Antecedents and consequences of the perceived usefulness of smoking cessation online health communities

Purpose –An empirical study investigated the antecedents to perceived usefulness (PU) and its consequences in the context of smoking cessation online health communities (OHCs). Design/methodology/approach – To validate a research model for perceived informational support, perceived emotional support and perceived esteem support, the authors conducted a partial-least-squares analysis of empirical data from an online survey (N 5 173) of users of two smoking cessation OHCs. The proposed model articulates these as antecedents to PU from a social support perspective, and knowledge sharing and continuance intention are expressed as consequences of PU. Findings – The empirical study identified that the PU of smoking cessation OHCs is influenced by perceived emotional support and perceived esteem support, and perceived informational support indirectly affects PUvia these factors. In turn, PU exerts a positive influence on both knowledge sharing and continuance intention. Also, knowledge sharing positively affects continuance intention. Originality/value – The study contributes to scholarship on users’ postadoption behavior in the context of smoking cessation OHCs by disentangling the antecedents to PU from a social support perspective and pinpointing some important consequences of PU. The research also has practical implications for managing smoking cessation OHCs.


Introduction
Internet-based smoking cessation interventions that allow social interactions among smokers have ballooned in popularity in recent years. It is estimated that more than 12 million adult smokers in the USA sought related assistance via the Internet in 2017 alone (Graham and Amato, 2019). Recently, smoking cessation online health communities (OHCs) have received considerable attention from academics and practitioners too. Smoking cessation OHCs can be defined as social networks in which individuals can interact with each other with regard to kicking the smoking habit, seeking or offering related social support (Chen et al., 2019;Mpinganjira, 2018). Smoking cessation OHCs provide several benefits to smokers. Firstly, they offer smokers a communication channel through which they can interact with thousands of current smokers or ex-smokers without facing space and time restrictions. Also importantly, these OHCs allow users to remain anonymous by hiding their identity when online. This can help smokers maintain their privacy and avoid smoking-related stigma (e.g. blame, shame or negative stereotypes). Prior research suggests that participation in smoking cessation OHCs may lead at least to positive outcomes such as abstinence in the short term (Graham et al., 2015).
Quitting smoking is more like a marathon than a sprint. Even though some smokers may not have smoked for a while, they still need constant assessment and repeated interventions to prevent relapse. A longitudinal study spanning 25 years found that about 39% of former smokersthose who had quit smoking successfullyreported relapsing at least once during the smoking cessation process (Caraballo et al., 2014). For those who have quit in the recent past, the use of smoking cessation OHCs can help sustain the abstinence and aid in becoming permanently free of smoking (Cheung et al., 2015(Cheung et al., , 2020. In addition, they may also support other users through sharing tips and experiences of the smoking cessation journey (Dickerson et al., 2016;White et al., 2020). Obviously, smoking cessation OHCs can benefit both current and former smokers. Though the potential benefits for users are clear, a challenge remains: low participation levels (Saul et al., 2016). There are unanswered questions about how smokers can be motivated to keep using the OHCs as their smoking cessation process unfolds and about how to inspire them to contribute knowledge to the OHCs. Information systems (IS) scholars have posited that users' continuance intention toward an IS is critical for its success and sustainability (Bhattacherjee, 2001). In addition, knowledge sharing has been identified as important for the long-term success and sustainability of online communities (Chiu et al., 2006). Therefore, it is essential to investigate the factors influencing users' intention to continue using smoking cessation OHCs and their knowledge-sharing behavior in these communities.
Although research has paid a large amount of attention to either knowledge sharing in OHCs (e.g. Yan et al., 2016;Zhang et al., 2020;Zhang et al., 2017) or behavior related to their continuance intention (e.g. Song et al., 2018;Wu, 2018), little research has investigated the links between these distinct postadoption behaviors. Given the importance of bothusers' continuance intention toward smoking cessation OHCs and knowledge sharing in these OHCsfor the sustainability of smoking cessation OHCs, investigation of the link between the two should fruitfully advance understanding of the interdependence of these postadoption behaviors in smoking cessation OHCs.
Researchers have argued that perceived usefulness (PU) is a crucial motivator for continuance intention toward an IS (e.g. Bhattacherjee, 2001;Davis, 1989;Venkatesh and Davis, 2000) and an important driver of knowledge sharing in online communities (e.g. Hashim and Tan, 2018;Yuan et al., 2016). While their studies have provided important insights into the role of PU in postadoption behaviors, prior research has ignored the specific context of smoking cessation OHCs and the needs of users of these OHCs. Unlike diseases that rely predominantly on physical treatment, many issues or problems relevant to smoking cessation could be alleviated via behavioral interventions, such as counseling and social support. By affording such interventions, smoking cessation OHCs can be an important and effective part of improving abstinence , yet prior research has produced model of IS continuance, PU has been posited to be a dominant factor in predicting intentions to continue using an IS (Bhattacherjee, 2001). The association between PU and continuance intention has been validated in various contexts, such as e-government (Hamid et al., 2016), e-learning (Alraimi et al., 2015) and general OHCs (Wu, 2018). Also, a link has been found between PU and other postadoption behaviors. For instance, Li and Liu (2014) discovered that it influences the word-of-mouth behavior of e-service users, and the findings of Yuan et al. (2016) suggest that PU has a positive impact on knowledge sharing in online travel communities. A study by Hashim and Tan (2018) identified users' intention to share knowledge in online business communities as driven by the PU of the community.
