Explaining Older Adults’ Continuance Intention Toward Smart Homes: Integrating the Expectation–Confirmation Model of Information Systems and the Technology Acceptance Model

Abstract

China is experiencing rapid population aging and is actively promoting smart home–based eldercare. Smart homes offer a promising means of supporting older adults in aging in place. However, low adoption and limited sustained use constrain their potential benefits, thereby exacerbating social, economic, and healthcare burdens. This study examined factors influencing older adults’ continuance intention to use smart homes in Shandong Province, China, by integrating the Expectation–Confirmation Model of Information Systems and the Technology Acceptance Model and incorporating China-specific contextual antecedents, including government policy, intergenerational technical support, compatibility, and cost. Data were collected using an online questionnaire survey of older adults aged 60 years and older with prior smart home experience (n = 421) and analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). Results showed that perceived usefulness, perceived ease of use, satisfaction, and cost directly affected continuance intention, whereas government policy, compatibility, and intergenerational technical support influenced continuance intention through perceived usefulness, perceived ease of use, and confirmation. Based on these results, this study proposes a conceptual framework for understanding older adults’ continuance intention toward smart homes. The findings provide implications for inclusive policy, user-centered design, and family-supported digital aging in rapidly aging societies.

1. Introduction

Population aging is accelerating worldwide due to sustained declines in fertility and mortality. In 2022, approximately 771 million people aged 65 years and older were estimated, representing about 10% of the global population. This share is projected to reach 16% by 2050 and 24% by 2100 [1]. China, which has the world’s largest aging population, had 264 million people aged 60 years and older in 2021 (18.7%), a figure that is expected to reach 380 million (approximately 28%) by 2050 [2]. As a result, China faces mounting social, economic, and healthcare pressures, alongside heightened physiological, social, and psychological vulnerabilities among older adults [3,4,5]. Under China’s “9073” elderly care model, roughly 90% of older adults are expected to age in place, 7% to receive community-based care, and only 3% to live in institutions. Consequently, as functional decline increases the need for assistance with activities of daily living, demand for sustainable, home-based care solutions continues to intensify [5].

Smart homes have been widely promoted as a core technological response to demographic and care pressures associated with population aging [6,7]. A smart home is a residence equipped with networked sensors, devices, and appliances that enable remote monitoring, control, and personalized services [8]. For older adults, smart home services can be organized into five dimensions: health monitoring, environment monitoring, companionship, social communication, and recreation and entertainment, as shown in Figure 1 [9]. Smart homes can learn users’ habits [10], support multimodal interaction through sound, light, and spatial cues [11], and connect to external platforms to deliver these services [9]. Evidence indicates that smart homes enhance independence, well-being, and quality of life among older adults, thereby facilitating aging in place [12,13,14].

In China, smart elderly care has been positioned as a strategic national priority, supported by subsidies, standardization, and promotional initiatives [7,15,16,17]. Older adults also show a strong interest in smart homes [18], and adoption rates among older adults are increasing [19]. As shown in Figure 2, China’s smart home market for older adults reached approximately CNY 156 billion in 2024, with a compound annual growth rate (CAGR) of 21.2% from 2020 to 2024 [20]. However, Figure 3 shows that, in 2023, the continued use rate of smart homes among older adults in China was only 30%, while the discontinued use rate reached 57%. The intermittent use rate was 13% [21]. This indicates that more than half of older adults discontinued smart home products after initial use and did not resume use. Moreover, 83% of smart home products were discontinued within three months of initial use by older adults. Devices such as smart wristbands, health monitors, and companion robots often became underused or idle after a short period of use, and in some cases were described as “electronic waste” [21].

Therefore, the central problem addressed in this study is not merely initial adoption, but rather why older adults discontinue using smart homes after adoption and what drives their continued use in the Chinese context. From a practical perspective, as older adults’ educational attainment continues to improve, they are increasingly able to adopt and learn to use smart home products, and an increasing number of adult children select smart home products that better fit older adults’ needs [19,20,21]. However, the continued use rate of smart homes among older adults is only about 30%, which constrains the extent to which smart homes can support aging in place [19]. Thus, maintaining users’ continuance intention is critical [22]. Continuance intention is a key predictor of sustained use of smart homes [21,22]. In older adults’ post-adoption use, factors such as perceived ease of use, perceived usefulness, compatibility, and cost are key concerns [23,24,25,26,27,28].

From a theoretical perspective, perceived ease of use and perceived usefulness, core constructs of the Technology Acceptance Model (TAM), are consistently identified as key determinants of older adults’ behavioral intention toward smart homes in the pre-adoption stage [23,24,25,26]. However, it is unclear whether these determinants continue to affect older adults’ continuance intention after they have begun using smart homes. Existing studies on older adults’ acceptance of smart homes have also relied heavily on pre-adoption models such as the TAM [23,24,29,30]. By contrast, the Expectation-Confirmation Model of Information Systems (ECM-IS), which has been widely used to explain continuance intention in information systems research through constructs such as confirmation, satisfaction, and perceived usefulness, has not been systematically applied to older adults’ continuance intention toward smart homes in the post-adoption stage. As a result, existing adoption-oriented studies provide limited explanatory power for the post-adoption use patterns observed in China, where rapid expansion does not necessarily translate into sustained use.

In addition, several contextual antecedents that are particularly salient in China’s smart eldercare environment remain underexplored in relation to continuance intention. Government policy plays a decisive role in shaping the smart home market for older adults in China [31]. However, its impact on older adults’ continuance intention remains unclear. Compatibility across devices and platforms is frequently cited as a determinant of older adults’ acceptance of smart homes [27,28]. However, its influence on continuance intention remains underexplored. Previous studies have consistently identified cost as a significant determinant of behavioral intention in the pre-adoption stage [9,32,33,34], yet whether cost affects older adults’ continuance intention in the post-adoption stage remains largely unknown.

Furthermore, 64% of older adults in China were first introduced to smart homes by younger family members, and 80% reported that they prioritized seeking help from their children when they encountered operational difficulties [35]. Intergenerational technical support from adult children has been shown to influence older adults’ behavioral intention toward smart homes during the pre-adoption stage [24,29]. However, its contribution to continuance intention after adoption has received limited empirical attention. Collectively, these factors reflect the socio-technical and policy context within which smart homes are promoted in China, but they have rarely been examined in an integrated post-adoption framework.

To address these gaps, this study integrates the Expectation-Confirmation Model of Information Systems and the Technology Acceptance Model with contextual antecedents, including government policy, intergenerational technical support, compatibility, and cost. This study empirically focuses on older adults in Shandong Province, China, a rapidly aging and policy-active region, and pursues three objectives:

• To identify the key factors influencing older adults’ continuance intention to use smart homes in Shandong Province, China.
• To examine the relationships between the key factors and older adults’ continuance intention to use smart homes in Shandong Province, China.
• To develop a conceptual framework for older adults’ continuance intention to use smart homes in Shandong Province, China.

The remainder of this paper is organized as follows. Section 2 reviews the relevant literature and proposes the conceptual framework. Section 3 describes the research methodology. Section 4 presents the data analysis and results, including demographic information and PLS-SEM findings. Section 5 discusses the key results, while Section 6 concludes by outlining the theoretical contributions, practical implications, limitations, and recommendations for future research.

2. Literature Review

2.1. Older Adults’ Acceptance of Smart Homes

Empirical research on older adults’ acceptance of smart homes has focused primarily on the pre-adoption stage. Most studies have examined intention-related dependent variables, including behavioral intention, intention to use, willingness to adopt, and intention to accept [23,24,25,26,30,36,37,38,39,40]. For example, Maswadi, Ghani and Hamid [39] identified factors influencing older adults’ behavioral intention toward smart home technologies in Saudi Arabia. Zhou, Qian and Kaner [23] examined older adults’ intention to use smart homes. However, research on older adults’ continuance intention toward smart homes remains scarce [22]. Among the independent variables, perceived ease of use and perceived usefulness are the most consistently examined variables, underscoring the enduring influence of the Technology Acceptance Model (TAM) in studies of older adults’ acceptance of smart homes [23,24,25,26].

From a theoretical perspective, prior work has predominantly used the Technology Acceptance Model (TAM) and the Unified Theory of Acceptance and Use of Technology (UTAUT) to explain older adults’ intention to use smart homes in the pre-adoption stage [23,24,25,26,30,38,39,41]. TAM explains technology acceptance primarily through perceived usefulness and perceived ease of use, whereas UTAUT extends this perspective by incorporating performance expectancy, effort expectancy, social influence, and facilitating conditions [42,43]. Although these models have strong explanatory power for pre-adoption acceptance and intention to use, they are less suitable as the primary framework for the present study, which focuses on post-adoption continuance intention among older adults, as they do not explicitly model expectation confirmation and post-use satisfaction. By contrast, the Expectation-Confirmation Model of Information Systems (ECM-IS), which was developed to explain continuance intention in the post-adoption stage, is more theoretically aligned with the present study. However, it has rarely been applied to older adults’ intention to continue using smart homes [22]. UTAUT was excluded from the proposed model to maintain theoretical parsimony and because prior empirical evidence in this context has consistently highlighted the salience of perceived usefulness and perceived ease of use relative to the broader set of UTAUT constructs.

Although ECM-IS is well-suited to explaining post-adoption continuance, its core mechanism primarily emphasizes confirmation, perceived usefulness, and satisfaction. It provides a limited explicit explanation of perceived ease of use as a post-adoption cognitive mechanism. This limitation is particularly important in the smart home context, where ease of use remains a key concern for older adults [19,44]. TAM, by contrast, offers a stronger account of how perceived ease of use shapes users’ evaluations of technology use [45]. Therefore, this study adopts ECM-IS as the core post-adoption framework and selectively incorporates perceived ease of use from TAM. In this integrated framework, ECM-IS explains the continuance mechanism, while perceived ease of use extends it by capturing usability-related cognition.

