Although Q&A review is a special type of online reviews, it more attributes to social Q&A when considering its design features. For example, potential consumers can ask questions related to target products in Q&A reviews. Then those already-purchased consumers will receive invitations, which randomly sent by e-commerce platform systems, and decide whether or not to answer the questions. Finally, other consumers can give feedback (e.g., likes or comments) to those answers. Therefore, in this study we use consumer knowledge-sharing intention to reflect consumers’ answering intention in Q&A reviews.

Q&A reviews generally have the following advantages. First, Q&A reviews can help consumers to improve efficiency (Shah and Kitzie, 2012). Information on the internet is always overloading and disordered. Traditionally, consumers have to spend extra time in reading numerous sources of information and finding answers on online reviews, but in Q&A reviews potential consumers can access this demanding through direct interaction. The cognitive pressure from information integration also can be reduced (Bleda et al., 2021). Second, Q&A reviews can help consumers to improve the perception of credibility and usefulness. Q&A reviews that provide accurate information reduce perceived uncertainty (Reimer and Benkenstein, 2016). For example, potential consumers can ask already-purchased consumers about specific and individual questions. Moreover, in traditional online reviews, any consumers can provide their evaluations. But as the rules of Q&A reviews, when questioners post their questions, e-commerce platforms will send random invitations to some of the consumers who have already purchased the products. Then those eligible consumers can give their answers voluntarily. In this sense, some unfairly negative reviews can be filtered, and certain perceived risks can be reduced. Third, Q&A reviews realize direct interaction between consumers and thereby satisfy their social needs (Jin et al., 2015). Not only can questioners receive responses from answerers, but answerers can get feedbacks (e.g., comments and likes) from other consumers.

In this article, we regard loyalty to platform as a consumer’s intention to repeat visiting an e-commerce platform (Cyr et al., 2005). Loyalty to platform reflect the level of consumer preference for a certain e-commerce platform (Jones and Taylor, 2012). Yao et al. (2015) conducted a field survey of 222 online community members and found knowledge-sharing was positively related to members loyalty. Similar conclusion was presented by Swanson et al. (2020), although they discussed the topic within organizations.

In this study, we predicted that consumers’ knowledge-sharing intention will positively affect consumers’ loyalty to platform based on the following reasons: First, the process of Q&A reviews contains social capital, and social capital can affect consumer loyalty (Jones and Taylor, 2012). The knowledge shared by consumers can be regarded as representing the utilitarian values (i.e., social capital) of e-commerce platforms and thus lead to higher loyalty to the platforms (Wang and Fesenmaier, 2004a). Second, social interaction will directly or indirectly affect consumer loyalty to platforms (Koh and Kim, 2004). Consumers’ knowledge-sharing is a kind of direct social interaction between consumers. These interactions can induce consumers’ social sentiments and make them feel as they are talking to real people in person. Moreover, consumers also tend to deem the platform as an enthusiastic, generous and social place, and are more willing to maintain the relationship with the platform (Nadeem et al., 2020). Finally, after sharing knowledge, consumers will have a sense of responsibility and ownership toward the platform, and in turn induce a higher level of loyalty (Yao et al., 2015). Therefore, in this study we hypothesized:

In addition, we also take an initial step on assessing the moderator role of feedback between variables. In this study, we define feedback as a post-replying behavior, i.e., behavior of responding to the answers. Feedback is a key function and an important social interaction indicator in Q&A reviews, but currently few studies have examined this topic (Fang et al., 2018). Traditionally, consumers who write online reviews cannot obtain feedback from others, but in Q&A reviews, other consumers or sellers can express their opinions to answerers using the “like” or “reply” functions. This feedback mechanism can further deepen the degree of social interaction on e-commerce platforms and increase consumer engagement and loyalty (Jin et al., 2015; Guan et al., 2018).

In this study we predicted that feedback (i.e., likes and comments) can moderate the relationship between consumers’ motivations and knowledge-sharing intention. Firstly, Q&A reviews are actually a form of social interaction. If consumers are motivated to share knowledge, it means that they have a need for social interaction. Feedback is a key indicator of social dynamics (Bateman et al., 2011) and can be regarded as a dynamic communication process between consumers. Therefore, we believe that feedback can satisfy consumers’ social interaction needs and thus promote higher levels of participation (Stavrositu and Sundar, 2012). Secondly, feedback not only can enhance the willingness of people to share knowledge (e.g., Brzozowski et al., 2009), but also is a form of motivational affordance, which may affect consumers’ intrinsic motivations (e.g., Fong et al., 2019). Even more, Chen et al. (2019a) explored the interaction effect of two motivational affordances (i.e., votes and comments) and found the number of comments (which can be seen as feedback) moderated the relationship between usefulness voting and knowledge contribution.

