Empirical Study on the Factors Affecting User Stickiness of Online Visual Art Platform From the Perspective of User Experience

This paper aims to analyze the factors affecting user stickiness of the online visual art platform by developing a Push-Pull-Mooring (PPM) framework from the perspective of user experience. The study collected 332 valid responses through an online questionnaire and tested the research model using structural equation modeling. The results show that push effects (convenience and perceived security risk), pull effects (function experience, content experience, interactive experience, emotion experience), and mooring factor (relationship inertia) positively impact user stickiness. These findings are helpful for the online visual art platform to understand the influencing factors of user stickiness and adjust strategies from the perspective of user experience. The research contributes to the field of online visual art platform by proposing and verifying hypotheses on user stickiness through the PPM framework and constructing a structural equation model from the perspective of user experience.


I. INTRODUCTION
The emergence of online visual art platforms in large numbers reflects the rapid development of network technology. These platforms have become a mainstream way for users to learn, obtain information, and communicate online, resulting in a significant increase in user engagement. The value of these platforms is increasingly prominent and has emerged as a crucial topic of user research in the context of the internet.
In this study, the visual arts platform mainly refers to the sharing platform of digital art works (such as CG paintings and design works), not the art trading platform. Online visual art platforms are distinctive from other online platforms due to their strong visual appeal and unique user retention mechanisms. Consequently, user behavior on such platforms is characterized by a lower degree of interaction compared to other social or content platforms, as users primarily browse and seek inspiration from the artworks displayed. This may The associate editor coordinating the review of this manuscript and approving it for publication was Yilun Shang. result in relatively uniform user behavior. However, online visual art platforms typically have unique brands and styles, leading to high user stickiness, and frequent returns to the platform to browse and collect art, increasing user retention and visit duration. Additionally, sensory stimulation from visuals dominates the overall platform experience, and users rely heavily on the visual appeal of the artworks. Thus, the success of online visual art platforms is closely related to the display of artworks and user experience. Consequently, platform operators need to focus on enhancing the user experience and interface design, improving user satisfaction and loyalty.
Therefore, optimizing the multidimensional user experience of visual art platforms is crucial in the platform construction process. This study aims to investigate the impact of user experience on user stickiness of online visual art platforms. Previous research has shown that user experience is a critical factor in contemporary visual art market [1]. Moreover, it has been suggested that businesses should pay close attention to user experience in their development process to promote user satisfaction and platform sustainability [2]. However, academic research on online visual art platforms is still limited, with most studies focusing on interface interaction design and user experience evaluation. Consequently, research that specifically addresses user experience with relevance to user stickiness is still needed. Addressing the challenges of motivating users, ensuring platform sustainability, and promoting user stickiness remains a major concern for online visual art platforms. As such, this paper presents an empirical study to explore the relationship between user experience and user stickiness in the context of online visual art platforms.
In previous studies, user stickiness has been commonly defined as user loyalty, which refers to the likelihood of a customer to repeatedly engage with a store or platform. Originally, user stickiness was introduced in the field of e-commerce and has since been widely applied to other fields. Specifically, user stickiness is defined as the ability to attract and retain users' attention, which encompasses both the frequency and duration of website visits [3], [4]. It is regarded as a significant indicator of user loyalty to a website [5], [6], engagement in further interactions [7], and an entrenched commitment to the website [8].
In this study, we conceptualize user stickiness as a user's willingness to stay on a website, reuse its features, and extend their duration of stay, which is similar to the usability research approach [7], [9]. It is synonymous with consumer loyalty in marketing literature [10] and is a crucial predictor of the continuous use intention of IT systems [11], [12]. It is important to note that stickiness reflects user interest, availability, and potential profitability, which are not limited to usage intention or continuous intention. The former involves cognitive involvement and engagement, while the latter measures short-term or long-term use intention, of which stickiness can be a potential antecedent [4], [6].
The online platform's operational mode can be regarded as a new technological product or service from an academic perspective. Previous studies have adopted various theoretical models, including innovation diffusion theory, task-technology fit (TTF) theory, technology acceptance model (TAM), and the unified theory of acceptance and use of technology (UTAUT), to explore users' usage intention and behavior towards new technology products. TAM mainly focuses on investigating the correlation between users' attitudes towards technology and their perceived usefulness and ease of use, its ability to elucidate variations in the intention to adopt or actual utilization of technology, and its simplicity in terms of specification within structural equation modeling frameworks [13], [14]. This study primarily focuses on examining the influence of user experience on user stickiness across various dimensions, without directly addressing attitudes and intentions towards technology usage. For UTAUT, it comprehensively considers multiple factors that affect user acceptance, including personal, social, and technical aspects. The UTAUT model offers a more comprehensive explanation of the process through which user acceptance is formed. According to UTAUT, the four key constructs (performance expectations, effort expectations, social influence, and facilitating conditions) directly impact behavioral intention and subsequent behavior, while these constructs are influenced by variables such as gender, age, experience, and voluntary use [15]. Nonetheless, the UTAUT model is relatively intricate and involves numerous factors, necessitating a larger dataset and additional resources for validation and application. Consequently, the UTAUT model was deemed unsuitable for this study. Nevertheless, this study suggests a similarity in the underlying mechanisms between user stickiness and user transfer willingness. In this study, the push-pull-mooring (PPM) theory is used to examine user stickiness. PPM theory is the main paradigm of population migration research and explains the comprehensive influence of push and pull forces on migration [16]. As online user switching behavior and population migration share similar characteristics of moving from one place to another, PPM theory is more suitable for understanding user switching behavior [17]. The study argues that when users have the intention to continue using a platform, it implies the opposite intention to migrate and generates stickiness. Thus, the intention to stay and the intention to migrate are two opposite outcomes of the same user behavior, both influenced by push and pull factors, similar to user switching intention.
Compared to previous studies on user behavior of online platforms, this study has significant innovations in several aspects. Firstly, this research introduces the PPM framework to analyze user stickiness in the context of an online visual art platform, and explores the influencing factors of user stickiness from the perspective of push, pull, and mooring factors. The proposed hypotheses are verified, and the PPM framework has played a crucial role in this study. Secondly, previous studies mainly focused on investigating the adoption and continued usage of e-learning systems using TAM and UTAUT [18], [19], but ignored the impact of user experience. This study suggests that user experience, as a pull factor, plays a positive role in user stickiness. Specifically, the user experience dimensions of online art platforms, including content experience, function experience, interaction experience, and emotion experience, are considered in this study [20]. Thirdly, previous research on PPM theory mainly focused on the analysis of influencing factors of user switching intention, while this research innovatively applies PPM to the influencing factors of user stickiness (i.e., visit duration and user retention), which proves that PPM theory can well analyze the push, pull, and mooring factors generated by user stickiness. Since the research object is no longer user switching behavior, the mooring effect of switching cost is weakened. Therefore, this study considers relationship inertia as the mooring factor of user stickiness (visit duration and user retention), and convenience of use and risk perception as the pull factor. Perceived convenience and low risk of switching to other platforms when users use the current platform will positively promote their continued use of the current platform, resulting in user stickiness. This paper is structured into five main sections. The first section introduces the background of the research topic, providing the rationale and motivation for the study. The second section presents the relevant theoretical background and literature review used in this study, including the pushpull-mooring (PPM) theory and previous research on user stickiness. The third section outlines the proposed structural equation model based on PPM framework and the hypotheses between potential variables. The fourth section describes the methodology of the study, including the distribution of questionnaires and the analysis of data results using statistical software. The final section provides a discussion of the research findings, including the implications and limitations of the study, and proposes recommendations for future research in this area.