In 2024, the satisfaction rate among Chinese older adults with prior smart home experience was below 40%, indicating a substantial gap between the actual user experience of smart home products and older adults’ prior expectations [46]. This finding suggests that, in the post-adoption stage, satisfaction and expectation confirmation are important factors influencing older adults’ continuance intention. Compared with factors such as technology anxiety, privacy, and security, which are more commonly discussed in the pre-adoption stage [9,45], older adults in the continuance stage are more concerned with perceived ease of use, perceived usefulness, compatibility, and cost, with perceived ease of use and perceived usefulness being particularly salient [44,46].

Meanwhile, intergenerational technical support plays a prominent role in China. Survey data show that 64% of Chinese older adults were introduced to smart homes through younger family members, and 80% of older adults prioritize seeking help from their children when they encounter operational difficulties [35], indicating that family-based intergenerational support remains the primary source of support for older adults’ smart device use [47]. In addition, government policy plays an important role in promoting smart homes for older adults, regulating markets, and supporting continued use [18]. Overall, perceived ease of use, perceived usefulness, compatibility, cost, government policy, and intergenerational technical support are all important factors influencing older adults’ continuance intention toward smart homes. However, existing empirical research on older adults’ continuance intention toward smart homes has provided only a limited systematic examination of these factors. Although aging-specific constructs such as technology anxiety, perceived risk, and age-related cognitive and physical decline are also relevant to older adults’ acceptance of smart homes, the present study focuses on post-adoption continuance. It therefore prioritizes the most salient post-adoption cognitive factors and China-specific contextual antecedents.

Several gaps remain. First, most empirical research has focused on the pre-adoption stage, with limited attention given to continuance intention in the post-adoption stage [48]. Second, it remains unclear whether perceived ease of use and perceived usefulness still continue to influence continuance intention during the post-adoption stage [23,24,25,26]. Third, ECM-IS has been underutilized in this context, despite its relevance to the post-adoption stage [22]. Finally, the contextual antecedents, including government policy, compatibility, cost, and intergenerational technical support, remain insufficiently examined [24,27,29,31,33]. In response, this study develops a conceptual framework that integrates ECM-IS with the TAM-derived construct of perceived ease of use and incorporates four China-specific contextual antecedents to explain older adults’ continuance intention toward smart homes in the post-adoption stage.

2.2. Conceptual Framework and Hypothesis Development

2.2.1. Expectation-Confirmation Model of Information Systems

Expectation-Confirmation Theory (ECT) explains consumers’ post-adoption behavior by positing that continuance intention is primarily determined by satisfaction with prior use [49]. ECT has been widely applied to customer satisfaction, post-purchase behavior, and service marketing [50]. The theory comprises five constructs: expectation, perceived performance, confirmation, satisfaction, and repurchase intention. Users first form expectations before adoption, then evaluate the perceived performance after use; comparing these expectations and performance yields confirmation, which affects satisfaction and, in turn, continuance intention [51].

To address the limitations of ECT in explaining technology discontinuance and capturing post-adoption psychological processes in information systems, Bhattacherjee [49] proposed the Expectation-Confirmation Model of Information Systems (ECM-IS) (Figure 4). ECM-IS shifts the focus from general consumer satisfaction to post-use beliefs in the context of information systems. In ECM-IS, perceived usefulness and satisfaction are modeled as the direct antecedents of continuance intention. Moreover, confirmation influences both perceived usefulness and satisfaction, which, in turn, together affect continuance intention [49].

ECM/ECM-IS has been widely applied across business, information science, computer science, and management to explain continuance intention [52,53,54,55,56]. However, its application to older adults’ intention to continue using smart homes remains limited [57,58]. This suggests that the effects of confirmation, perceived usefulness, and satisfaction on older adults’ continuance intention toward smart homes have not yet been systematically examined.

2.2.2. Technology Acceptance Model

The Technology Acceptance Model (TAM), derived from the Theory of Reasoned Action (TRA), explains users’ behavioral intention to use information technology systems voluntarily [28,59,60]. In the Technology Acceptance Model (TAM), perceived ease of use and perceived usefulness are core determinants of behavioral intention [61]. Extensive empirical work supports the model’s explanatory power and robustness (Figure 5) [62].

In early TAM formulations, attitude mediated the effects of perceived ease of use and perceived usefulness on behavioral intention. Subsequent research, however, has shown that attitude often adds little incremental predictive value and, accordingly, has frequently been excluded from later TAM applications [34,45]. Today, TAM is among the most widely used and influential models for studying user acceptance [23,62].

Perceived usefulness is a shared construct in both TAM and ECM-IS, whereas satisfaction is a core construct in ECM-IS but not in the original TAM. In TAM, PU primarily captures pre-adoption beliefs about the expected utility of a technology. In ECM-IS, by contrast, perceived usefulness reflects a post-adoption re-evaluation of utility shaped by actual use experience and expectation confirmation. Moreover, unlike TAM, ECM-IS places satisfaction at the center of the post-adoption process, conceptualizing it as an affective evaluation formed after expectation confirmation and actual use, which in turn directly drives continuance intention [49,60].

TAM has been applied across multiple domains, including education [63,64], business [65,66,67], computer science and information systems [68,69], and green and sustainable technologies [42,70]. In the context of smart homes for older adults, TAM has been used to examine the adoption of smart home services, IoT health technologies, and smart home healthcare systems [25,59]. These studies consistently show that perceived ease of use and perceived usefulness predict behavioral intention at the pre-adoption stage [24,25].

However, because the present study focuses on continuance intention rather than initial adoption, ECM-IS provides the primary theoretical foundation. Accordingly, perceived usefulness and satisfaction are modeled following the ECM-IS post-adoption logic, while perceived ease of use is selectively incorporated from TAM into the model as an additional usability-related cognitive antecedent. This extension is theoretically important because ease of use remains a salient concern for older adults as they continue using smart homes in their daily lives [19,44]. Therefore, this study extends the ECM-IS framework by incorporating perceived ease of use to examine whether usability perceptions continue to shape older adults’ continuance intention to use smart homes.

2.2.3. China-Specific Contextual Antecedents

Beyond the core post-adoption mechanism explained by ECM-IS and the usability-related cognition captured by the TAM-derived construct of perceived ease of use, older adults’ continuance intention toward smart homes is also shaped by broader contextual antecedents. In this study, government policy, compatibility, cost, and intergenerational technical support are conceptualized as China-specific contextual antecedents that shape older adults’ post-adoption perceptions and evaluations [18,35,44,46].

The development of smart homes for older adults in China has been predominantly policy-driven, with government policy playing a decisive role in enabling rapid growth [31,71]. At the same time, diffusion has been constrained by limited market incentives and low public awareness [31]. To address these barriers, the government issued a series of national measures to stimulate innovation and market uptake in smart health and elderly care, including the “Notice on the Action Plan for the Development of the Smart Health and Elderly Care Industry (2021–2025)” [72], the “Notice on Organizing the Application Process for the 2022 Smart Health and Elderly Care Products and Services Promotion Catalogue” [17], and the “Three Ministries Increase Efforts to Promote Smart Health and Elderly Care Products and Services” [16]. These policies collectively provide strategic guidance, financial support, and technical standards to support the national objectives. However, empirical research has rarely modeled government policy as an explicit contextual determinant of older adults’ continuance intention.

Compatibility is consistently identified as a key factor in acceptance [27,28]. In the Chinese smart home market, the absence of unified technical standards for elderly-oriented services has led to platform fragmentation. Providers often build proprietary ecosystems that weaken interoperability across devices and services, creating integration barriers for end users [31,73]. Policies such as “Standardization and Improvement of Smart Technologies, Reshaping Lifestyle-Smart Home, Lighting Up “E-life” were introduced to encourage standardization [15]. Although compatibility has been shown to affect behavioral intention in the pre-adoption stage [27], its influence on continuance intention remains underexplored.

Adoption and continued use of smart homes entail not only initial purchase costs but also installation, maintenance, and repair costs and, in some cases, increased energy expenses [58,74,75]. Cost has repeatedly been identified as a barrier to acceptance [9,32,76] and it significantly influences pre-adoption behavioral intention [33,34]. However, its effect on older adults’ continuance intention after adoption has not yet been systematically examined.

Assistance provided by family members across generations is another distinctive feature of the Chinese context and has been shown to influence older adults’ acceptance of smart homes [23,24,29]. Cultural norms around family caregiving vary across countries. In many Western settings, resource transfers tend to flow from parents to adult children [77], whereas in Chinese families, adult children commonly provide sustained instrumental and technological support to aging parents, reflecting filial piety [78,79] and thereby leading to prevalent upward intergenerational support [80]. While previous studies have demonstrated that such support affects behavioral intention, its influence on post-adoption continuance intention remains underexplored.

Overall, this study identifies several gaps. First, most previous studies have primarily emphasized potential users’ behavioral intention, with comparatively limited attention to actual users’ continuance intention after initial adoption [48]. Second, it remains unclear whether perceived ease of use and perceived usefulness continue to influence continuance intention after adoption. Third, although the Expectation–Confirmation Model of Information Systems provides a theoretically grounded explanation of post-adoption behavior, it has rarely been applied in the context of smart homes for older adults. Finally, China-specific contextual antecedents, including government policy, compatibility, cost, and intergenerational technical support, remain underexamined in explaining older adults’ continuance intention. In response, this study integrates the ECM-IS with TAM and incorporates China-specific contextual antecedents to examine older adults’ continuance intention toward smart homes.

