variable
Internet Usage Frequency	−0.103 *** (0.016)	0.286 *** (0.045)	−0.093 *** (0.016)
Log Personal Annual Income	—	—	−0.032 *** (0.005)
Observations	5332	5332	5332
R-Squared	0.179	0.146	0.188
Panel B: Depression Level as Dependent Variable
Internet Usage Frequency	−0.068 *** (0.015)	0.286 *** (0.045)	−0.059 *** (0.015)
Log Personal Annual Income	—	—	−0.029 *** (0.005)
Observations	5332	5332	5332
R-Squared	0.071	0.146	0.081

Notes: (1) Robust standard errors are reported in parentheses; (2) *** p < 0.01; (3) All models include control variables: Male, Age, Age Squared, Han Ethnicity, Years of Education, Married, Divorced/Widowed, CCP Member, Urban Residence; (4) Step 1 tests the total effect of internet usage on health outcomes (c pathway); (5) Step 2 tests the effect of internet usage on income (a pathway); (6) Step 3 tests the direct effect of internet usage on health outcomes controlling for income (c′ pathway) and the effect of income on health outcomes (b pathway).

Table 5. Decomposition of income mediation effects.

Effect Type	Health Impact on Work			Depression Level
	Coefficient	SE	p-Value	Coefficient	SE	p-Value
Total Effect (c)	−0.103	0.013	0.000	−0.068	0.012	0.000
Direct Effect (c′)	−0.093	0.013	0.000	−0.059	0.012	0.000
Indirect Effect (a × b)	−0.009	0.002	0.000	−0.008	0.002	0.000
Mediation Proportion	8.47%	—	—	12.44%	—	—

Notes: (1) SE = Standard Error; (2) Mediation Proportion = |Indirect Effect/Total Effect| × 100%.

For physical health, information technology’s total effect equals −0.103 (Step 1); after incorporating income variables, the direct effect decreases to −0.093 (Step 3), with an indirect effect of −0.009 and a mediation effect proportion of 8.47%. While this proportion may appear modest, it reveals an important theoretical mechanism: information technology affects health not merely through direct channels of information access and social connections but also indirectly promotes health improvement through economic income enhancement.

For mental health, the mediation effect proportion reaches 12.44%, indicating that economic status improvements exert more substantial effects on mental health. This finding aligns closely with theoretical expectations in health economics: economic pressure alleviation can significantly improve individuals’ psychological states, with information technology—as an economic empowerment tool—partially realizing its mental health effects through economic mechanisms.

The discovery of this mediation mechanism validates Hypothesis 4—namely that information technology partially improves health status through economic income enhancement. From a theoretical standpoint, this finding supports the fundamental cause theory in health inequality research, which considers socioeconomic status as the fundamental determinant of health differences. As an emerging socioeconomic resource, information technology’s health effects necessarily relate closely to economic mechanisms.

While income’s mediating role proves significant, its proportion of total effects remains relatively limited, indicating that information technology’s health impacts operate primarily through non-economic mechanisms. This finding reminds us that in promoting digital health development, we cannot focus exclusively on information technology’s economic functions but must also emphasize its direct roles in health information dissemination, social support network construction and health behavior guidance.

4.5. Urban–Rural Gap Decomposition: Information Technology’s Contribution

The Blinder–Oaxaca decomposition results in Table 6 provide precise econometric evidence for quantifying information technology’s role in urban–rural disparities. The decomposition results reveal that the total urban–rural income gap equals −1.802 (logarithmic value), with endowment effects of −0.708, coefficient effects of −1.539 and interaction effects of 0.445.

Table 6. Comprehensive Blinder–Oaxaca decomposition results.