II. LITERATURE REVIEW AND HYPOTHESES A. PUSH-PULL-MOORING FRAMEWORK
The PPM framework is widely recognized as a prominent paradigm in migration research [14]. The framework's pushpull factors are based on Ravenstein's seminal work on the ''Laws of Migration,'' which serves as the theoretical basis for the push-pull model [21]. Originally, the theory of population migration was developed to understand human migration behavior [22], [23], wherein people move from one place to another for a certain period of time [24]. This migration can be either short-term or long-term, with short-term migration referring to individuals who temporarily work and live away from home and then return to their place of origin, while long-term migration means individuals leave their place of residence permanently and never return [25].
According to Bogue, migration is the result of the interaction between the push effect of the origin and the pull effect of the new destination [26]. The push effect refers to the negative factors that encourage people to leave their country of origin (such as economic opportunities, natural disasters, etc.), while pull factors attract people to move to the new destination, such as a high-quality education environment, political freedom, etc. [27]. Moon revised the push-pull model and incorporated the mooring effect, which refers to factors related to migration behavior that may either hinder or promote decision-making behavior [23]. The decision of immigrants to migrate from one area to another is affected by the push, pull, and mooring effects.
Previous studies utilizing the PPM framework should consider the research context to identify the push, pull and mooring factors in different environmental settings [28]. The PPM model is a practical tool that enables further empirical research on user behavior intentions. For instance, Balakrishnan et al. used the PPM model to elucidate the transition from e-learning to social media-based learning, and found that push factors (perception of e-learning), pull factors (convenience, social impact, academic reasons, ease of use and social network), and mooring factors (barriers) had varying degrees of influence on the switching intention [29]. Liao et al. employed the PPM model to investigate online learning transitions between different social network-based platforms and discovered that push factors (social interaction and service quality), pull factors (attraction of new services and social effect), and mooring factors (switching cost and previous switching experience) had different effects on the switching intention [30]. Similarly, Chen and Keng analyzed the intention to switch from traditional physical English learning to online learning platforms using the PPM framework [31]. Recently, some researchers have applied the PPM model to other types of user intentions, such as entrepreneurial willingness [32] and unfollowing intention [33]. In order to analyze user stickiness (visit duration and user retention) of online art platforms from a user experience perspective, the PPM framework is employed to investigate the relationship between various factors through the pull, push, and mooring factors.