Another stream of research focuses on investigating antecedents to PU of an IS from various perspectives. For instance, Agarwal and Karahanna (2000) found the PU of the World Wide Web to be influenced by the individual users and situational factors, such as the individual-specific traits of playfulness, personal innovativeness and user experience. In addition, Zhang et al. (2012) suggested that system characteristics affect the PU of computerbased communication systems. In the general OHCs context, Wu (2018) found that social support, information quality and service quality influence PU of the OHCs. Also, user perceptions of the hedonic and utilitarian aspects of an IS could affect the PU of that ISfor instance, curiosity, information quality and enjoyment affect PU of travel-review websites (Wang and Li, 2019).
Recent research has paid increasing attention to the PU of IS in the specific context of smoking cessation. For instance, Ali et al. (2019) found that PU of mobile health and quickresponse code technologies to be positively associated with smokers' use intention and actual use of both technologies. In research on digital educational games for students' smoking cessation, the PU of such games showed a positive association with the students' intention to quit smoking . However, all these studies focused on outcomes from PU and ignored the antecedents to it, at least with regard to smoking cessation OHCs. While scholars have investigated PU from multiple anglesamong them user characteristics (Agarwal and Karahanna, 2000), features of the technology (Zhang et al., 2012) and hedonic and utilitarian value (Wang and Li, 2019) and although prior research on smoking cessation OHCs has highlighted the importance of online social support for enhancing users' success in quitting smoking (Cheung et al., 2015;Graham et al., 2016), no empirical evidence has attempted to answer the question of whether social support can predict PU in the context of smoking cessation OHCs. This gap prompted us to examine the role of social support in predicting PU in the specific context of smoking cessation OHCs.

Social support theory
Social support refers to information and actions that lead an individual to believe that he or she is "cared for and loved, esteemed and valued" and "belongs to a network of communication and mutual obligation" (Cobb, 1976, p. 300). Prior literature suggests that social support affects human health and serves as a stress buffer (Cohen, 2004;Cohen and Wills, 1985). Social support has been found to be associated with positive outcomes in various health domains, such as alcohol withdrawal (Peirce et al., 2000) and smoking cessationspecifically, smokers are more likely to show improved smoking cessation performance when receiving active social support via strong social ties to partners, family members and close friends (Wagner et al., 2004;Westmaas et al., 2010). Likewise, social support expressed along weaker social ties, such as those in smoking cessation OHCs, has been suggested to lead to positive outcomes. For instance, Graham et al. (2015) stated that smokers who participate in smoking cessation OHCs might be more likely than nonmembers to stop smoking within three months.
Numerous studies have investigated social support in OHCs. These studies, summarized in Table 1, represent two major streams of research. One stream, examining how social support is exchanged in OHCs, employs various typologies of social support to categorize the support via content analysis, with one frequently used typology being the social support behavior code (SSBC) developed by Cutrona and Suhr (1992). According to the SSBC, there are five types of social support: (1) Informational support involves communicating facts or suggestions. Often, the informational support in OHCs includes messages about diseases, treatments and how one can cope with stress caused by illness. (2) Emotional support communicates love and caring. Such support usually produces a sense of being cared about by other users. In OHCs, sympathy and empathy shown through the communication are common examples of emotional support.
(3) Esteem support involves communicating confidence and respect for others' abilities. In OHCs, esteem support often is given when the messages conveyed state or imply that the reader is capable of and competent in dealing with a disease. Such support is generally intended to enhance users' self-confidence. (4) Network support encourages a sense of belonging to a social network of people with similar health concerns. Finally, (5) tangible support is the provision of goods or financial support needed in a stressful environment. In various contexts, such as OHCs related to HIV/AIDS (Coursaris and Liu, 2009), autism spectrum disorders (ASDs) (Mohd Roffeei et al., 2015) and smoking cessation (Zhang and Yang, 2015), informational support and emotional support have been found to be the two main types of social support exchanged, followed by esteem support and network support, while tangible support is quite uncommon. This might be because OHC users are generally dispersed geographically and stay anonymous online; only rarely do they provide material goods physically or directly supply financial support via the OHC (Huang et al., 2014(Huang et al., , 2019. The other research stream focuses on the role of social support in OHCs. For instance, Wang et al. (2017) investigated which types of social support affect users' participation and found that informational support, seeking emotional support and companionship are three important determinants of users' continued participation in breast-cancer OHCs. The findings of Chen et al. (2019) indicate that the exchange of social support is determined by the structural capital developed in OHCs. Also, they found that social support has a positive influence on users' health literacy and health-attitude valence. The work of Huang et al. (2019), in turn, identified that structural capital, cognitive capital and relational capital all facilitate the provision of emotional support, whereas only cognitive capital promotes the provision of informational support.
The literature shows that social support theory may be amenable to explaining users' perceptions of the usefulness of smoking cessation OHCs from the individual perspective. Firstly, the literature on social support points to a positive correlation between that support and health. This may partly explain the positive impact of social support from OHCs on the success of one's smoking cessation efforts. Secondly, social support theory is useful in identifying the types of social support in smoking cessation OHCs and examining their roles in smoking cessation OHCs. While these OHCs are collectives of users with a common goal of quitting smoking, users differ in the types of social support they need for coping with the stresses and uncertainties related to reaching that goal. Since social support from smoking cessation OHCs might increase user perceptions of the usefulness of smoking cessation OHCs, thereby further affording abstinence, social support theory represents a suitable theoretical framework for examining the determinants of PU with regard to smoking cessation OHCs. Informed by findings from prior research on several types of social support in OHCs, our study focused on three important types of social support identified in the literature: informational, emotional and esteem support. We excluded tangible support because of its rarity in OHCs (Huang et al., 2019;Wang et al., 2017). Also, we omitted network support from consideration in our study because it has been argued to be distinct from social support and functions differently in OHCs (Huang et al., 2019). Since social support is a Perceived usefulness of smoking cessation OHC mechanism for reducing uncertainty and stress, the analysis could have been unnecessarily complicated by the inclusion of network support, which scholars regard as the shared activities for their own sake rather than for buffering against a stressful situation (Albrecht and Adelman, 1987;Huang et al., 2019;Rook, 1987;Thoits, 1986).