Component	Income Gap (Log)	Health Impact on Work Gap	Depression Level Gap
Overall Decomposition
Rural (Group 1)	7.169 *** (0.090)	2.602 *** (0.030)	2.397 *** (0.027)
Urban (Group 2)	8.971 *** (0.065)	2.029 *** (0.020)	1.988 *** (0.018)
Difference (Urban–Rural)	−1.802 *** (0.111)	0.573 *** (0.036)	0.409 *** (0.033)
Endowment Effects	−0.708 *** (0.070)	0.158 *** (0.021)	0.096 *** (0.018)
Coefficient Effects	−1.539 *** (0.135)	0.389 *** (0.044)	0.278 *** (0.041)
Interaction Effects	0.445 *** (0.106)	0.026 (0.034)	0.036 (0.032)
Endowment Decomposition
Internet Usage Frequency	−0.195 *** (0.037)	0.043 *** (0.011)	0.040 *** (0.010)
Coefficient Decomposition
Internet Usage Frequency	−0.094 (0.320)	−0.302 *** (0.103)	−0.102 (0.096)
Constant	−0.953 (1.229)	−0.929 ** (0.396)	−0.803 ** (0.372)
Interaction Decomposition
Internet Usage Frequency	0.016 (0.056)	0.053 *** (0.018)	0.018 (0.017)
Observations	5332	5332	5332

Notes: (1) Robust standard errors are reported in parentheses; (2) *** p < 0.01, ** p < 0.05; (3) All models include control variables: Male, Age, Age Squared, Han Ethnicity, Years of Education, Married, Divorced/Widowed, CCP Member; (4) Group 1 = Rural residents; Group 2 = Urban residents; (5) The endowment effect captures differences due to different characteristics between groups; the coefficient effect captures differences due to different returns to characteristics between groups; the interaction effect captures the interaction between endowment and coefficient differences.

Within endowment effects, information technology factors contribute −0.195, accounting for 27.5% of total endowment effects and 10.8% of overall gaps. This finding indicates that urban–rural differences in information technology access levels constitute an important factor in income disparities, although their impact magnitude remains relatively circumscribed. Greater impacts derive from coefficient effects, suggesting systematic differences in returns to information technology and other resources between urban and rural areas.

In health gap decomposition, information technology factors assume more prominent roles. Within physical health gap endowment effects, information technology contributes 0.043, representing 27.2% of endowment effects; for mental health gaps, information technology contributes 0.040, representing 41.7% of endowment effects. These findings indicate that information technology plays crucial roles in urban–rural health gap formation, with digital divide bridging holding important implications for improving urban–rural health equity.

The policy implications of these decomposition results are clear: narrowing the urban–rural gaps requires dual-track strategies. On the one hand, continued promotion of information infrastructure development in rural areas and enhancement of rural residents’ information technology access levels remain essential; on the other hand, improving information technology utilization efficiency in rural areas and ensuring that rural residents derive benefits from information technology comparable to urban residents assume even greater importance.

The theoretical value of this decomposition analysis lies in its transformation of the abstract digital divide concept into quantifiable empirical phenomena, providing important methodological demonstration for digital inequality research. Simultaneously, it reminds us that while information technology constitutes an important factor affecting urban–rural gaps, it is not the sole determinant; narrowing the urban–rural disparities continues to require comprehensive reforms across multiple domains, including education, healthcare and infrastructure.

5. Conclusions and Discussion

As we reflect on the core findings of this study, what emerges as most striking is an unexpected pattern of differentiation: information technology’s impact on urban–rural income gaps exhibits relative homogeneity, while its impact on urban–rural health gaps demonstrates marked heterogeneous characteristics. This finding of “one technology, two faces” provides crucial insights for understanding the complexity of social stratification in the digital age. The principal empirical results can be summarized as follows:

• Information technology access demonstrates substantial positive effects on both economic outcomes (β = 0.286, p < 0.001) and health indicators, with each unit increase in usage frequency associated with approximately 28.6% income enhancement and significant reductions in health-related work impairment.
• While urban–rural differentials in technology’s income effects prove statistically insignificant (β = −0.114, p > 0.10), health effects exhibit pronounced heterogeneity, with rural residents experiencing markedly greater benefits (physical health interaction: β = 0.079, p < 0.001; mental health interaction: β = 0.054, p < 0.05).
• Economic income serves as a partial mediating mechanism, accounting for 8.47% of technology’s physical health effects and 12.44% of mental health effects, indicating the operation of both economic and direct health pathways.
• Decomposition analysis reveals that information technology factors contribute 10.8% to urban–rural income disparities while assuming more substantial roles in health gaps (27.2% of physical health differences, 41.7% of mental health differences).
• Technology effects conform to domain-specific patterns: relative convergence in market-standardized economic domains versus compensatory mechanisms in geographically dependent health domains.

This challenges a long-standing yet rarely explicitly articulated assumption in digital stratification research—the assumption of technological impact uniformity. Our research demonstrates that different life domains operate according to distinct institutional logics, resource allocation mechanisms and degrees of marketization; consequently, the embedding processes and action mechanisms of technology within these domains necessarily differ. In relatively standardized economic domains, market returns to technical skills tend toward homogenization; in health domains that depend more heavily on geographic proximity, technology’s compensatory functions prove more pronounced in resource-scarce areas [20,62].

This study’s findings carry paradigm-shifting implications for the digital inequality research field. Traditional research has primarily centered on “access divides” and “usage divides”, while we illuminate the significance of “benefit divides”: even under similar levels of technological access, different groups may experience substantial differences in actual benefits derived from information technology. This finding opens new analytical dimensions for digital divide research, suggesting that future inquiry must shift from simple “have versus have-not” binary oppositions toward in-depth analysis of technology benefit realization mechanisms. Our urban–rural decomposition analysis indicates that bridging digital divides requires not merely improving technological access levels in rural areas, but more crucially, enhancing the efficiency of technology utilization and benefit conversion mechanisms. This theoretical perspective shift also provides fresh insights for reconceptualizing digital inclusion: genuine digital inclusion may depend more heavily on differences in various groups’ capacities to utilize technology for value creation [63,64]. At the policy level, our findings provide theoretical foundations for developing differentiated digital policies. Policymakers must abandon “one-size-fits-all” approaches and craft targeted intervention strategies based on domain-specific characteristics. In the health domain, emphasis should be placed on digital health service provision in rural areas, fully leveraging information technology’s compensatory functions; in the economic domain, greater attention should be directed toward skills training and platform development.

Beyond theoretical contributions, this study also makes significant academic and practical contributions to the field of sustainable development. Academically, it expands digital inequality research by linking information technology’s differentiated effects to sustainable urban–rural development, thereby providing a novel framework for understanding how digital tools can mitigate structural disparities. Practically, our findings demonstrate that improving rural residents’ digital access and utilization efficiency can not only reduce income and health inequalities but also contribute directly to the realization of sustainable development goals (SDGs), particularly in health equity, poverty reduction and inclusive digital transformation. Therefore, the study underscores the necessity of integrating digital strategies into broader sustainability policies and highlights the importance of domain-specific interventions to achieve long-term social and health equity.

Nevertheless, we must acknowledge our research limitations. The relationship between economic resources and technology adoption exhibits the endogeneity common to many economic phenomena: income facilitates technology access, while technology use enhances earning capacity, creating feedback effects that generate cumulative advantages over time. Our focus on technology’s distributional consequences, while necessarily abstracting from these simultaneous equations, reflects both the policy urgency of understanding digital inequality and the analytical constraints inherent in cross-sectional identification strategies. Cross-sectional data constrain causal inference capabilities; measurements of technology usage depth and quality remain relatively simplified; domain-specific theoretical explanations retain largely speculative characteristics. These limitations simultaneously point toward important directions for future research. Future work should prioritize panel data analysis of technology adoption dynamics, quasi-experimental evaluation utilizing exogenous policy variation and deeper investigation of the market and institutional mechanisms that produce domain-specific technology effects.