B. APPLICATION OF PPM MODEL
Previous studies have predominantly utilized the PPM model to examine user switching intention, which arises when customers evaluate a product or service negatively, leading them to consider switching to an alternative [34]. The marketing literature has explored numerous factors that influence switching behavior [35].
In recent years, PPM models have been applied to analyze various factors that influence users' willingness, in addition to the study of users' switching intention. For instance, Ojiaku et al. used the PPM model to analyze the entrepreneurial intention of young graduates [32], while Tang and Chen applied it to examine the factors influencing brand microblog user unfollowing intention [33]. Furthermore, PPM models have been used to study users' attitudes towards the opening of parks [36]. Table 1 summarizes the key findings of previous studies that have used the PPM model. As user research continues to deepen, the PPM model is expected to be applied to the analysis of other factors that influence users' willingness.

C. USER STICKINESS
The widespread use of the Internet has led to its application in various fields. User stickiness, which includes the frequency and duration of website visits, refers to the ability of a website to attract and retain users' attention [3], [4]. According to Marchand and Davenport, stickiness implies the capacity of a website to create a favorable impression on users and encourage them to stay on the site [44]. Websites with high user stickiness tend to have users who frequently visit and spend more time on the website [45]. Even though users may switch to alternative sites, they prefer to revisit and use their favorite sites [8]. Lin defined stickiness as the willingness of users to revisit their favorite websites [6]. Lu and Lee described user VOLUME 11, 2023  stickiness as the amount of time users spend commenting, browsing, and staying within the community [46].
The concept of user stickiness refers to the loyalty and usage frequency of users towards a particular product or service, which is typically reflected in the length of time users spend accessing the product or service and their retention rate over a given period. The length of time users spend accessing a product or service is an indicator of their usage frequency and loyalty towards it. An increase in user access time may indicate an increase in their interest and loyalty towards the product or service, while a decrease in access time may indicate a loss of interest in the product or service or a greater interest in competing products. User retention refers to the continuation of product or service use by users over a given period, which is typically an indicator of user loyalty. An increase in user retention rate may indicate an increase in user loyalty and satisfaction with the product or service, while a decrease in retention rate may indicate a loss of interest in the product or service or a greater interest in competing products.
The duration of user visits and the retention rate are important indicators for evaluating user stickiness towards a product or service. A high duration of user visits and retention rate indicate that users have high loyalty and frequency of use towards the product or service, which indicates good user stickiness. Conversely, a low duration of user visits and retention rate indicate that users have low loyalty and frequency of use towards the product or service, which indicates poor user stickiness. These indicators can provide important insights for businesses to improve their user engagement and retention strategies. Previous studies have suggested that users who spend a significant amount of time on a website tend to revisit it [47]. Further research by Liu et al. has supported this claim, demonstrating that users who spend more time on blogs are likely to revisit them in the future [48]. Xu et al. have also found a positive relationship between users' visit duration and retention [45]. Therefore, based on these findings, this study puts forward the following hypothesis: H1. The time that a user spends on the platform will have a positive impact on the retention of users on the platform. 60766 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply.