Therefore, this study develops a conceptual framework that integrates the ECM-IS and the TAM to explain older adults’ continuance intention to use smart homes in the post-adoption stage. Specifically, ECM-IS provides the core post-adoption explanatory mechanism, while TAM complements the model by capturing usability-related cognitions. In addition, four China-specific contextual antecedents are incorporated as exogenous conditions that shape older adults’ post-adoption perceptions and evaluations.

2.2.4. Conceptual Framework

By integrating ECM-IS with the TAM-derived construct of perceived ease of use and incorporating China-specific contextual antecedents, this study proposes a conceptual framework to explain older adults’ continuance intention toward smart homes in Shandong Province, China (see Figure 6). Rather than treating all predictors as parallel independent variables, the framework is hierarchically structured. Continuance intention serves as the focal outcome variable. Perceived ease of use, perceived usefulness, confirmation, and satisfaction constitute the core post-adoption cognitive and evaluative mechanisms (derived from ECM-IS and the TAM-based usability cognition pathway). By contrast, government policy, compatibility, intergenerational technical support, and cost are positioned as exogenous contextual antecedents. These antecedents represent institutional, technological, family-support, and economic conditions, respectively. They shape older adults’ post-adoption cognitions and evaluations and, in some cases, also exert direct effects on continuance intention. Among these constructs, government policy, satisfaction, perceived usefulness, perceived ease of use, and cost are specified as direct paths to continuance intention, whereas government policy, compatibility, intergenerational technical support, confirmation, perceived usefulness, and perceived ease of use are specified as mechanism paths that support the theorized explanatory mechanism of continuance intention.

2.2.5. Hypotheses Development

Direct-Effect Hypotheses on Continuance Intention

Satisfaction and continuance intention

In ECM-IS, satisfaction is a key factor influencing continuance intention, a relationship that has been supported across multiple post-adoption settings [22,81,82]. For example, Kaium, Bao, Alam and Hoque [81] proposed that satisfaction positively influences continuance intention in mobile health services. This relationship was confirmed by Bae, Jo, Jung and Lee [82] and Wang, Kim and Kim [22] in studies on mobile social network services and IoT devices. In the context of smart homes for older adults, this relationship is salient because continuance intention depends on whether continued use yields stable, satisfactory experiences in daily life. Given the post-adoption nature of continuance intention, satisfaction is expected to remain a direct determinant of continuance intention [22]. Accordingly, this study proposes the following hypothesis.

Hypothesis 1 (H1):

Satisfaction has a positive effect on older adults’ continuance intention toward smart homes.

Perceived usefulness and continuance intention

Within the ECM-IS framework, perceived usefulness is a central determinant of continuance intention [49]. This association has been consistently validated in empirical studies [82,83]. Xie, Jia and He [83] confirmed a positive effect of perceived usefulness on continuance intention in older adults’ shared nursing services. In research on mobile social network services, Bae, Jo, Jung and Lee [82] reported similar findings. In the smart home context, older adults are more likely to continue using technologies that deliver tangible benefits in everyday life, such as health management support, improved quality of life, and greater convenience. Accordingly, when smart homes are perceived as practically beneficial rather than merely novel, continuance intention should be enhanced. Therefore, this study proposes the following hypothesis.

Hypothesis 2 (H2):

Perceived usefulness has a positive effect on older adults’ continuance intention toward smart homes.

Government policy and continuance intention

Government policy refers to targeted actions undertaken to solve, mitigate, or prevent problems [84,85]. In this study, government policy is operationalized at the individual level and measured by older adults’ perceptions of policy support related to smart homes, rather than by an objective macro-level policy index. In China, policy support has played an important role in promoting the development of smart homes for older adults [31]. Recent policy initiatives in China (e.g., public promotion, subsidies, and trade-in support for age-friendly smart products) further indicate ongoing institutional efforts to create a more supportive environment for smart home use among older adults [86,87]. Official policy reports have also noted that such measures may help reduce product idleness and encourage intention to continue using smart homes [88,89].

Prior research in other digital technology contexts has also reported that government policy positively influences users’ continuance intention [79]. In the smart home context, when older adults perceive clear policy endorsement, public promotion, and supportive measures, they may develop greater confidence in the reliability and long-term viability of smart home services, thereby directly enhancing their intention to continue using smart homes. Accordingly, this study proposes the following hypothesis.

Hypothesis 3 (H3):

Government policy has a positive effect on older adults’ continuance intention toward smart homes.

Perceived ease of use and continuance intention

In TAM, perceived ease of use is the extent to which users believe that interacting with a product or system requires minimal effort [60]. Previous studies have suggested that perceived ease of use positively influences continuance intention [83,90]. For example, Xie, Jia and He [83] proposed that perceived ease of use positively affects continuance intention in shared nursing services for older adults. Jia, Pang, Huang and Hou [90] also suggested that perceived ease of use positively influences continuance intention in live streaming. In the context of older adults’ use of smart homes, this relationship is particularly important because continuance intention may be weakened when routine operation, setup, troubleshooting, or maintenance are perceived as cognitively demanding. Therefore, in the post-adoption stage, lower effort expectancy is expected to strengthen continuance intention. Accordingly, this study proposes the following hypothesis.

Hypothesis 4 (H4):

Perceived ease of use has a positive effect on older adults’ continuance intention toward smart homes.

Cost and continuance intention

Costs include purchase, installation, maintenance, repair, and other ongoing expenditures. Users consider both upfront outlays and long-term financial burdens when evaluating smart homes [58,74]. Smart home devices may also increase operating expenses, such as energy consumption [75]. Prior research found that cost negatively affects continuance intention [43,74]. In this context, this relationship may be particularly salient for older adults, especially those with relatively fixed or limited incomes, because smart home use involves not only the initial purchase cost but also post-adoption expenditures, such as ongoing maintenance and potential repair costs. As these expenses accumulate over time, they may increase uncertainty regarding continued use. Therefore, cost is expected to remain a relevant inhibitor of continuance intention even after initial adoption [58]. Accordingly, this study proposes the following hypothesis.

Hypothesis 5 (H5):

Cost has a negative effect on older adults’ continuance intention toward smart homes.

Mechanism-Linking Hypotheses Supporting Continuance Intention

Government policy, perceived usefulness, confirmation, and perceived ease of use

In China, government policy plays a critical role in driving the development of smart homes for older adults [31]. Prior research in the digital industry has shown that government policy is positively associated with perceived ease of use, perceived usefulness, and confirmation, thereby supporting the extension of this logic to the smart home context [79]. However, in the context of smart home adoption among older adults, government policy should not be treated merely as a macro-level factor facilitating technology diffusion. Rather, it can shape older adults’ subjective perceptions through specific mechanisms, including information dissemination, user guidance, onboarding support, standardization and certification, and financial support.

First, government policy may strengthen perceived usefulness by increasing the visibility, legitimacy, and practical relevance of smart home functions. In China, policy-led promotion of smart eldercare services and age-friendly home-retrofitting initiatives may help older adults better understand how smart homes can support daily living, safety, and independence. When such policy support improves awareness and reduces uncertainty about the functional value of smart homes, older adults are more likely to perceive them as useful [72]. Accordingly, this study proposes the following hypothesis.

Hypothesis 6 (H6):

Government policy has a positive effect on older adults’ perceived usefulness of smart homes.

Second, government policy may enhance confirmation by reducing the gap between pre-use expectations and the actual experience. Policy-related actions, such as standards development, testing, certification, and interoperability protocols, can improve product reliability and service consistency, helping older adults experience smart homes in ways that align with what they have been led to expect. In addition, policy tools such as trade-in programs, subsidies, and trial initiatives may reduce barriers to trying smart homes and improve the quality of early use experiences, thereby strengthening expectation confirmation [91]. Accordingly, this study proposes the following hypothesis.

Hypothesis 7 (H7):

Government policy has a positive effect on older adults’ confirmation of smart homes.

Third, government policy may improve perceived ease of use by shaping the external support conditions that surround smart home use. Chinese policy initiatives often emphasize age-friendly transformations and digital technology training for older adults, which, in principle, can reduce cognitive burden and operational barriers [72]. Although the implementation and reach of such support may vary across local contexts, the theoretical expectation remains that stronger perceived policy support is associated with greater perceived ease of use. Accordingly, this study proposes the following hypothesis.

Hypothesis 8 (H8):

Government policy has a positive effect on older adults’ perceived ease of use of smart homes.

Compatibility and confirmation

Prior work on chatbot services has shown that compatibility enhances users’ post-use reactions and evaluations [92]. Strengthening functional compatibility can, therefore, help to confirm users’ initial expectations. Moreover, Park and Lee [93] examined AI and systemic factors that improve chatbot sustainability and identified compatibility as a key determinant of confirmation. Extending this logic to the Chinese smart home context, smart door locks and voice-enabled devices are widely adopted by older adults because they support biometric authentication and voice control. In these usage settings, compatibility across devices and platforms, as well as the fit between smart home functions and older adults’ usage habits, becomes especially important for ensuring that actual use experiences align with prior expectations [46]. Therefore, greater compatibility is expected to strengthen confirmation. Accordingly, this study proposes the following hypothesis.

Hypothesis 9 (H9):

Compatibility has a positive effect on older adults’ confirmation of smart homes.

Intergenerational technical support and perceived ease of use

Intergenerational technical support is increasingly common in families in the digital age [24,29,94,95]. The study by He and Huang [29] highlighted the importance of this approach in facilitating smartphone use among urban older adults. Zhou, Qian and Kaner [23] conducted a TAM-based survey on smart home use and found that intergenerational technical support positively affects perceived ease of use. In the context of smart home use in China, intergenerational support remains a prominent source of practical assistance for older adults. Through intergenerational technical support, older adults may receive help in reducing operational uncertainty, understanding device functions, and troubleshooting usage problems. Consequently, they are more likely to perceive smart homes as manageable and less effortful to use [47]. Accordingly, this study proposes the following hypothesis.