3. The research model and hypotheses 3.1 The proposed model Proceeding from the literature on postadoption behaviors (Bhattacherjee, 2001;Venkatesh and Davis, 2000;Yuan et al., 2016), we expected to find PU to be an important factor in predicting both continuance intention and knowledge sharing in smoking cessation OHCs and to find a link between these two postadoption behaviors. Furthermore, research examining social support in OHCs led us to posit that three particular types of perceived social support (perceived informational support, perceived emotional support and perceived esteem support) are central antecedents to the PU of smoking cessation OHCs. In addition, we hypothesized that perceived informational support influences perceived emotional and perceived esteem support in these OHCs. Age, gender, country and smoking cessation stage were tested as possible moderators. Table 2 presents the definitions of the constructs in the proposed research model. Figure 1 summarizes the model itself.

Hypotheses
Emotional supportthat is, communicating encouragement, concern, understanding, sympathy and even love to others (Cutrona and Suhr, 1992)can help individuals restore

Construct Definition
Continuance intention (CI) Willingness to continue using the smoking cessation OHC (Bhattacherjee, 2001) Perceived emotional support (PEMS) Users' perceptions of the care, empathy, encouragement and even love received in the smoking cessation OHC (Cutrona and Suhr, 1992) Perceived esteem support (PESS) Users' perceptions surrounding respect and confidence gained in their abilities via the smoking cessation OHC (Cutrona and Suhr, 1992) Perceived informational support (PIS) User perceptions connected with the information on smoking cessation received in the smoking cessation OHC, such as advice, facts and referrals (Cutrona and Suhr, 1992) Knowledge sharing (KS) The behavior of exchanging information, experience and skills related to smoking cessation in the smoking cessation OHC (Hsu et al., 2007) Perceived usefulness (PU) The degree to which a user believes that using the smoking cessation OHC will enhance his or her success in ceasing to smoke (Davis, 1989) Table 2.
Constructs in the research model Figure 1. The proposed research model their emotional stability by reducing such signs of emotional distress as anxiety and sorrow (Huang et al., 2019). Smokers who are trying to kick the habit often feel disappointed over their relapses and become anxious about the repeated failure. Smoking cessation OHCs offer a friendly environment in which smokers can disclose their negative feelings and ask for emotional support from people who have experienced similar situations (Huang et al., 2019;Zhang and Yang, 2015). Members of a smoking cessation OHC can receive empathy from peers who truly understand their negative emotions related to the smoking cessation process. In addition, users can obtain encouragement from other users that bolster their confidence in achieving abstinence. Moreover, the anonymity and privacy protections developed for smoking cessation OHCs allow freely sharing personal emotions without many security or privacy risks. Emotional support from these OHCs may assist in users' efforts to reduce the stress they face on the smoking cessation journey and to restore their emotional stability (Granado-Font et al., 2018;Rocheleau et al., 2015;Zhang and Yang, 2015). A positive correlation between emotional support and smoking cessation success has been reported in the context of telephone-based interventions (Burns et al., 2014), providing further reason to expect perceived emotional support from smoking cessation OHCs to help smokers regain emotional stability, which may lead to greater success in kicking the habit. Accordingly, the more emotional support users can obtain from the smoking cessation OHC, the more useful we would expect them to find the OHC. We formed the following hypothesis: H1. Perceived emotional support is positively associated with the PU of a smoking cessation OHC.
Esteem support provides compliments and releases from blame (Cutrona and Suhr, 1992). This support can help smokers elevate their belief in themselves and their abilities with regard to quitting smoking Huang et al., 2014). Specifically, users of smoking cessation OHCs often receive congratulations and positive feedback when sharing their achievements (e.g. a month of being tobacco-free). This can help them cultivate a positive self-image and believe in their ability to quit smoking and in their capabilities for doing so. Furthermore, peers' expressions of forgiveness might alleviate users' feelings of guilt associated with relapse and motivate them to move past failures without blaming themselves unfairly. Studies have identified compliments as one type of partner support that promotes success in smoking cessation in offline settings (Cohen and Lichtenstein, 1990). As for online contexts in general, the literature suggests that esteem support is a social factor that supports health-related behavior changes, such as increased physical activity (Cavallo et al., 2014). Therefore, it is reasonable to expect perceived esteem support from a smoking cessation OHC to have positive effects on users' perceptions of the usefulness of the OHC. The more esteem support one receives from the smoking cessation OHC, the more useful that OHC is perceived to be. We developed the following hypothesis accordingly: H2. Perceived esteem support is positively associated with the PU of a smoking cessation OHC.