D. EFFECT OF PUSH FACTORS ON PPM FRAMEWORK
Within the PPM (Push-Pull-Mooring) framework, the push factor pertains to negative influences that drive users away from current services [23]. With regard to the push factor, the switching effect in information systems typically involves negative perceptions of the service provider, such as service failures, employee issues, or pricing strategies, which ultimately lead to lower levels of satisfaction [49]. Previous research on push effects has examined different concepts [33], [40], [50]. However, this study considers the unique characteristics of user stickiness on online art platforms. Therefore, this study analyzes the impact of the push effect on user stickiness (measured by visit duration and user retention) based on the dimensions of convenience of use and perceived security risks.

1) PERCEIVED SECURITY RISK
Security risk refers to factors that create potential uncertainty for consumers in online transactions, which can negatively affect their intention to adopt e-services and reduce their willingness to continue using a service in the future [51], [52]. However, when a new service is perceived as more secure, users may switch to it [53]. While previous studies have defined security risk as consumers' belief in the potentially uncertain negative results of online transactions [51], Cheng et al. have classified security risk as a push factor, describing it as the perception of high-security risk that leads consumers to switch to an alternative with a lower security risk [54]. In this study, perceived security risk is defined as the unknown risk that may be associated with using other platforms, which may lead to consumer stickiness to the currently used online platform. Therefore, perceived security risk is considered one of the factors that influence the push effect in this study.

2) USING CONVENIENCE
Previous literature has defined convenience as the perceived time and effort required by consumers when using or purchasing services [55]. Online shopping provides consumers with greater convenience and saves time and search costs compared to traditional shopping behavior [56]. In the digital age, convenience refers to the ability to learn online through various devices at any time and in any location, resulting in more efficient time and cost savings [57]. Prior research has indicated that convenience level has a significant impact on user satisfaction and influences consumers' behavioral intention to select physical stores or online shopping [58], [59]. Previous studies also found that convenience is a significant push factor that affects online platform users' switching behavior [60], [61]. This study aims to investigate the factors that impact user's continuous use behavior, which is similar to user stickiness. Therefore, this study proposes that using convenience is an important push factor that affects user stickiness. In this study, convenience refers to the degree of convenience, and when the convenience of using this platform decreases, it will drive users to leave the current platform.
H2. Using convenience and perceived security risk of online visual art work platform will have a positive impact on the visit duration.
H3. Using convenience and perceived security risk of online visual art work platform will have a positive impact on the user retention.

E. EFFECT OF PULL FACTORS ON PPM FRAMEWORK
In the PPM framework, pull factors are positive elements that motivate people to shift from one service to another [22]. In the literature on marketing and information systems, the concept of pull mainly focuses on alternative attraction to explain the willingness of users to switch from offline to online consumer behavior [33]. Previous studies have suggested that when consumers are unable to fulfill their needs in physical stores, they seek alternative services to replace the original service pattern [39]. Therefore, this study examines the pull factors that influence user stickiness of online art platforms from the perspective of user experience and identifies four dimensions of user experience as pull factors.

1) USER EXPERIENCE
User experience, according to Li and Zhou, encompasses users' emotions, cognition, behavior, and other aspects before, during, and after using a product or system [62]. Initially, research on user experience was conducted in psychology and related fields. In the mid-1990s, Norman et al. introduced and popularized the concept of user experience, extending its applicability to all aspects of products and services [63]. By integrating multiple disciplines, such as engineering, marketing, and industrial design, they designed products that met users' needs, resulting in greater satisfaction and enjoyment. This study focuses on the user experience of an online visual art platform.
In recent years, there has been a growing interest in user experience, and scholars have expanded their research on this topic into various fields. User experience has become increasingly important in the field of enterprise marketing [64], [65]. According to Schmitt, user experience is defined as a direct observation and feeling that arises from an individual's participation in an activity, which stimulates the senses, thoughts, and emotions to induce personal perception of events [66]. When a person reaches a certain degree of emotion or feeling under the influence of an experience, a good feeling will be generated in his mind [67], [68]. Li et al. posit that user experience has a clear beginning and end, and includes all the feelings that users gain during these times [69]. The user experience from enterprises is considered as the intuitive experience of users when using software systems or hardware systems (interface design, voice conversion and text expression). This represents the psychological reaction that individuals get from their contact with the organization [70], [71]. The user experience in this study refers to the feeling in various dimensions when using the online art work platform. Adopts the results of Wang and Liu dividing the user experience by dimensions (function experience, content experience, interactive experience, and emotion experience) [20]. Therefore, we propose the following hypotheses: H4. The user experience of function, content, interactive, and emotion of online visual art work platform will have a positive impact on the visit duration.
H5. The user experience of function, content, interactive, and emotion of online visual art work platform will have a positive impact on the user retention.