Hypothesis 10 (H10):

Intergenerational technical support has a positive effect on older adults’ perceived ease of use of smart homes.

Perceived usefulness and satisfaction

Within the ECM-IS framework, perceived usefulness is a key determinant of satisfaction [49], and this linkage has been consistently validated [82,96,97]. For example, Wu, Song and Lin [96] proposed that perceived usefulness positively influences satisfaction in health management apps. This relationship was further supported by Bae, Jo, Jung and Lee [82] and Rahi, Mansour, Alharafsheh and Alghizzawi [97]. In the context of older adults’ use of smart homes, satisfaction is expected to increase when older adults perceive that smart homes deliver meaningful benefits in their daily lives. Therefore, perceived usefulness is expected to positively influence satisfaction in the post-adoption stage. Accordingly, this study proposes the following hypothesis.

Hypothesis 11 (H11):

Perceived usefulness has a positive effect on older adults’ satisfaction with smart homes.

Confirmation, perceived usefulness, satisfaction, and perceived ease of use

Empirical studies have supported the relationships among confirmation, perceived usefulness, perceived ease of use, and satisfaction [81,83,98]. For instance, Zhu, Jiang and Cao [98] found that confirmation significantly influences perceived usefulness in the context of non-face-to-face telemedicine. Confirmation is expected to enhance perceived usefulness. When older adults find that smart homes perform as expected or better than expected, they are more likely to revise their judgments in favor of the technology’s practical value. In the Chinese smart home context, this mechanism is important because uncertainty about product reliability and real-world utility remains a barrier for many older adults. Once usage experiences confirm initial expectations, perceived usefulness is likely to increase. Accordingly, this study proposes the following hypothesis.

Hypothesis 12 (H12):

Confirmation has a positive effect on older adults’ perceived usefulness of smart homes.

Furthermore, confirmation is expected to strengthen satisfaction. Prior research has shown that confirmation positively influences satisfaction in post-adoption contexts [81]. ECM-IS posits that satisfaction arises when actual performance meets or exceeds prior expectations, a logic that is especially relevant to older adults’ post-adoption evaluations of smart homes. When older adults perceive that smart home use aligns with or exceeds their expectations, they are more likely to form a positive affective evaluation. Accordingly, this study proposes the following hypothesis.

Hypothesis 13 (H13):

Confirmation has a positive effect on older adults’ satisfaction with smart homes.

Moreover, confirmation is expected to improve perceived ease of use. Prior research has identified confirmation as a significant determinant of perceived ease of use in post-adoption service contexts [83]. In the smart home context, older adults’ judgments of perceived ease of use are shaped not only by interface design and service support, but also by whether their actual interaction experience aligns with their expectations. When older adults encounter fewer unexpected difficulties than initially feared, they are more likely to perceive smart homes as easier to use. Accordingly, this study proposes the following hypothesis.

Hypothesis 14 (H14):

Confirmation has a positive effect on older adults’ perceived ease of use of smart homes.

Perceived ease of use, perceived usefulness, and satisfaction

Prior research has confirmed the influence of perceived ease of use on perceived usefulness and satisfaction [9,25,90]. For example, Pal, Funilkul, Vanijja and Papasratorn [9] and Yan and Lee [25] showed that perceived ease of use positively affects perceived usefulness in the contexts of smart home services and smart home healthcare systems. In the post-adoption context of smart homes for older adults, when older adults can operate smart homes easily and independently, they may spend less time and effort on operation and rely less on others for external assistance. This can improve usage efficiency and daily convenience, thereby increasing the likelihood that smart homes are judged as useful. Accordingly, this study proposes the following hypothesis.

Hypothesis 15 (H15):

Perceived ease of use has a positive effect on older adults’ perceived usefulness of smart homes.

In addition, perceived ease of use is expected to enhance satisfaction. For instance, Jia, Pang, Huang and Hou [90] suggested that perceived ease of use positively influences satisfaction in post-adoption use contexts. In smart home use, low operational friction can reduce frustration, dependence on others, and perceived burden, thereby improving older adults’ overall affective evaluation. This mechanism is especially important because sustained satisfaction often depends on whether smart homes can be used smoothly in daily life. Therefore, this study proposes the following hypothesis.

Hypothesis 16 (H16):

Perceived ease of use has a positive effect on older adults’ satisfaction with smart homes.

3. Research Methodology

3.1. Geographic Context of the Study

The study was conducted in Shandong Province, eastern China (Figure 7). Shandong has the largest aging population in China (20.9%; approximately 21.2 million) [2,99]. It is also a national hub for the smart-home industry (≈22,600 firms) with leading manufacturers such as Haier and Hisense [100]. As the birthplace of Confucian culture, Shandong is characterized by strong norms of filial piety that foster intergenerational support [101]. These features provide a pertinent context for examining older adults’ continuance intention toward smart homes.

3.2. Research Method

This study adopted a quantitative approach. Data were collected using an online questionnaire survey, which has been widely used in this research area [23,24,25,33,39]. Data screening and descriptive statistics were conducted using Statistical Package for the Social Sciences (SPSS), and structural relationships were examined using Partial Least Squares Structural Equation Modeling (PLS-SEM) [102]. PLS-SEM was selected because it is suitable for predictive research with complex models and does not require multivariate normality [102,103].

3.3. Measures and Instruments

Quantitative research is typically designed to generalize from samples to populations [104] and commonly relies on questionnaires to generate analyzable data [105]. Consistent with prior studies in this field, this study employed a structured online questionnaire with closed-ended items [24,25,39].

The instrument was first developed in English, translated into Chinese to ensure comprehensibility for older adults in China, and then back-translated into English to verify semantic equivalence. It comprised two sections. Section 1 captured demographic information using single-choice items. Section 2 measured the study constructs using the scale items listed in

Table 1. Scale items of constructs.

Constructs	Items	Questions (7-Point Likert Scale)	References
Government Policy (GP)	GP 1	Government-led campaigns and educational initiatives have raised public awareness of smart homes.	[85,94]
	GP2	Substantial incentives (subsidies or tax breaks) offered by the government have lowered the costs of adopting smart homes.
	GP 3	Government policies have promoted the public’s adoption of smart homes.
Compatibility (COM)	COM 1	Different smart home devices can operate seamlessly with one another.	[9,30,94]
	COM 2	Smart homes from different vendors can interoperate seamlessly.
	COM 3	The use of smart homes aligns with all aspects of my daily life.
Intergenerational Technical Support (ITS)	ITS 1	My children guide me in learning how to use smart homes.	[23,24,27]
	ITS 2	With my children’s guidance, I can easily operate smart homes.
	ITS 3	My children assist me when I encounter difficulties using smart homes.
Perceived Usefulness (PU)	PU 1	I think smart homes can improve my health.	[83]
	PU 2	I think smart homes can enhance my quality of life.
	PU 3	Smart homes give me more options.
	PU 4	I think smart homes are highly useful.
Perceived Ease of Use (PEOU)	PEOU 1	The operation process of smart homes is simple.	[83]
	PEOU 2	I do not have to think hard when using smart homes.
	PEOU 3	I find smart homes easy to use.
	PEOU 4	I find it easy to get smart homes to do what I want it to do.
Cost (CS)	CS 1	I believe that purchasing smart homes requires a substantial financial investment.	[58]
	CS 2	I find maintaining the operation of smart homes to be a financial burden.
	CS 3	I find the potential repair costs of smart homes to be high.
Confirmation (CNF)	CNF 1	The service level provided by smart homes was better than what I expected.	[97,98]
	CNF 2	The various features of smart homes were better than what I expected.
	CNF 3	Smart homes were easier to use than what I expected.
	CNF 4	Overall, most of my expectations toward smart homes were confirmed.
Satisfaction (SAT)	SAT 1	The decision to use smart homes is right and wise.	[83]
	SAT 2	The experience of using smart homes will make me feel happy.
	SAT 3	Overall, I am satisfied with smart homes.
	SAT 4	I am confident in the development of smart homes.
Continuance Intention (CI)	CI 1	I am interested in smart homes.	[83]
	CI 2	I would like to continue using smart homes.
	CI 3	I would recommend smart homes to my family and friends.
	CI 4	I think we should encourage people to use smart homes.

. All items were rated on a 7-point Likert scale, ranging from 1 (“strongly disagree”) to 7 (“strongly agree”). Although the government policy items refer to policy campaigns, incentives, and policy promotion, respondents rated them based on their own awareness and perceptions. Accordingly, government policy was treated as an individual-level perceptual construct rather than an objective macro-level policy indicator. Although satisfaction and continuance intention are both post-adoption constructs and may be positively related, they capture different aspects of user responses. In this study, satisfaction was operationalized as respondents’ post-use evaluative and affective appraisal of their smart home experience, whereas continuance intention was operationalized as a future-oriented behavioral intention to continue using and recommending smart homes. The corresponding items were adapted from validated prior scales to preserve comparability with existing research.

3.4. Sampling Technique

Quantitative research requires a sampling technique that yields a sample sufficiently representative of the target population [106]. In this field, the convenience sampling technique has been most commonly used [23,24,37,107]. Guided by prior research, this study employed a convenience sampling technique. This technique selects respondents based on accessibility and willingness to participate because it is simple, cost-effective, and efficient [108].

To reduce potential sampling bias and enhance representativeness, proportional quotas were implemented based on the distribution of older adults across the province’s four cities with the highest proportions of older adults (Linyi, Weifang, Qingdao, and Jinan) [67], with the number of respondents from each city matching its share of the older adult population. Using the sample size estimation formula, a minimum of 384 valid responses was required; therefore, city-level quotas were rounded up to ensure this minimum was met.