Scholars have identified informational support as another major type of social support in smoking cessation OHCs (Granado-Font et al., 2018;Rocheleau et al., 2015). This form of support provides users with information on problem-solving (Huang et al., 2019). Users of a smoking cessation OHC may receive information on the benefits of quitting and the negative consequences of continuing to smoke (Granado-Font et al., 2018;Rocheleau et al., 2015;Zhang and Yang, 2015). This might assist smokers in developing firmer intentions to stop smoking and get ready for truly quitting (World Health Organization, 2014). Additionally, users can get suggestions, such as tips on coping with cravings and withdrawal symptoms, and read personal success stories addressing how to quit (Granado-Font et al., 2018;Rocheleau et al., 2015;Zhang and Yang, 2015). With this support, smokers may gain skills for quitting and Perceived usefulness of smoking cessation OHC better optimize their quitting strategies and plans. In addition, unlike general guidelines produced by professionals, the informational support in smoking cessation OHCs is largely based on real-world experiences so may better match individual smokers' practical needs. Informational support from smoking cessation OHCs may help users prepare, plan and act to stop their tobacco use. Therefore, it is reasonable to expect that the perceived informational support from the OHC will lead them to perceive the OHCs as useful. The more informational support one can receive from it, the more useful it is perceived to be. Accordingly, we hypothesized thus: H3a. Perceived informational support is positively associated with the PU of a smoking cessation OHC.
Psychology literature shows that the information individuals have received can affect their emotions (Joseph et al., 2020;Westermann et al., 1996). For instance, Zupan and Babbage (2017) found that reading information (e.g. narratives or stories) can elicit emotions such as sadness, anger and happiness. Familiar events and situations depicted in written stories lead readers to sympathize with the characters, thereby evoking emotional responses (Oatley, 1999;Zupan and Babbage, 2017). Much of the informational support received in smoking cessation OHCs takes this formnot only smoking cessation tips, advice and facts, but also personal experience and stories (Cheung et al., 2017), which may trigger emotional reactions and help cultivate experiences of emotional support (Derks et al., 2008;Verheyen and Goritz, 2009). For instance, personal stories about quitting posted by other members of the OHC may remind users that they are not alone in their struggle and encourage them to feel a sense of companionship. Meanwhile, others' achievements and victories might also support rebuilding a user's confidence in continuing the fight against nicotine addiction, having the effects of esteem support in smoking cessation OHCs. At the same time, those who benefit from such informational support may give supportive feedback to its providers, expressing congratulations and thanks in return. These factors led us to expect perceived informational support to have an influence on perceived emotional and esteem support, so we proposed the following hypotheses: H3b. Perceived informational support is positively associated with perceived emotional support in a smoking cessation OHC.
H3c. Perceived informational support is positively associated with perceived esteem support in a smoking cessation OHC.
In the literature, some have argued that PU is the primary determinant of knowledgesharing behavior in online communities. For instance, Yuan et al. (2016) found that PU affected it in the context of online travel-oriented communities. In addition, some work has found that PU is the predominant driver of users' intention to continue using the given IS (Bhattacherjee, 2001). Hence, we expected to find that PU affects both continuance intention and knowledge-sharing behavior in smoking cessation OHCs, and we proposed the following two hypotheses: H4. PU is positively associated with users' knowledge sharing related to smoking cessation OHCs.
H5. PU is positively associated with users' intention to continue using the smoking cessation OHC.
Prior literature highlights contributors to different post-IS-adoption behaviors, among them continuance intention and knowledge sharing, but minimal attention has been paid to associations between these distinct postadoption behaviors. Li and Liu (2014) have suggested that there is value in investigating these relationships for purposes of examining their possible interdependence. Studying online auction communities, Wang and Chiang (2009) found that users who are more engaged in online communities (e.g. asking and/or answering questions) are more likely to continue using them. Therefore, it is reasonable to argue that the more knowledge users of smoking cessation OHCs share (whether sharing tips/advice/ experience or asking/answering questions), the more likely they will be to keep using the OHCs. We developed this hypothesis: H6. Users' knowledge sharing is positively associated with their intention to continue using the smoking cessation OHC.
Finally, considering possible effects of user characteristics such as age and gender as moderators has been recommended for anyone testing whether social support affects smoking cessation outcomes (Westmaas et al., 2010). Furthermore, research indicates that an additional factor, the individual's stage on the smoking cessation journey, has an association with social support in OHC contexts (Ploderer et al., 2013). Since we collected our empirical data in two countries with different cultural backgrounds, we considered the country as another possible moderator. We hypothesized that country, age, gender and smoking cessation stage moderate the proposed relationships in our model.

Research method 4.1 Development of the measurement technique
To guarantee the reliability and validity of the measurements for each construct in the proposed model, we employed previously validated instruments. The items for each construct were reworded for the context of smoking cessation OHCs. A five-point scale, ranging from "1 5 strongly disagree" to "5 5 strongly agree," was used to measure all the construct items in the study. The source items for perceived informational support and perceived emotional support were informed by the research of Liang et al. (2011), the measurement items for perceived esteem support were adapted from work by Oh et al. (2013), PU and continuance intention were measured with items adapted from Bhattacherjee (2001) and items for knowledge sharing were adapted from the work of Hsu et al. (2007). The Appendix presents details of the construct items.

The data-collection process
Two nonprofit smoking cessation OHCs, one in Finland (Stumppi.fi) and the other in China (a smoking cessation bar on Baidu Post Bar), were selected for this research. Even though the smoking cessation OHCs operated in very different countries and were hosted by separate organizations, they showed some similarities in platform structure and functions. Both OHCs provided users with basic functions, such as starting a new discussion, submitting questions to seek help, commenting or replying in a discussion thread and sending private messages, and both are easy to use. We employed our online survey to collect the data after having received ethics approval from the corresponding author's home university. The survey questionnaire for collecting empirical data was developed in English and then translated into Finnish and Chinese. Then, IS researchers who are native speakers of the respective languages reviewed the questionnaire in all three variants to verify the validity of the content and its translation. After this, we conducted a pilot study with 20 users of Stumppi.fi to test the questionnaire in Finnish. We modified the questionnaires in all three languages further on the basis of their feedback.