F. EFFECT OF MOORING FACTOR ON PPM FRAMEWORK
Mooring factors can be defined as personal and social influences that prompt an individual to either leave or stay in their place of residence [23]. In the context of this study, mooring factors are defined as the factors that promote or hinder the generation of user stickiness. Based on previous research and the characteristics of the user experience on online art platforms, this study adopts relationship inertia as a mooring factor. According to Lin et al., relationship inertia affects mooring in online art platforms [60]. Therefore, this study defines relative inertia as the mooring factor.

1) RELATIONSHIP INERTIA
Relationship inertia is a concept that explains customers' preference for staying in their current transaction relationship instead of actively seeking out alternative channels. This tendency reduces the likelihood of service provider switching [72]. Previous studies have shown that both consumer inertia and satisfaction positively influence repeat purchase intention [73]. In tourism research, inertia has been found to play a crucial role in predicting the quality of tourist relationships and their intention to revisit [74]. In the context of information systems, the relationship between e-commerce and inertia has been found to be a key factor affecting customer selection, and conversion cost is an important variable affecting customer relationship management performance, satisfaction, and relationship inertia [75], [76]. Conversely, inertia has a negative impact on users' intention to switch to other mobile instant messaging providers [77]. In this study, relationship inertia is defined as the degree of inertia of the user using the current online art platform, and is taken as the mooring factor. Therefore, the following hypotheses are proposed: H6. Relationship inertia of online visual art work platform will have a positive impact on the visit duration.
H7. Relationship inertia of online visual art work platform will have a positive impact on the user retention.

III. RESEARCH METHODOLOGY A. RESEARCH FRAMEWORK
Following the PPM framework, this study proposes three key effect factors, namely push effects (perceived security risk and using convenience), pull effects (function experience, content experience, interactive experience, and emotion experience) and mooring effects (relationship inertia). These seven factors can be grouped into three dimensions, namely push, pull, and mooring dimensions. Figure 1 depicts our conceptual framework, including all the hypotheses formulated in this study.

B. SAMPLING AND DATA COLLECTION
Due to the lack of previous research on user stickiness of online visual art platforms, there is currently no established questionnaire to measure this concept. Additionally, different types of products have varying evaluation criteria for user experience. Most existing questionnaires for measuring user experience are designed for physical interactive products. To address this issue, we adapted the four dimensions of user experience from Wang and Liu's research on online art platforms (function experience, content experience, interactive experience, and emotion experience) as the evaluation criteria for our questionnaire [20]. The specific items included in the questionnaire were derived from previous studies [20], [78].
This study developed an initial questionnaire for online platform user stickiness by referring to questionnaires from renowned scholars and taking into account cultural differences and the real situation. The questionnaire was developed through several rounds of discussions with experts in design and psychology. Scholars specializing in structural equation modeling suggested that a larger-scale questionnaire provides better reliability and validity, as demonstrated by empirical research [79], [80]. Therefore, this study employed a seven-point Likert scale (ranging from 1= strongly disagree to 7= strongly agree) and a total of 37 items measuring nine latent variables.
The research team conducted a pre-test of the questionnaire by distributing it to potential participants and collecting 54 responses for analysis. Based on the results of the analysis, five items with low reliability and validity and low factor loading were removed, resulting in a final questionnaire with 32 items. The items and their sources can be found in Table 2: The survey was conducted between January 2023 and February 2023 using the online survey website SoJump in China. Prior to completing the questionnaire, a pre-question was used to determine whether participants were regular users of online visual art platforms. Participants who answered negatively were excluded from the study. The survey recorded the time taken to complete the questionnaire to ensure data quality. A total of 340 responses were collected from visual art platform users across various universities and companies in China, with five excluded due to never using online visual art platforms. Two screening methods were applied to ensure the quality of the sample [84]: first, participants completing the questionnaire within one minute were deemed to have filled it out irresponsibly and were excluded; second, reverse questions were included, and participants who answered inconsistently were also excluded. A total of 332 valid responses were used for formal data analysis.
This study considered the sample size required for the analysis. Previous studies, such as Loehlin, found that the median sample size of paper data was 198 based on 72 SEM papers [85]. Barrett suggested that the sample size should be at least 8 times greater than the number of variables [86]. However, when using the maximum likelihood method in SEM, a sample size greater than 500 can cause the chi-square value to expand excessively, leading to poor model fit. To address this, Zhang Weihao, an authoritative scholar in SEM, proposed that the sample size should be limited to 500. Therefore, the sample size of 332 collected in this study questionnaire is considered sufficient for the analysis.
A total of 332 valid questionnaires were collected, consisting of 188 males (56.63 percent) and 144 females (43.37 percent). The proportion of males was slightly higher but not significantly different from that of females.