Table 2. Quota distribution of the minimum required sample size by city in Shandong Province.

City in Shandong	Older Adults	Proportion of Older Adults	Sample Size
Linyi	2,163,415	27%	104
Weifang	2,043,071	25%	96
Qingdao	2,042,649	25%	96
Jinan	1,837,112	23%	89

reports the minimum quotas by city: Linyi (104, 27%), Weifang (96, 25%), Qingdao (96, 25%), and Jinan (89, 23%); totals may differ slightly due to rounding.

3.5. Respondent

Inclusion criteria were defined based on age, prior experience, and region, in line with the field’s conventions. Respondents were aged 60 years and older [23,25,27,40,109], had prior experience using smart homes to ensure the relevance of their responses to continuance intention in the post-adoption stage [25], and resided in Shandong Province, specifically in Linyi, Weifang, Qingdao, and Jinan [99]. Shandong was selected because it has one of China’s largest aging populations, concentrated in these cities [99], ranks second nationally in the number of smart home-related companies [100], and is deeply rooted in Confucian culture that emphasizes filial piety and intergenerational support [101].

3.6. Sample Size

Determining an appropriate sample size helps minimize sampling error and improve the reliability of inferences [110,111]. For large or effectively infinite populations, this study used the standard formula [112]:

$$n_{60}={\displaystyle \frac{z^2\times p\times \left(1-p\right)}{e^2}},$$

where n₆₀ is the required sample size for individuals aged 60 years and older, z is the z-score for the desired confidence level, p is the estimated population proportion, and e is the margin of error (precision). Using z = 1.96 (95% confidence), p = 0.5 (maximum variability), and e = 0.05 (±5 percentage points), the calculation yielded

$$n_{60}={\displaystyle \frac{{1.96}^2\times 0.5\times \left(1-0.5\right)}{{0.05}^2}}\approx 384.$$

Thus, at least 384 valid responses were required to achieve the desired level of precision.

3.7. Data Collection

The research was reviewed and approved by the Biomedical Ethics Committee of Qufu Normal University (Approval No. 2025107). Before accessing the questionnaire, all respondents provided written electronic informed consent. An online information page outlined the study objectives, the voluntary nature of participation, confidentiality safeguards, and participants’ rights. Access to the survey was technically restricted so that only individuals who selected “I agree” on a mandatory consent checkbox could proceed; without indicating consent, the questionnaire could not be opened or submitted.

Recruitment took place from 25 March 2025 to 9 April 2025. This study administered the questionnaire via WJX.cn, which generated a unique survey link. To prevent duplicate entries, IP address restrictions were enabled in the platform backend. Before starting the survey, respondents were asked to answer a screening question about their experience with smart homes. Only older adults with prior experience using smart homes were eligible to proceed, ensuring alignment with the study’s objective. The questionnaire was administered anonymously to protect privacy; no personally identifiable information was collected, and responses were not traceable to individuals. Respondents were informed that participation was voluntary and that they could withdraw at any time without providing a reason. Because the survey was administered online and anonymously, we could not directly verify whether each questionnaire was completed independently by the respondent or with assistance from family members or others. In addition, participation required basic internet access and the ability to use online tools, which may have made the sample more likely to include older adults with relatively stronger digital abilities. After data collection, responses were screened using pre-specified quality-control criteria independent of the measurement model results, including IP-based geographic eligibility checks, a minimum completion-time threshold, and routine checks for missing values and outliers. Importantly, no responses were removed based on factor loadings, reliability, or model-fit results.

3.8. Analysis Instruments

Data were analyzed using SPSS 24.0 and SmartPLS 4.0. In SPSS, data preparation included data entry, screening, handling of missing values, outlier diagnostics, and recoding where appropriate, and descriptive statistics were generated for the demographic variables [113]. PLS-SEM was then conducted in SmartPLS 4.0 to estimate and evaluate both the measurement and structural models [102,114]. Measurement model evaluation assessed the reliability and validity of constructs and indicators, while structural model evaluation tested the proposed hypotheses and assessed the overall model [39].

3.9. Pilot Study

A face validity review was conducted by a panel of five experts using criteria related to grammar, clarity, and precision, spelling, sentence structure, formatting consistency, and content relevance/adequacy [115]. Based on their feedback, several refinements were made.

Content validity was then assessed by six domain experts using the Item-Level Content Validity Index (I-CVI) and the Scale-Level Content Validity Index (S-CVI/Ave). Following Lynn [116], I-CVI was calculated from ratings on the four-point relevance scale recommended by Davis [117], with acceptable thresholds of 1.00 for ≤5 judges and ≥0.78 for ≥6 judges. All items met the I-CVI criteria, and the S-CVI/Ave was 0.98, exceeding the commonly accepted benchmark of 0.90 [116,117].

Before conducting the formal survey, a pilot study was conducted to evaluate completion time and to preliminarily assess reliability and validity [94]. Using the 10–20% guideline [118] and a minimum main-sample requirement of 384, the target range was 38 to 77 participants. A total of 77 responses were collected via WJX to maximize statistical power [119]. A subset of 14 pilot respondents completed the pilot questionnaire through a WJX link distributed via WeChat and provided feedback on item clarity and comprehensibility. In addition, the average completion time was 2–4 min, and questionnaires completed in less than 2 min were considered invalid.

4. Data Analysis and Results

4.1. Demographic Information

The online survey was administered on WJX.cn from 25 March to 9 April 2025. After data collection, the raw data were cleaned in Microsoft Excel and SPSS 24.0. Records were removed if the IP address fell outside the questionnaire’s target region or if the completion time was less than 2 min. Missing values and outliers were then assessed in SPSS using the “Frequencies” procedure for categorical variables, whereas the “Descriptive” procedure was used for continuous variables based on the valid N and inspection of the minimum and maximum values. After cleaning, 421 questionnaires were retained, yielding an effective response rate of 84.2%.

Descriptive statistics for the sample are presented in

Table 3. Demographic information for the respondents in the quantitative method.

Sample	Category	Frequencies (n)	Percentages (%)
Age	60–65 years	220	52.3
	66–70 years	155	36.8
	>70 years	46	10.9
Gender	Male	212	50.4
	Female	209	49.6
Education Level	Junior high school or below	118	28.0
	High school	173	41.1
	Junior college diploma	71	16.9
	Bachelor’s degree or above	59	14.0
Household Size	1 person	26	6.2
	2 persons	52	12.4
	3–4 persons	185	43.9
	≥5 persons	158	37.5
Employment Status	Employed	52	12.4
	Retired	369	87.6
Types of Occupation	Private enterprise	175	41.6
	Foreign-funded enterprise	64	15.2
	Government or public institution	83	19.7
	Freelance	99	23.5

. Most respondents were aged 60–65 years (52.3%), followed by those aged 66–70 years (36.8%) and those > 70 years (10.9%), indicating a predominance of younger seniors who may be more able or inclined to engage with smart homes. To facilitate efficient recruitment and data collection and to align with prior research in this domain, the questionnaire was administered online. This pattern may partly be attributable to the online survey format, which may have reduced participation among adults aged > 70 years. The gender distribution was nearly balanced (male: 50.4%, female: 49.6%). With respect to education level, the distribution was as follows: high school, 41.1%; junior high school or below, 28.0%; junior college diploma, 16.9%; and bachelor’s degree or above, 14.0%. Household size was reported as follows: 3–4 persons, 43.9%; ≥5 persons, 37.5%; 2 persons, 12.4%; 1 person, 6.2%. Regarding employment status, most respondents were retired (87.6%), whereas 12.4% were employed. Regarding occupational type, respondents were distributed as follows: private enterprise (41.6%), foreign-funded enterprise (15.2%), government or public institution (19.7%), and freelance (23.5%).

4.2. Common Method Bias (CMB) Testing

Common Method Bias (CMB) was assessed to address potential same-source bias in the survey data [120]. This study applied the full collinearity Variance Inflation Factor (VIF) procedure and adopted the recommended cutoff value of 3.33 [39,102,121]. As reported in

Table 4. Full collinearity VIF values for assessing common method bias.

Construct	CI	CNF	COM	CS	GP	ITS	PEOU	PU	SAT
Continuance Intention (CI)
Confirmation (CNF)							1.786	2.009	2.106
Compatibility (COM)		1.272
Cost (CS)	1.452
Government Policy (GP)	1.639	1.272					1.627	1.628
Intergenerational Technical Support (ITS)							1.318
Perceived Ease of Use (PEOU)	1.929							1.589	1.595
Perceived Usefulness (PU)	1.850								1.760
Satisfaction (SAT)	1.984

Note: All constructs exhibited VIF values below 3.33, indicating no problematic collinearity.

, all constructs exhibited VIF values below 3.33, indicating the absence of problematic collinearity and suggesting that CMB is unlikely to pose a substantive threat to the validity of the findings.

4.3. PLS-SEM Analysis

The SEM analysis consisted of two components: the measurement model and the structural model analysis. In addition, this study conducted the Importance-Performance Map Analysis (IPMA).

4.3.1. Measurement Model Analysis

To establish the quality of the measurement model, reliability and validity were assessed. Reliability refers to the internal consistency of the constructs, whereas validity refers to the extent to which the measurement items accurately represent the intended constructs [122].

Reliability

Reliability was assessed using Cronbach’s alpha and Composite Reliability (CR). For Cronbach’s alpha, values of at least 0.70 are considered acceptable, and values of 0.80 or higher are preferred [103,123]. CR accounts for unequal indicator loadings and is regarded as an appropriate reliability index for SEM. Values of at least 0.70 indicate adequate reliability [122,124,125].

Table 5. Reliability analysis.