The full-scale online survey was launched on November 23, 2018 in China and December 17, 2018 in Finland. We recruited participants by making the questionnaire available via the two target smoking cessation OHCs. In all, 235 users had responded by April 30, 2019 (48 in Perceived usefulness of smoking cessation OHC Finland and 187 in China). Each respondent received an incentive for participating in the online survey.
After exclusion of replies that indicated an unwillingness to participate in the survey (2 in Finland, 48 in China) and unreliable replies, such as ones with the same answer option marked for all measurement items (12 in China), we had 173 forms as a valid sample for data analysis.
All respondents were smokers at different stages in smoking cessation. As for the sample's demographic breakdown, most respondents were between 25 and 44 years old (67.6%), and 37.0% were female, 59.5% were male and 3.5% concealed their gender. Table 3 presents all respondents' demographic information and smoking cessation stage.

Measurement invariance, common-method variance and collinearity
Because the data were collected from different smoking cessation OHCs, in two countries, we conducted an invariance test to check whether the construct measurements were understood similarly by the two samples, following the measurement invariance of composite models (MICOM) assessment procedure proposed by Henseler et al. (2016b). The results of permutation testing show that all c values, the difference in mean values and the variance of composites between the two countries fall between the upper and lower bound for the 95% confidence interval, as recommended by Henseler et al. (2016b). Thus, the testing established that we achieved measurement invariance, indicating that we could safely pool the data from the two sources and proceed with the analysis.
We used Harman's single-factor test (Podsakoff et al., 2003) to check for common-method variance (CMV). The highest total variance for any factor is 45.8%, which is below the recommended maximum of 50% (Podsakoff et al., 2003), thereby indicating minimal concern about CMV. Further, we measured collinearity via partial least squares (PLS), following the suggestion of Kock and Lynn (2012). All variance inflation factors from the full collinearity test are below the recommended upper limit of 3.3 (Kock and Lynn, 2012), so the research model is free of collinearity.

Data analysis
We used the PLS implementation of SmartPLS 3.0 to test both the measurement model (this involved assessment of convergent validity and discriminant validity) and the structure model. To test convergent validity (Chin, 1998;Hulland, 1999;Tenenhaus et al., 2005) Table 3.
Respondents' demographic data and stage in the smoking cessation process the factor loading for each item, composite reliability (CR) and average variance extracted (AVE) for each construct. We removed two items, PEMS3 and PIS1, because their loadings were lower than the recommended minimum. As Table 4 shows, each item's factor loading exceeded 0.70, and the AVE and CR figures met the recommended criteria: the threshold values are 0.5 and 0.7 (Chin, 1998;Fornell and Larcker, 1981), respectively. This indicates adequate convergent validity.
To evaluate discriminant validity, we calculated the square root of the AVE for all constructs in the research model. We then conducted a comparison between the loading of each item for an associated construct and its cross-loadings on other constructs. For each construct, the value of the square root of the AVE is higher than the correlation with other constructs (See Table 5). As shown in Table 6, the factor loading of each construct item for the relevant construct is higher than the cross-loadings on the other constructs. Thus, the discriminant validity of all constructs in our proposed model is supported (Chin, 1998 (Geisser, 1974;Hair et al., 2017;Stone, 1974), which was measured by means of SmartPLS 3.0's blindfolding technique. The Q 2 for knowledge sharing was 0.255, the Q 2 for continuance intention was 0.221 and the Q 2 for PU was 0.279, indicating good predictive relevance.
Finally, we tested the goodness of fit by measuring the standardized root mean square residual or SRMR (Henseler et al., 2016a). The result was 0.064, which is lower than the maximum acceptable value of 0.08 proposed by Hu and Bentler (1999). Our model showed a good fit.
We applied the bootstrapping procedure of SmartPLS 3.0 to test the structural model, including the path significance and the hypotheses' effects. The overall explanatory power and estimated path coefficients are presented in Figure 2. As postulated, perceived emotional support (β 5 0.267, p < 0.05) and perceived esteem support (β 5 0.367, p < 0.001) had a significant positive correlation with PU. We did not find a significant association between  perceived informational support and PU, but the former was significantly correlated with perceived emotional support (β 5 0.743, p < 0.001) and perceived esteem support (β 5 0.766, p < 0.001). PU showed a significant positive correlation both with knowledge sharing (β 5 0.572, p < 0.001) and with continuance intention (β 5 0.405, p < 0.001). Finally, knowledge sharing showed a significant association with continuance intention (β 5 0.238, p < 0.05). Therefore, H1, H2, H3b, H3c, H4, H5 and H6 are supported. Our model explains 48.6% of the variation in the PU of smoking cessation OHCs, 32.7% of knowledge sharing, 33.1% of continuance intention, 55.2% of perceived emotional support and 58.7% of perceived esteem support.

Moderation analysis
To test for moderating effects of age, gender, country and smoking cessation stage, we performed multigroup analysis (MGA) to investigate whether the paths' strengths differ with the user group, after evaluating the measurement invariance by means of the aforementioned MICOM procedure.
Since most respondents were aged 25-44 (N 5 117) and the numbers in other age bands were relatively small, we balanced the samples in size by dividing the respondents into two groups: group A includes those aged 25-44, and group B includes all those under 25 or over 44. Regarding the smoking cessation stage, we divided the sample into three groups in line with the six stages on the journey described by Prochaska and Velicer (1997), (1) before-action users, encompassing all those intending to quit but not having acted on this intention yet and covering the contemplation and preparation stages; (2) in-action users, for those who had entered a stage of action; and (3) after-action users, covering those who had not smoked for at least six monthsindividuals in the maintenance or temptation stage. Note that the four responses from those in the precontemplation stage, without an intention to quit smoking, were excluded from the analysis.