C. PRELIMINARY INSPECTION 1) NON-RESPONSE BIAS
A total of 380 young and middle-aged users were invited to participate in the questionnaire survey, out of which 345 users responded. To check for non-response bias, we conducted t-tests to compare the demographic information of the responding and non-responding users in terms of gender, educational background, and platform use. The results showed no significant difference between the two groups (p > 0.05), indicating that non-response bias was not a serious concern [87]. Table 3 presents the demographic characteristics of the participants in this study.

2) COMMON METHOD BIAS
As all data in our study were self-reported in the survey, there may be common method bias due to consistency motives and social desirability [88]. To reduce common method bias to some extent, we ensured anonymity for the survey respondents [89]. This study tested common method variance analysis using Harman's single-factor test, following the suggestions of Delerue and Lejeune for the total variance explained by one common factor [90]. Unrotated exploratory factor analysis revealed a total of 9 factors with a characteristic root greater than 1, and the maximum factor variance interpretation rate was 32.16% (<40%). Thus, there is no single principal component that can explain most of the variation in this study.
In this study, the Variance Inflation Factor (VIF) was utilized to analyze the correlation between each construct, as suggested by prior literature [91]. The results in Table 4 show that the VIF values of all indicators are well below the acceptable threshold of 3.3, indicating that collinearity is not VOLUME 11, 2023  a major concern in this study. Therefore, based on these two testing methods, it can be concluded that common method bias is not a significant issue in this study.

IV. RESULTS
Structure equation modeling has two common methods, covariance-based SEM (CB-SEM) and partial least square SEM (PLS-SEM). This study adopts the CB-SEM due to two considerations.
CB-SEM is a confirmatory approach that requires a complete theoretical model, making it suitable for testing established theories [92]. On the other hand, PLS-SEM is suitable for early-stage theory development and testing, making it appropriate for exploring complex structural models [93]. Given the good theoretical foundation of the model in this study (PPM theory) and the determined direction of the relationship between variables (factors influencing user stickiness), CB-SEM was deemed more suitable. Regarding sample size, PLS-SEM can handle small samples without the assumption of distribution, while CB-SEM requires larger samples, but has lower variability in the case of large sample size. With a sample size over 300, CB-SEM was deemed appropriate for this study. The data analysis process included two stages: (1) measurement modeling to assess the reliability and validity of the measurement scale and (2) structural modeling to test hypotheses.

A. MEASUREMENT MODEL ASSESSMENT
The aim of the measurement modeling stage is to assess the measurement scales' reliability and validity. The reliability of the measures was evaluated using composite reliability (CR) and Cronbach's Alpha. The recommended values for CR and Cronbach's Alpha are 0.7 or higher. As presented in Table 5, the results for both CR and Cronbach's Alpha exceeded the recommended values, indicating that all constructs were reliable.
In this study, confirmatory factor analysis was employed to evaluate convergent and discriminant validity. Convergent validity measures the degree of correlation between the same latent variable, while discriminant validity measures the difference between different latent variables. Average variance extracted (AVE) and item loading significance were used to assess the convergent validity of the constructs. The recommended AVE value is 0.5 [94], and the minimum AVE value in Table 5 is 0.571, indicating satisfactory convergent validity. Furthermore, all item loadings were greater than 0.7, demonstrating satisfactory convergent validity [95].
The assessment of discriminant validity can be performed by examining the correlation matrix and the square root of the average variance extracted (AVE) for all constructs. Good discriminant validity requires that the square root of the AVE for each construct should exceed the correlation between that construct and all other constructs in the model. In Table 6, it is evident that the square root of the AVE value for each construct is greater than the correlation with any other factor, indicating that the measurement model exhibits good discriminant validity.