Constructs	Items	Cronbach’s Alpha	Composite Reliability (CR)
Continuance Intention (CI)	CI1	0.942	0.958
	CI2
	CI3
	CI4
Confirmation (CNF)	CNF1	0.935	0.953
	CNF2
	CNF3
	CNF4
Compatibility (COM)	COM1	0.919	0.949
	COM2
	COM3
Cost (CS)	CS1	0.925	0.952
	CS2
	CS3
Government Policy (GP)	GP1	0.923	0.951
	GP2
	GP3
Intergenerational Technical Support (ITS)	ITS1	0.917	0.948
	ITS2
	ITS3
Perceived Ease of Use (PEOU)	PEOU1	0.945	0.960
	PEOU2
	PEOU3
	PEOU4
Perceived Usefulness (PU)	PU1	0.931	0.951
	PU2
	PU3
	PU4
Satisfaction (SAT)	SAT1	0.930	0.950
	SAT2
	SAT3
	SAT4

reports the reliability statistics for each construct. Cronbach’s alpha and CR were consistently above the recommended thresholds for all constructs, indicating excellent internal consistency and supporting the reliability of the measurement model.

Validity

Validity was assessed by examining convergent and discriminant validity [125].

Convergent validity

Convergent validity evaluates whether multiple indicators of a construct reflect the same latent variable [124,125]. This study assessed convergent validity using standardized factor loadings and the Average Variance Extracted (AVE) method. Commonly used criteria are factor loadings of at least 0.70 and AVE values of at least 0.50 to indicate adequate convergent validity [124,125]. As reported in

Table 6. Convergent validity analysis.

Constructs	Items	Factor Loadings	Average Variance Extracted (AVE)
Continuance Intention (CI)	CI 1	0.931	0.852
	CI 2	0.923
	CI 3	0.913
	CI 4	0.926
Confirmation (CNF)	CNF 1	0.908	0.836
	CNF 2	0.917
	CNF 3	0.910
	CNF 4	0.923
Compatibility (COM)	COM 1	0.931	0.860
	COM 2	0.921
	COM 3	0.931
Cost (CS)	CS 1	0.933	0.870
	CS 2	0.933
	CS 3	0.931
Government Policy (GP)	GP 1	0.931	0.867
	GP 2	0.930
	GP 3	0.933
Intergenerational Technical Support (ITS)	ITS 1	0.930	0.858
	ITS 2	0.929
	ITS 3	0.921
Perceived Ease of Use (PEOU)	PEOU 1	0.931	0.858
	PEOU 2	0.919
	PEOU 3	0.925
	PEOU 4	0.930
Perceived Usefulness (PU)	PU 1	0.915	0.829
	PU 2	0.906
	PU 3	0.909
	PU 4	0.912
Satisfaction (SAT)	SAT 1	0.929	0.827
	SAT 2	0.905
	SAT 3	0.905
	SAT 4	0.897

, all standardized factor loadings ranged from 0.897 to 0.933, and AVE values exceeded 0.820, demonstrating strong convergent validity.

Discriminant validity

Discriminant validity was assessed using the Fornell and Larcker criterion [124]. For each construct, the square root of the Average Variance Extracted (AVE) should be greater than its correlations with all other constructs [124]. In

Table 7. Fornell-Larcker criterion.

Construct	CI	CNF	COM	CS	GP	ITS	PEOU	PU	SAT
Continuance Intention (CI)	0.923
Confirmation (CNF)	0.601	0.915
Compatibility (COM)	0.463	0.583	0.927
Cost (CS)	−0.491	−0.419	−0.343	0.933
Government Policy (GP)	0.488	0.611	0.462	−0.451	0.931
Intergenerational Technical Support (ITS)	0.564	0.475	0.447	−0.348	0.388	0.926
Perceived Ease of Use (PEOU)	0.572	0.598	0.479	−0.392	0.455	0.653	0.926
Perceived Usefulness (PU)	0.654	0.646	0.507	−0.493	0.567	0.495	0.481	0.911
Satisfaction (SAT)	0.642	0.562	0.454	−0.412	0.419	0.517	0.658	0.523	0.909

Note: Bold diagonal values indicate the square roots of AVE.

, the diagonal elements report the square roots of the AVE, and the off-diagonal elements report inter-construct correlations. For all constructs, the diagonal entries exceed the corresponding correlations in the same row and column, thereby indicating satisfactory discriminant validity.

Discriminant validity was further assessed using the heterotrait–monotrait (HTMT) ratio [126]. Adopting the conventional criterion of 0.85 [24,126], all HTMT estimates in

Table 8. Heterotrait–Monotrait Ratio (HTMT).

Construct	CI	CNF	COM	CS	GP	ITS	PEOU	PU	SAT
Continuance Intention (CI)
Confirmation (CNF)	0.640
Compatibility (COM)	0.497	0.627
Cost (CS)	0.526	0.451	0.372
Government Policy (GP)	0.523	0.656	0.500	0.488
Intergenerational Technical Support (ITS)	0.606	0.513	0.487	0.377	0.420
Perceived Ease of Use (PEOU)	0.607	0.636	0.514	0.419	0.487	0.701
Perceived Usefulness (PU)	0.698	0.693	0.546	0.531	0.611	0.535	0.513
Satisfaction (SAT)	0.685	0.602	0.491	0.443	0.451	0.558	0.701	0.562

were below this threshold (ranging from 0.372 to 0.701), confirming that the constructs are empirically distinct.

4.3.2. Structural Model Analysis

Hypothesis testing results

The structural model was estimated in SmartPLS 4.0 using a non-parametric bootstrapping procedure with 5000 resamples to ensure robust path estimates [102]. Hypotheses were evaluated using path coefficients (β), t-statistics, and p-values. Statistical significance was defined as p-values below 0.05, 0.01, and 0.001 for statistically significant, highly significant, and very highly significant effects, respectively [102]. Path coefficients indicate the direction and magnitude of relationships, with larger absolute β values indicating stronger effects. As summarized in

Table 9. Structural path results for the proposed model.

Path Type	Hypothesis	Path Coefficient (β)	t-Statistics	p-Values	Hypothesis Status
Direct paths to CI	H1: SAT → CI	0.300	5.053	0.000 ***	Supported
	H2: PU → CI	0.340	5.824	0.000 ***	Supported
	H3: GP → CI	0.051	0.950	0.342	Not supported
	H4: PEOU → CI	0.140	2.392	0.017 *	Supported
	H5: CS → CI	−0.122	2.623	0.009 **	Supported
Mechanism paths	H6: GP → PU	0.258	4.778	0.000 ***	Supported
	H7: GP → CNF	0.434	9.020	0.000 ***	Supported
	H8: GP → PEOU	0.071	1.486	0.137	Not supported
	H9: COM → CNF	0.382	8.526	0.000 ***	Supported
	H10: ITS → PEOU	0.467	9.821	0.000 ***	Supported
	H11: PU → SAT	0.198	3.409	0.001 **	Supported
	H12: CNF → PU	0.422	7.445	0.000 ***	Supported
	H13: CNF → SAT	0.152	2.321	0.020 *	Supported
	H14: CNF → PEOU	0.333	6.163	0.000 ***	Supported
	H15: PEOU → PU	0.111	2.223	0.026 *	Supported
	H16: PEOU → SAT	0.472	8.040	0.000 ***	Supported

Note: * p < 0.05, ** p < 0.01, *** p < 0.001. The blue fonts indicate hypotheses that are not supported.

, 14 of the 16 hypotheses were supported, whereas H3 and H8 were not supported.

Figure 8 presents the structural paths for the quantitative model. Of the 16 hypotheses, 14 were supported, whereas H3 and H8 were not supported. Among the supported paths, the path from cost to continuance intention was the only one with a negative effect. Satisfaction and perceived usefulness showed the most significant direct impacts on continuance intention, while government policy, compatibility, intergenerational technical support, confirmation, and perceived ease of use played prominent roles in the model’s mechanism-based pathways linking antecedents to continuance intention.

Model explanatory and predictive power

Table 10. Explanatory and predictive power of the conceptual framework.

Construct	R2	Q2	f2
Continuance Intention	0.580	0.490
Perceived Ease of Use	0.536	0.457	0.024
Satisfaction	0.500	0.407	0.108
Perceived Usefulness	0.473	0.387	0.149

summarizes the model’s explanatory and predictive performance. The coefficient of determination (R²) quantifies explained variance [102,112]. According to Hair et al. [102], values of 0.75 or above are considered substantial, those between 0.50 and 0.75 are moderate, those between 0.25 and 0.50 are weak, and values below 0.25 are very weak. The R² for continuance intention is 0.580, indicating moderate explanatory power. Effect sizes (f²) were calculated to quantify each predictor’s impact on continuance intention. According to Cohen [112], f² values of 0.02, 0.15, and 0.35 denote small, medium, and large effects, respectively. Using these benchmarks, perceived ease of use showed a small effect (f² = 0.024), satisfaction showed a medium effect (f² = 0.108), and perceived usefulness showed the largest effect (f² = 0.149). Predictive relevance, assessed with Stone-Geisser’s Q² through blindfolding [102], was greater than zero for continuance intention (Q² = 0.490), indicating strong predictive accuracy.

Model fit assessment

In PLS-SEM, model fit was primarily assessed using the Standardized Root Mean Square Residual (SRMR), which captures the standardized discrepancy between observed and model-implied correlations. Values less than 0.08 indicate a good fit [102]. The Normed Fit Index (NFI) was reported as supplementary evidence, with values of 0.90 or higher generally considered acceptable [102]. As shown in

Table 11. Model fit of the conceptual framework.

Model Fit Index	Estimated Model	Threshold	Interpretation
SRMR	0.069	<0.08	Good fit
NFI	0.917	>0.90	Acceptable fit

, SRMR was 0.069, and NFI was 0.917, indicating that the model fit the data satisfactorily.