As Table 7 illustrates, we verified full measurement invariance between Finnish and Chinese respondents and also with regard to different age groups. For the gender and smoking cessation stage, partial invariance was identified. Therefore, performing MGA can be considered acceptable in this case (Henseler et al., 2016b).
No significant difference was found between Finnish and Chinese users (see Table 8). A significant difference did appear between the two age classes, however, with specific regard to the relationship between knowledge sharing and continuance intention (see Table 9). Also, as Table 10, on gender, indicates, we found a significant difference between male and female users for the connection between perceived emotional support and PU: perceived emotional support was a significant driver of PU for female users (β 5 0.599, p < 0.001) but not for male users.
As Table 11 shows, we found a connection between perceived emotional support and PU for before-action users, and perceived esteem support was linked to PU for after-action users. Also, PU showed a significant correlation with knowledge sharing no matter the user's stage in the smoking cessation process. For before-action and in-action respondents alike, PU was significantly correlated with continuance intention, but knowledge sharing displayed a significant connection to continuance intention among only those users in the afteraction group.

Discussion
Our findings on the antecedents to the PU of smoking cessation OHCs and on its consequences raise several points that are of interest.
Firstly, perceived emotional support emerged as a determinant of the PU, particularly for users who want to quit smoking but have taken no action thus far. One possible explanation n.s 0.268* n.s Note(s): ***, p < 0.001; **, p < 0.01; *, p < 0.05; n.s., not significant Table 7. Measurement invariance testing via MICOM Table 8. Results from testing country as a moderator is that those users are still hesitant to initiate actions in this regard and, hence, need encouragement or expressions of care from others, to dispel any misgivings about smoking cessation and to enhance their confidence in such actions. Also noteworthy is the gender difference we uncovered in the relationship between perceived emotional support and PU. A possible explanation for it being an antecedent to PU among female but not male users is that women value emotional support more than males do, particularly in stressful situations (Matud, 2004;Tamres et al., 2002). Female users of smoking cessation OHCs might be likely to view emotional support as crucial for reducing smoking-related stress than male users. Accordingly, when experiencing high levels of emotional support via a smoking cessation n.s n.s p < 0.05 0.737*** 0.645*** 0.392*** H5 n.s p < 0.05 p < 0.001 0.718*** 0.692*** n.s H6 n.s p < 0.05 n.s n.s n.s 0.404*** Note(s): ***, p < 0.001; **, p < 0.01; *, p < 0.05; n.s., not significant Table 9. Results from testing age as a moderator Table 10. Results from testing gender as a moderator Table 11. Results from testing the user's stage in smoking cessation as a moderator Perceived usefulness of smoking cessation OHC OHC, female users are likely to perceive the smoking cessation OHC as useful for improving the outcome of their smoking cessation efforts.
Secondly, our findings suggest that perceived esteem support is another important antecedent to the PU of smoking cessation OHCs, especially for users who have been able to avoid smoking for at least six months. It may be that these users in particular are likely to blame themselves for repeated lapses, which can occur easily. Esteem support from others can reduce their self-blame, and positive feedback recognizing what they have achieved on the journey of quitting smoking may well enhance their self-confidence. Our findings in relation to this contribute significantly to the literature. Little prior research has examined the association between perceived esteem support and the PU of OHCs. This might be because most scholars focus on emotional support when considering the social support associated with users' personal emotions in OHCs and overlook the importance of esteem support.
Contrary to our hypothesis, perceived informational support did not emerge as an antecedent to the PU of the OHCs in our research context. This finding is at odds with research by Wu (2018), who identified informational support as an important driver of the PU of general OHCs. This discrepancy might arise from the indirect influences that perceived informational support exerts on PU, via the effects of both perceived emotional support and perceived esteem support on the PU of the smoking cessation OHCs. Indeed, the post hoc mediation analysis showed that the impact of perceived informational support on PU was fully mediated by perceived emotional support (direct effect β 5 0.127, p > 0.05; indirect effect β 5 0.199, p < 0.05) and by perceived esteem support (direct effect β 5 0.127, p > 0.05; indirect effect β 5 0.278, p < 0.001). The informational support experienced from smoking cessation OHCs may not always be adequate to enhance users' perception of the OHC's usefulness, since they can obtain similar information from alternative sources, such as online self-help materials or advice from doctors. In addition, informational support in smoking cessation OHCs might feed into users' perceptions of emotional and esteem support instead. For instance, reading other users' stories in a smoking cessation OHC may help to produce a sense of stress release in the course of smoking cessation and a sense of being cared for by other members of the smoking cessation OHC. Meanwhile, the stories of success shared by other users could also shore up their confidence in quitting smoking. Through these mechanisms, perceived informational support has an indirect influence on the PU of smoking cessation OHCs via perceived emotional support and perceived esteem support.
Fourthly, we found significant impacts of PU on both continuance intention and knowledge sharing. Our findings on the former relationship are consistent with those from research in the domains of online banking (Bhattacherjee, 2001), e-learning (Alraimi et al., 2015), e-government (Hamid et al., 2016) and general OHCs (Wu, 2018): The PU strongly influences users' intention to continue using the smoking cessation OHC. Our findings on the association between PU and knowledge sharing are in line with prior research too, such as the work of Yuan et al. (2016). Looking at the context of online travel communities, they found that the PU positively influences knowledge sharing in said communities. In smoking cessation OHCs, when users perceive the OHC to be useful for faring better with the smoking cessation process, they are more likely to continue using the OHC and contribute their knowledge to the OHC.