B. STRUCTURAL MODEL ASSESSMENT
The assessment of the structural model aims to examine the significance and strength of each hypothesis and the variables that explain the dependent variable. This step involves the empirical testing of hypotheses using AMOS software and structural equation model (SEM). Initially, the goodness-offit and overall explanatory power of the model are evaluated. Goodness-of-fit analysis refers to the degree of fit between the model and the observed data [96].
This study uses AMOS 24.0 to conduct confirmatory factor analysis on structural equation model (SEM). In order to ensure the consistency of the research data with hypotheses, the fit-of-goodness should meet the relevant requirements: CMIN/DF < 3.00 [97]; GFI and AGFI > 0.90, RMSEA < 0.08 (Reasonable fit) or < 0.05 (Favorable fit); NFI, IFI and CFI > 0.90 [95]; SRMR<0.08 [98]. As shown in Table 7: Figure 2 presents the results of the structural model. All the hypotheses have expected effects on switching intension at different significances, as seen in Table 8. Table 8 provides support for the validity of all hypotheses proposed in this study. The results indicate that visit duration (β = 0.199, p=0.001) has a significant positive impact on user retention. Push factor has a significant positive impact on both visit duration (β = 0.266, p<0.001) and user retention (β = 0.194, p=0.010). Additionally, push factor has a positive impact on user retention (β = 0.153, p=0.021), although this effect is not statistically significant. Furthermore, mooring factor has a positive impact on visit duration (β = 0.154, p=0.011), and a weak impact on user retention (β = 0.124, p=0.042). Therefore, based on these findings, we can conclude that hypotheses H1-H7 proposed in this study are valid.

A. DISCUSSION
In the context of intense competition and a highly segmented market, achieving sustainable development and retaining users has become a difficult challenge for online visual art platforms. To address this issue, this study investigates the factors that affect user stickiness on these platforms using  the PPM framework. Empirical findings indicate that the PPM theory can effectively explain the formation of user stickiness from the user experience perspective. Furthermore, all proposed hypotheses are supported by the research results. In the following section, we will discuss the implications of these findings. VOLUME 11, 2023 60771 Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply.

1) PUSH EFFECTS
The push effect in our study refers to the push factor that influences user stickiness to the online visual art platform. When users feel that other platforms or transfer processes have poor convenience and low security risks, it will create stickiness towards the current platform. These results are consistent with previous studies on switching intentions based on the PPM model [60], [61]. Hypothesis 2 and 3 have been validated, indicating that when users feel the convenience of use on the current platform, they will increase their access time, thereby increasing the platform's user retention; Similarly, if users experience inconvenience, they will switch to other platforms. If users feel high security risks (such as information leakage, copyright protection, etc.), they will leave the current platform and switch to other platforms. Therefore, it is essential to make users aware of the potential risks of leaving the current platform and improved convenience during their use of the online visual art platform.

2) PULL EFFECTS
The findings indicate that the pull effect is also a crucial factor in enhancing user stickiness on the online visual art platform. The pull effect is the outcome of the attraction of the destination's attributes and characteristics. This study's pull effect is based on the user experience perspective when using the online visual art platform. Hypothesis 4 and 5 have been validated, indicating that user experience, as a driving factor for user stickiness, attracts users to continue using the current platform and generates user stickiness. When the user experience improves, users will increase access time and retention rate, thereby improving the user stickiness of the platform. The four dimensions of function experience, content experience, interactive experience, and emotional experience are essential in evaluating user experience [20]. Consequently, the online visual art platform should enhance the user experience of the platform based on these four dimensions. It should refine the adjustment measures in terms of function, content, interaction, and emotion to entice users to increase their visit duration and user retention, thereby strengthening their stickiness.

3) MOORING EFFECTS
The study finds that mooring factors positively impact user stickiness, including visit duration and user retention, which aligns with prior research by Chen and Keng [31]. Hypothesis 6 and 7 have been validated, however, the effect size is not as significant compared to push and pull factors, indicating that relationship inertia is not the primary factor driving user stickiness. This could be due to users placing greater emphasis on platform services and content, rather than social attributes, which are not a central aspect of the online visual art platform. Moreover, contemporary users tend to be more rational and are less likely to let their previous habits impact their perception of platform quality.

4) USER STICKINESS
The research results indicate that the platform's visit duration has a positive impact on user retention. Hypothesis 1 has been validated. Visit duration and user retention are important factors that reflect user stickiness, and this study uses these two factors to reflect the platform's user stickiness. In the internal influence mechanism of user stickiness, visit duration positively promotes user retention, indicating that visit duration is a more direct and effective factor in improving user stickiness. Starting from improving visit duration, user activity and stickiness can be adjusted, while user retention more directly reflects the platform's level of user stickiness.