In addition, the Goodness-of-Fit (GoF) index was reported as a supplementary, holistic indicator of overall model fit [102,127]. GoF was computed as the geometric mean of the Average Variance Extracted (AVE) and the average R² [24,128]. The GoF of 0.666 exceeded the 0.36 benchmark, providing additional evidence of a strong overall model fit [24,128].

$$GoF=\sqrt{\overline{AVE}\times \overline{R^2}}=\sqrt{0.851\times 0.522}\approx 0.666$$

4.4. Importance-Performance Map Analysis (IPMA)

Importance-Performance Map Analysis (IPMA) was used to identify drivers of older adults’ continuance intention toward smart homes. IPMA evaluates each construct on two dimensions: importance, defined as the total effect on continuance intention, and performance, defined as the latent variable score on a 0–100 scale [129]. As shown in

Table 12. Importance-Performance Map Analysis of Factors Influencing Continuance Intention.

Latent Constructs	Importance (Total Effect of the Latent Variable Continuance Intention)	Performance (Index Values)
Perceived Usefulness (PU)	0.400	63.091
Perceived Ease of Use (PEOU)	0.326	59.593
Confirmation (CNF)	0.323	62.974
Government Policy (GP)	0.318	66.152
Satisfaction (SAT)	0.300	65.779
Intergenerational Technical Support (ITS)	0.152	59.084
Compatibility (COM)	0.123	56.68
Cost (CS)	−0.122	39.114

and Figure 9, priority should be given to factors that exhibit high importance but relatively low performance, as they have a substantial impact and the greatest potential for improvement. Based on these criteria, perceived ease of use emerges as the top priority for improvement. It shows high importance (importance score of 0.326) but relatively low performance (performance score of 59.6/100), suggesting that older adults do not yet perceive smart homes as sufficiently easy to use. This may reflect persistent usability and interaction barriers, particularly during onboarding and configuration, as noted in prior research [130]. By contrast, perceived usefulness has the highest importance (≈0.40) and relatively high performance (≈63.1), suggesting that improving it may be a lower short-term priority than improving other dimensions. Therefore, enhancing perceived ease of use is likely to yield substantial improvements in older adults’ continuance intention to use smart homes.

5. Discussion

This study examines older adults’ continuance intention toward smart homes from a post-adoption perspective. Overall, continuance intention is directly influenced by perceived usefulness, satisfaction, perceived ease of use, and cost, while contextual antecedents such as government policy, compatibility, and intergenerational technical support primarily operate through mechanism-based pathways.

5.1. Discussion of Direct Paths to Continuance Intention

H1: Satisfaction had a significant positive effect on continuance intention (β = 0.300, p = 0.000). This result is consistent with the Expectation-Confirmation Model of Information Systems (ECM-IS), which positions satisfaction as a central driver of continuance intention and aligns with previous studies [22,81,82]. In the smart home context, this finding suggests that older adults’ willingness to continue using smart homes depends substantially on their cumulative post-use evaluations. In other words, continued use is not sustained merely by initial adoption, but rather by whether the actual experience remains satisfactory over time.

H2: Perceived usefulness had a significant positive effect on older adults’ continuance intention toward smart homes (β = 0.340, p = 0.000). This is consistent with ECM-IS and prior post-adoption studies [82,83]. The finding indicates that older adults are more likely to continue using smart homes when they perceive clear practical benefits, such as convenience, safety, and support for independent living. This extends prior post-adoption findings to the context of smart homes and highlights that perceived usefulness remains central to sustaining long-term use.

H4: Perceived ease of use had a significant positive effect on older adults’ continuance intention toward smart homes (β = 0.140, p = 0.017). This result is consistent with prior research that identified perceived ease of use as a determinant of continuance intention among older adults [83,90]. This interpretation is further supported by usability research showing that smart home onboarding and setup can be time-consuming and unintuitive, with common friction points including complex Wi-Fi provisioning, unclear network support, insufficient error-recovery guidance, and confusing permission requests during device pairing and configuration [130]. For older adults, continuance intention depends not only on initial learning to use smart homes but also on whether routine operation, configuration, troubleshooting, and maintenance remain manageable.

H5: Cost had a significant negative effect on older adults’ continuance intention toward smart homes (β = −0.122, p = 0.009). This result is consistent with Liang, Li and Wei [43] and Park, Kim, Kim and Kwon [74], and extends prior evidence to the post-adoption stage of smart home use. Although the effect size is modest, its statistical significance indicates that cost remains relevant to continuance intention after initial adoption. In the post-adoption context, older adults may evaluate not only the purchase price but also the broader financial burden associated with continued use, especially when expenses are recurring or difficult to anticipate [131,132,133].

This pattern suggests that cost should be understood across the entire product life cycle rather than only at the point of sale. Beyond upfront expenditures, older adults may face ongoing costs for maintenance, repairs, upgrades, and service fees [131,132]. For older adults with relatively fixed or limited budgets, the accumulation of these costs over time may weaken their willingness to continue using smart homes. Taken together, these findings support conceptualizing cost in post-adoption research as a persistent economic constraint, rather than merely an adoption-stage barrier.

5.2. Discussion of Mechanism Paths

H6: Government policy had a significant positive effect on older adults’ perceived usefulness of smart homes (β = 0.258, p = 0.000), aligning with Arkanuddin, Susanti and Broto [85]. This suggests that supportive policies, market regulation, and related institutional arrangements may enhance older adults’ perceptions of the practical value of smart homes. In the post-adoption context, government policy appears to strengthen perceived usefulness indirectly by improving the broader environment in which smart home technologies are delivered and supported.

H7: Government policy had a significant positive effect on older adults’ confirmation of smart homes (β = 0.434, p = 0.000), aligning with prior research [85]. This suggests that policy support plays an important role in shaping whether older adults’ expectations are confirmed in actual use. A plausible explanation is that policy-driven regulation, service improvement, and market standardization can make smart home experiences more reliable and predictable, thereby strengthening confirmation.

H9: Compatibility had a significant positive effect on older adults’ confirmation of smart homes (β = 0.382, p = 0.000). This aligns with post-adoption research indicating that compatibility enhances confirmation [92,93]. This finding indicates that older adults are more likely to confirm their expectations when smart homes fit their routines, habits, and usage contexts. In the smart home setting, technology–life fit appears to be a key condition for positive post-adoption evaluation.

H10: Intergenerational technical support had a significant positive effect on older adults’ perceived ease of use of smart homes (β = 0.467, p = 0.000). This finding is consistent with He and Huang [29] and Zhou, Qian and Kaner [23]. This suggests that support from younger family members substantially reduces operational uncertainty for older adults. In the post-adoption stage, intergenerational technical support appears to help older adults manage smart homes more effectively and perceive them as easier to use in daily life.

H11: Perceived usefulness had a significant positive effect on older adults’ satisfaction with smart homes (β = 0.198, p = 0.001). This result is consistent with previous studies [82,96,97]. It suggests that when older adults perceive smart homes as genuinely beneficial in daily life, they are more likely to feel satisfied with their use. In the post-adoption context, perceived usefulness contributes not only directly to continuance intention (H2) but also to satisfaction, which in turn contributes to continuance intention.

H12: Confirmation had a significant positive effect on older adults’ perceived usefulness of smart homes (β = 0.422, p = 0.000). This finding aligns with Xie, Jia and He [83] and Zhu, Jiang and Cao [98]. This result indicates that when older adults’ initial expectations are confirmed through actual use, their evaluation of smart homes’ usefulness becomes more favorable. This supports the ECM-IS logic that confirmation plays a central role in shaping post-adoption cognitive appraisal.

H13: Confirmation had a significant positive effect on older adults’ satisfaction with smart homes (β = 0.152, p = 0.020). This result is consistent with prior research [81,83]. The effect size is smaller than some other paths, but it remains meaningful in the post-adoption chain. In the smart home context, when older adults perceive that the actual use experience matches or exceeds their prior expectations, they report higher satisfaction, which, in turn, supports continuance intention.

H14: Confirmation had a significant positive effect on older adults’ perceived ease of use of smart homes (β = 0.333, p = 0.000). This finding is consistent with Xie, Jia and He [83]. It suggests that when actual smart home use aligns with older adults’ expectations, they are more likely to perceive smart homes as manageable and user-friendly. The result reinforces the idea that perceived ease of use in the post-adoption stage is shaped not only by interface features but also by expectation-consistent experiences.

H15: Perceived ease of use had a significant positive effect on older adults’ perceived usefulness of smart homes (β = 0.111, p = 0.026). This finding is consistent with prior research [9,25]. Although the effect size is relatively small, the result indicates that usability continues to contribute to value perception in the post-adoption stage. For older adults, smart homes that are easier to operate are also more likely to be perceived as useful in practical terms.

H16: Perceived ease of use had a significant positive effect on older adults’ satisfaction with smart homes (β = 0.472, p = 0.000). This result is consistent with Jia, Pang, Huang and Hou [90]. This is one of the strongest paths in the model, indicating that usability is especially salient in shaping older adults’ post-adoption evaluation of smart homes. The result suggests that ease of operation and reduced interaction friction are central to maintaining satisfaction during the long-term use of smart homes.

5.3. Discussion of Non-Supported Paths

H3: Government policy did not have a statistically significant effect on older adults’ continuance intention to use smart homes (β = 0.051, p = 0.342). This finding contrasts with Arkanuddin, Susanti and Broto [85]. One possible explanation is that government policy primarily shapes the upstream conditions of smart home use (e.g., promotion, subsidies, service accessibility, and market order), rather than directly influencing continuance intention. This distinction may be particularly salient in the post-adoption stage: while policy support may facilitate acceptance, continuance intention is more likely to be influenced by older adults’ accumulated perceived usefulness, perceived ease of use, and satisfaction. In addition, policy effects may not be highly visible or directly attributable at the user level, as older adults often evaluate continued use based on immediate product and service experiences rather than on broader institutional arrangements. The present findings support this interpretation, as government policy significantly affects perceived usefulness (H6) and confirmation (H7), suggesting that its influence operates primarily through these constructs during the post-adoption process. Taken together, these results suggest that government policy functions as a distal enabling condition, whereas continuance intention is shaped more directly by proximal post-adoption cognitions and evaluations.