Another important finding is that knowledge sharing exerts positive effects on continuance intention with regard to smoking cessation OHCs. Our results demonstrate that the more knowledge users share with the smoking cessation OHC, the stronger their intention to continue using it. This might be explained by the fact that users who have devoted time and effort to sharing knowledge in a smoking cessation OHC tend to form a bond with it. Such connections may render them more likely to continue their use. Interestingly, an age difference was evident in the relationship between knowledge sharing and continuance intention. The two showed a correlation among users aged 25-44 but not among those of other ages. There are several possible explanations. One is that those aged 25-44, who constitute the majority of users of smoking cessation OHCs, form a useencouraging psychological bond with the OHC via their knowledge-sharing activities, a bond perhaps fortified by attention to age-specific concerns. The other users, on the other hand, might not share much knowledge in these forums, with their need for social support being the main factor in their wish to keep using them.
Finally, we should discuss the differences connected with users' stage in quitting smoking with regard to the impacts of PU on continuance intention and knowledge sharing. Specifically, PU exerts stronger significant effects on the continuance intention of beforeaction users than on that of in-action users but no significant effects for after-action users in this regard, and significant differences are visible both between the in-and after-action group and between the before-and after-action group. These significant differences might be due to the following factors: Before-action users may be new to the smoking cessation OHC, turning to it as they seek further guidance and social support to prepare themselves and to translate their quitting intentions into specific actions. When before-action users view the smoking cessation OHC as useful, they have much stronger intentions to continue using it than in-action users; they know that they will need even more social support from it when they enter the action stage. Finally, though after-action users perceive smoking cessation OHCs as useful in supporting their long-term goals, they have already reached the goal of abstinence, so PU might not be a large factor in their continuing use of the smoking cessation OHC.
We also found significant stage-related differences when examining perceived informational support's effects on perceived emotional support and perceived esteem support. Specifically, it has its strongest influence on the former among before-action users (β 5 0.845, p < 0.001), followed by in-action users (β 5 0.802, p < 0.001) and then after-action users (β 5 0.623, p < 0.01); a significant difference exists between before-and after-action users. Obviously, after-action users have been members for some time and achieved abstinence. Though they can still get informational support from the OHC, they are familiar with the guidance and tips most commonly provided, so the influence of perceived informational support on perceived emotional support is weaker for them than for the other users. Before-action users, in contrast, are in more need of informational support, to help them decide to move on to actions, than are in-action users. The informational support they receive from smoking cessation OHCs gives them a stronger sense of empathy, encouragement and so on. from the OHC's other users. Therefore, it may be little surprise that perceived informational support has bigger impacts on perceived emotional support for before-action users relative to in-action users, who have already received considerable informational support that assists them in quitting smoking. Hence, the information they keep receiving might not confer as much emotional support as before-action userswho are in great need of informational supportexperience.
As for the impact of perceived informational support on perceived esteem support, the influence is strongest for before-action users (β 5 0.887, p < 0.001), followed by after-action users (β 5 0.705, p < 0.001) and then in-action users (β 5 0.660, p < 0.001). We found significant differences between before-and in-action users and between before-and afteraction users both. One possible explanation for these is that users at different stages differ in the social support they need, for overcoming different challenges. Those who intend to quit but have not yet acted need guidance and advice, to build their confidence for doing so; therefore, perceived informational support influences these users most strongly, with regard to perceived esteem support. Those who have already quit smoking have experienced difficulties in keeping this up and need information that maintains their confidence in their ability to stay smoking-free. Finally, perceived informational support affects perceived esteem support least for members of the in-action group, who have established their confidence in quitting smoking but have not experienced all the hardships of the process. Perceived usefulness of smoking cessation OHC Stage-related differences emerged in relation to knowledge sharing too, with PU having the strongest connection with it among before-action users (β 5 0.737, p < 0.001), then in-action users (β 5 0.645, p < 0.001) and after-action users (β 5 0.392, p < 0.001). We identified a significant difference between the before-and the after-action group. The users who had already abstained from smoking for at least six months perceived smoking cessation OHCs as useful for their success in quitting smoking, but they were less likely than other users to contribute knowledge to the OHC. One possible reason is that successful quitters use such OHCs less often than others do.
When examining the relationship between the two postadoption behaviors, we found a positive correlation between knowledge sharing and continuance intention only for afteraction users, with a significant difference visible between the in-and the after-action group. It might be that users who have already abstained for a while have solid practical tips and advice for other users, based on their experience, and also share more knowledge in the smoking cessation OHC, to demonstrate reciprocity and support the users and OHC that supported them on the path to quitting smoking. When they share more knowledge in the smoking cessation OHC, they feel a strong bond with the group and a sense of solidarity, so they are more likely to keep using the OHC.
6. Conclusions 6.1 Theoretical contributions The research findings have several implications for scholarship. Firstly, we extended the application of social support theory to explain IS postadoption behaviors in the context of smoking cessation OHCs from the view of social support in explaining PU of smoking cessation OHCs. Our consideration for the roles of distinct components of social support in explaining IS postadoption behaviors enriches IS research from the view of the social support of smoking cessation OHCs in addressing societies' public-health and well-being issues.
Secondly, whereas prior literature on external factors affecting PU has almost exclusively taken a technological perspective (Zhang et al., 2012), our work enriches PU literature by employing social support theory to investigate the antecedents to the PU of smoking cessation OHCs. Our findings on perceived emotional and esteem support as two important determinants of PU in this context suggest that PU can be explained well from the perspective of social support, which is especially relevant in a smoking cessation OHC context. By looking at users' perceptions of emotional support and esteem support as factors in explaining PU in this specific context, we advance understanding of PU of such OHCs and provide evidence of the value of contextualizing the antecedents to PU from the social support view with regard to specific research settings.