B. THEORETICAL SIGNIFICANCE
This study provides a comprehensive investigation into the factors that impact user stickiness on online visual art platforms, based on the existing literature and the unique features of such platforms. A PPM framework was constructed and empirically tested, revealing the significant effects of push, pull, and mooring factors on user stickiness. The innovative application of PPM theory in the context of online art platforms has yielded promising outcomes. Overall, this research has contributed to the understanding of user stickiness and 60772 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply. provided practical implications for enhancing user experience and retention in the online art industry. The following content is the theoretical significance and unique innovation points of this study. Previous studies have primarily applied the PPM model to analyze user switching intention. However, with the increasing focus on user research, the PPM model has been applied to analyze other factors that influence user behavior. For instance, it has been used to analyze the entrepreneurial intention of young graduates [32], the factors that influence user unfollowing intention on brand microblogs [33], and users' attitudes towards park opening [36]. In this study, we apply the PPM model to analyze the factors that influence user stickiness, which is similar to the concept of continuous use intention. This is one of originalities of the study.
Furthermore, prior studies that utilized the PPM model to examine online platform user switching intentions or other intentions have mainly employed alternative attraction, TAM, or Task-Technology Fit as a pull factor. In contrast, this study creatively adopts the dimensions of user experience (function experience, content experience, interactive experience, and emotion experience) as the pull factor. This study focuses on the impact of user experience on user stickiness, no longer focusing on the use of technology, but innovatively using dimensions related to user experience (content experience, functional experience, interactive experience, emotional experience) as pulling factors. Empirical analysis shows that the four dimensions of user experience have a significant pulling effect on user stickiness (access duration, user retention).

C. PRACTICAL SIGNIFICANCE
The online visual art platform managers are concerned about how to attract users to generate continuous use intention. The pull effect is a crucial factor that positively influences the user experience on the platform. This study emphasizes the importance of improving service quality in four dimensions of user experience, namely, function experience, content experience, interactive experience, and emotion experience. To enhance the platform's functional experience, the managers should improve the functional modules and core functions' quality. Moreover, they should optimize the art works' quality, classification, and content to improve the content experience. Therefore, managers should prioritize improving the multi-dimensional user experience to increase the user's visit duration and improve user retention on the visual art platform.
Regarding the push effect, the online visual art platform can enhance user stickiness by improving the convenience and security of the platform. For instance, enhancing the ease of use by optimizing website links, module flow, and main functions. Additionally, the platform can mitigate perceived security risks by improving the uniqueness of functions, thus increasing the platform's irreplaceable ability and reducing the potential risk of user churn. Therefore, online visual art platform managers should focus on optimizing convenience and security to enhance user stickiness.

VI. LIMITATIONS AND FUTURE RESEARCH
The present study has certain limitations. Firstly, it focuses mainly on undergraduate and graduate students in school, and excludes data from teenagers and middle-aged users. Therefore, the analysis of the factors influencing user stickiness may be limited. Future research should consider expanding the research scope to include a wider range of user groups to better achieve the research objectives.
Secondly, the study's online questionnaire primarily targeted Chinese users, which may limit the generalizability of the research findings to users from other countries. Furthermore, the initial distribution of the questionnaire relied on the author's social network, which may have led to a homogenous sample in terms of age, education, and demographics. To enhance the diversity of the sample, future research should expand the number of respondents from different countries, such as the United States and Europe. This will ensure that the research findings are more representative of the global population of users on online visual art platforms.
Thirdly, previous research has mainly applied the PPM framework to the analysis of user switching intentions, whereas this study focuses on the user stickiness of online visual art platforms. Although our approach is innovative, the theoretical foundation of the PPM theory in relation to user stickiness is still limited. Hence, the scope and applicability of the PPM theory in understanding user stickiness needs to be further verified in future research.

VII. CONCLUSION
This study investigated the impact mechanism of user experience on user stickiness on online visual art platforms through the push pull mooring theory. The results showed that the push pull mooring theory can effectively verify the relationship between various influencing factors, proving that both user retention intention and user transfer intention are influenced by push, pull, and mooring factors. All hypotheses have been validated, and it has also been proven that the user experience of each dimension plays an important role in the formation of user stickiness. This result has positive significance for analyzing user behavior and enhancing user stickiness on online visual art platforms, providing scientific and effective theoretical guidance for designers and managers of related platforms, and also providing reference for subsequent research in this field. He is currently the Director/an Associate Professor with the Department of Industrial Design, School of Art Design and Media, East China University of Science and Technology. He has presided over several courses, such as special research on major disciplines of category IV peak plateau discipline construction in Shanghai. His current research interests include user research and industrial design.