H8: Government policy did not have a statistically significant effect on older adults’ perceived ease of use of smart homes (β = 0.071, p = 0.137), which also contrasts with findings of Arkanuddin, Susanti and Broto [85]. This suggests that, compared with policy support, older adults’ perceptions of how easy smart homes are to use are shaped more by actual usage experiences and immediate support conditions, such as their digital literacy or technological literacy, interface design quality, operational complexity, and intergenerational support [13,24,134,135]. Although technical training organized by governments and community organizations can, in theory, enhance older adults’ perceived ease of use, its practical impact may be relatively limited because, in reality, such training often suffers from insufficient resources, low frequency of provision, and uneven accessibility, making it difficult for many older adults to obtain sustained training support.

6. Contributions, Implications, and Limitations

6.1. Theoretical Contributions

This study identified key determinants of older adults’ continuance intention toward smart homes, examined their interrelationships, and developed and validated a conceptual framework. It makes four theoretical contributions to the literature on older adults’ continuance intention toward smart homes. Rather than proposing a fully generalizable theory, this study offers a context-bound theoretical extension by explaining how China-specific contextual antecedents shape older adults’ post-adoption cognition, evaluation, and continuance intention toward smart homes.

First, this study shifts the analytical focus from pre-adoption behavioral intention to post-adoption continuance intention. Existing smart home research on older adults has largely emphasized willingness to adopt or behavioral intention before sustained use experience is gained. By contrast, this study examines continuance intention in the post-adoption stage, where judgments are shaped by accumulated experience, confirmation, and satisfaction. This shift is theoretically important because the drivers of continuance intention differ from those of pre-adoption behavioral intention, and post-adoption theorizing is necessary to explain long-term engagement with smart homes.

Second, this study develops and validates a mechanism-level integrated framework for older adults’ continuance intention toward smart homes, grounded in ECM-IS and extended with the TAM-derived construct of perceived ease of use (PEOU). ECM-IS provides the core post-adoption continuance mechanism, while PEOU is selectively incorporated from TAM to capture usability-related cognition in older adults’ continuance intention toward smart homes. This integration clarifies the theoretical relationship between ECM-IS and TAM in the post-adoption context.

Third, this study extends post-adoption continuance theory by modeling China-specific contextual antecedents as exogenous conditions that shape post-adoption cognition and evaluation. Government policy, compatibility, and intergenerational technical support are not treated as parallel predictors with identical functions. Instead, the findings support a hierarchical explanatory structure in which these contextual antecedents influence continuance intention through confirmation, perceived usefulness, perceived ease of use, and satisfaction. This contributes to the literature by specifying where contextual antecedents enter the continuance process and how they operate through mechanisms in the Chinese smart home context.

Fourth, this study contributes a more nuanced conceptualization of cost in smart home continuance research. The significant negative effect of cost indicates that cost remains relevant after adoption, suggesting that cost should not be conceptualized solely as an adoption-stage barrier. Instead, in the post-adoption context, cost should also be understood as an ongoing economic burden that can weaken older adults’ intention to continue using smart homes. This refines smart home continuance theorizing by conceptualizing cost not only as an adoption-stage barrier but also as a persistent post-adoption constraint.

6.2. Practical and Policy Implications

This study offers practical and policy implications for government agencies, smart home providers, communities, families, and older adults.

6.2.1. Government-Level Policy Implications

The findings indicated that government policy operated mainly through upstream mechanisms rather than directly influencing continuance intention. Policy can play this role by reducing financial burden through targeted subsidies for older adults, standardizing product quality and interoperability through regulation, and promoting awareness and basic operational literacy through public campaigns and guided demonstrations. In rapidly aging regions, policy should not only subsidize access but also ensure that products are reliable, comprehensible, and perceived as relevant to daily living. Establishing and enforcing interoperability standards across vendors is especially important to reduce fragmentation, which is a persistent barrier for older adults. Policymakers should also consider the post-adoption costs of smart home use. Beyond subsidies and market regulation, policy efforts may further improve continuance intention by promoting pricing transparency (e.g., clearer disclosure of one-time and recurring costs) and supporting age-friendly service schemes for older adults with limited financial resources.

6.2.2. Organizational-Level Implementation Implications for Smart Home Providers

Satisfaction and perceived usefulness are key factors that influence continuance intention. Developers should therefore prioritize user-centered design for older adults by enabling simplified setup and operation, providing intuitive interfaces, and supporting voice or gesture control. Designs that emphasize reliability, autonomy, and companionship are more likely to sustain engagement. Improving perceived ease of use also remains crucial; this can be achieved by minimizing installation and configuration complexity and ensuring seamless integration across functions, which in turn supports long-term use.

Moreover, the negative effect of cost on continuance intention highlights the importance of lifecycle affordability. Developers should not only reduce upfront purchase cost but also lower ongoing financial burdens by offering transparent cost structures, modular service packages, and predictable maintenance or upgrade plans. Such strategies may help strengthen older adults’ intention to continue using smart homes.

6.2.3. Community- and Family-Level Support Implications

Intergenerational technical support emerged as a critical factor shaping perceived ease of use. Families, especially adult children, can reduce anxiety, assist older adults during onboarding, and reinforce confidence in everyday operation. Communities can complement family support by providing basic guidance, hands-on demonstrations, peer-led informal instruction, recurring age-friendly training sessions, simple troubleshooting workshops, and volunteer-based digital assistance. Such social and practical support may help mitigate age-related declines in memory or dexterity, thereby strengthening older adults’ intention to continue using smart homes.

6.2.4. User-Level Implications for Older Adults

For older adults, reflecting on personal needs and actual use experiences can help clarify which smart home functions are genuinely valuable and which barriers persist. Communicating these needs and difficulties to family members, community service providers, and smart home developers may facilitate more targeted support, reduce frustration during use, and enhance overall satisfaction. In addition, providing regular feedback on usability and service issues can help other stakeholders create a more user-centered and age-friendly living environment.

6.3. Limitations and Recommendations for Future Research

Several limitations should be acknowledged, and they also serve as directions for future research.

First, this study employed a cross-sectional design, which cannot capture how older adults’ perceptions evolve with prolonged exposure to smart homes. Future studies could adopt longitudinal designs to track these dynamics over time.

Second, the sample was drawn only from older adults in Shandong Province, China. Although this regional focus offers context-specific insights, it may limit the generalizability of the findings. Future research should include participants from diverse regions and cultural contexts to test the robustness of the proposed conceptual framework and examine potential socio-cultural moderators of continuance intention. Cross-regional and cross-cultural comparisons would strengthen external validity and yield more context-sensitive guidance for policy and design.

Third, this study relied exclusively on a quantitative survey design. While this approach enabled hypothesis testing and model estimation, it offered limited insight into older adults’ nuanced, context-specific experiences with smart homes. Future research could adopt mixed-methods designs that combine large-scale surveys with qualitative approaches, such as observations, interviews, and focus groups, to better uncover underlying mechanisms and enrich the interpretation of the statistical results.

Fourth, the sample was heavily concentrated in the 60–70 age range (89.1%), with only 10.9% of respondents aged 70 or older, thereby limiting the sample’s representativeness for older adults in more advanced age groups. Individuals over 70 are more likely to experience functional, sensory, and cognitive limitations that may shape their access to and acceptance of smart homes. Accordingly, the findings may primarily reflect the perspectives of relatively younger older adults and should be generalized to older age groups with caution. Future research should recruit a larger proportion of respondents aged 70 or older.

Fifth, because the survey was administered online and anonymously, we could not verify whether respondents completed their responses independently or with assistance from family members or others. This may have affected how some items were interpreted and answered. Future research should use mixed-mode data collection and explicitly record whether responses were completed independently or with assistance to improve response quality and external validity.

Sixth, the online survey design may have introduced selection bias, as older adults with limited digital literacy or restricted internet access were less likely to participate. As a result, the sample may overrepresent older adults with stronger digital capabilities. Therefore, the findings should be interpreted as most applicable to older adults with prior experience of using smart homes and at least basic digital literacy.

7. Conclusions

This study examined older adults’ continuance intention to use smart homes in the post-adoption stage, focusing on participants with prior smart home experience in Shandong Province, China. Drawing on the Expectation–Confirmation Model of Information Systems (ECM-IS) and the Technology Acceptance Model (TAM), the study developed and validated an integrated framework that combines post-adoption cognitive and evaluative mechanisms with China-specific contextual antecedents. The findings show that satisfaction, perceived usefulness, and perceived ease of use are the strongest positive predictors of continuance intention, whereas cost exerts a significant negative effect.

The results further indicate that China-specific contextual antecedents shape continuance intention mainly through mechanism-based pathways rather than solely through direct effects. In particular, government policy did not directly influence continuance intention; rather, it influenced upstream post-adoption perceptions via confirmation and perceived usefulness, while compatibility and intergenerational technical support exerted their effects via confirmation and perceived ease of use, respectively. These findings provide a context-bound theoretical extension to smart home continuance research by clarifying how institutional, technological, family support, and economic conditions jointly shape older adults’ post-adoption cognition, evaluation, and continuance intention in the Chinese context. Overall, the study highlights that sustaining smart home use among older adults requires not only usable and useful technologies but also supportive contextual conditions that remain salient after adoption.