Furthermore, our work contributes to social support theory through its investigation of the associations between distinct types of social support in smoking cessation OHCs. Our findings suggest that perceived informational support is a prerequisite for perceived emotional support and perceived esteem support in such OHCs. Though perceived informational support has no direct impact on PU, it exerts an indirect influence on the smoking cessation OHC's PU via perceived emotional and esteem support. With this new empirical insight, we offer a plausible explanation of how specific types of social support trigger users' perceptions as to the PU of smoking cessation OHCs. The findings on the relationships between perceived informational support and perceived emotional and esteem support also provide a comprehensive understanding of the role of informational support in these OHCs from a social support perspective.
The postadoption literature also benefits from our theory-grounded insights. While previous studies investigated the determinants of continuance intention (e.g. Song et al., 2018;Wu, 2018) and of knowledge sharing separately (e.g. Yan et al., 2016;Zhang et al., 2020;Zhang et al., 2017), our study incorporated these two postadoption behaviors into a single research model in the context of smoking cessation OHCs. Our findings related to how PU affects different postadoption behaviors (e.g. continuance intention and knowledge sharing) and the relationship between these distinct postadoption behaviors in the context of smoking cessation OHCs demonstrate that intentions to continue using such an OHC could be strengthened via knowledge-sharing activities. This study provides further evidence that knowledge-sharing behavior can trigger users' continuance intention with regard to smoking cessation OHCs while also suggesting that IS continuance research should consider the impacts of additional postadoption behaviors (knowledge sharing and others) on continuance intention, not merely users' motivations for continuance intention toward the IS.
Our final key theoretical contribution is related to the gender differences in the antecedents to PU and the moderating effects of smoking cessation stage on the consequences of PU. These findings highlight the crucial roles of user-specific factors (i.e. gender and smoking cessation stage) in explaining PU in such OHCs and for bringing new insights that can advance our understanding of the antecedents and consequences of PU for particular user groups in this specific context and others. In other words, this study provides further empirical evidence that user-specific factors are closely associated with user perceptions of the usefulness of smoking cessation OHCs and their postadoption behavior regarding such OHC while also suggesting that research on smoking cessation OHCs should consider the role of user-specific factors when investigating users' postadoption behaviors regarding such OHCs, not only the usefulness of such OHCs.

Practical implications
There are several practical implications also, principally for those managing and running smoking cessation OHCs. Our findings implying that PU affects continuance intention and knowledge sharing enable us to recommend that smoking cessation OHC service providers promote retention of users and sharing of knowledge by enhancing user perceptions of the usefulness of the smoking cessation OHCs. Specifically, our finding that perceived emotional and esteem support are antecedents to PU implies that smoking cessation OHC service providers should focus on their strategies and approaches for facilitating users' perceptions of emotional and esteem support in smoking cessation OHCs. For instance, smoking cessation OHC administrators should use warm and caring language when answering inquiries and also recommend users of such OHCs to apply warm messages in their communications with others to provide emotional support to other users. They can also offer templates expressing empathy and encouragement for users to use when they respond to other users. Users may find it more convenient to make selections from predefined message starters than to type out messages of emotional and esteem support from scratch.
Another recommendation, informed by the findings on the indirect impact of perceived informational support on PU, is that smoking cessation OHC service providers should continue to encourage information and knowledge sharing in these OHCs.
Since users in different stages on the smoking cessation journey differ in their needs for social support and their perceptions of the PU of the OHC, the service providers could offer differential social support, tailored to stage-specific needs, thereby nurturing knowledge sharing and intentions to continue using the smoking cessation OHC. For instance, we found that perceived emotional support has a positive influence on PU for users who intend to quit smoking but have taken no action thus far. The findings suggest that it is crucial to provide these users with encouragement or expressions of care in online interactions in smoking cessation OHCs, which might meet their needs and enhance their perceptions of the usefulness of smoking cessation OHCs in support their quitting of smoking.
One of the most important stage-specific practical implications arises from the association between knowledge sharing and continuance intention among the after-action users. Perceived usefulness of smoking cessation OHC This link suggests that smoking cessation OHC service providers should encourage users at this stage on the journey to share more knowledge in the OHC, thereby not only strengthening their intention to continue using the smoking cessation OHC but also producing informational support for other users.

Limitations and paths for future research
This study has several limitations, which represent directions for future research. Firstly, since we limited our consideration of the antecedents to the PU of smoking cessation OHCs to social-support exchange behavior, some other external determinants of PU could be investigated further. Research could examine the role of companionship activities in determining the PU of smoking cessation OHCs and evaluate the associated differences with regard to exchanging of social support. Secondly, while we examined how social support affects two distinct postadoption behaviors indirectly via PU, we did not test the direct impact of social support on these behaviors of smoking cessation OHC users. Therefore, future research could usefully investigate whether particular types of social support directly affect these two postadoption behaviorsand others. While we limited ourselves to investigating only two consequences of the PU of smoking cessation OHCs, further work could test whether our proposed model is appropriate for studying postadoption behaviors additional to continuance intention and knowledge sharing, such as governance-and recommendationrelated outcomes (Zou et al., 2018). Finally, we recruited a relatively restricted sample of users to participate in our survey. As people with certain experiences of using smoking cessation OHCs, the respondents from these two smoking-cessation OHCs suited our focus well. However, data could be collected from more countries to address the generalizability of our findings. Furthermore, our approach and findings point to the potential benefits of carrying out similar work with data from OHCs that are focused on other specific health concerns, such as problematic gambling or abuse of alcohol or other drugs.