In Q&A reviews, feedback usually takes the form of a “like” or a “comment.” As a positive form of feedback, likes can satisfy consumers’ self-competence needs and improve their level of intrinsic motivations (Fong et al., 2019), whereas comments are a direct means of social communication that can increase perceived involvement and perceived social presence, and further lead to a higher level of affective experience (Fortin and Dholakia, 2005), which in turn can influence consumers’ internal motivations and promote them to induce continuous intentions on knowledge-sharing. However, in terms of the external motivations in Q&A reviews, they mainly come from the pressure of potential consumers (or sellers), or from the rewards provided by the platform. External elements will not change along with the number of feedback responses. Therefore, we hypothesized that:

Moreover, in this study we propose that feedback can also moderate the relationship between consumers’ knowledge-sharing intention and loyalty to platform. Consumers who participate in knowledge-sharing can perceive utilitarian value, hedonic value, and social value through receiving feedback (Wang and Fesenmaier, 2004b). Those values may result in the sense of mutual benefit and consumers are thus more willing to use the e-commerce platform. For example, as a positive feedback, a “like” can enable answerers to confirm their values and make them feel social support and a sense of belonging (Chiu et al., 2006). Consumers may be more willing to participate in platform activities (Guan et al., 2018). On the other hand, the social interaction involved in feedback can increase consumers’ perceived intimacy toward the platform (Hassanein and Head, 2007), and additionally, consumers may also perceive social capital in the feedback they receive (Son et al., 2016). The content of the comments (i.e., feedback) generally supplements the incomplete information provided by original answerers or provides further discussion of the question. Comments can expand the level of knowledge and make the answerers perceive utilitarian values.

Therefore, these values may make consumers who participate in knowledge-sharing more likely to have positive attitudes toward the platform (Chung and Buhalis, 2008) and increase loyalty to platform. Thus, we hypothesized:

The hypothesized model is illustrated in Figure 1.

Although the measurement model yielded a significant chi-square (χ² = 536.95, df = 260, p < 0.05), the ratio of chi-square to degree of freedom (χ²/df = 2.07) was with the acceptable range of 2–5 (Marsh and Hocevar, 1985). Other fit statistics were also acceptable (GFI = 0.90, CFI = 0.92, AGFI = 0.90, TLI = 0.91, NNFI = 0.91, RMSEA = 0.06, SRMR = 0.05). This evidence set the foundation for the evaluations of the psychometric properties of the measurements.

In Table 1, we report the psychometric properties of the measurement model. We used several criteria to assess reliability and convergent validity. First, all Cronbach’s alphas of the measures for the latent constructs were larger than 0.70, indicating sound internal consistency reliability. Second, we calculated composite reliability using formula suggested by Fornell and Larcker (1981). The values of composite reliability ranged from 0.76 to 0.91, all were larger than 0.60, provided additional evidence for reliability (Bagozzi and Yi, 1988). Then, we calculated average variance extracted, the value ranged from 0.51 to 0.68, all were larger than 0.50, representing stricter evidence for convergence and reliability (Fornell and Larcker, 1981; Bagozzi and Yi, 1988). Finally, the factor loadings were assessed. All factor loadings were significant, and nearly all R² were larger than 0.50, representing an indication for convergent validity (Bagozzi et al., 1991).

We tested discriminant validity using method suggested by Fornell and Larcker (1981) that a scale possesses discriminant validity if the square root of average variance extracted by the underlying construct is larger than the correlations with other latent constructs. As shown in Table 2, each square roots of average variance extracted of latent constructs were larger than the correlations between possible pairs of latent constructs, thus provided supports for discriminant validity between the measures of the constructs.

In addition, we used Harman’s single-factor test to check for potential common method biases. All the indicators of the constructs were incorporated in the unrotated exploratory factor analysis. The result revealed that a single-factor only explained 34% of the variance, indicating no “general” factor in the data thus common method biases was not a big concern (Podsakoff and Organ, 1986).

In summary, the measurement model results demonstrated satisfactory psychometric properties required to proceed to the structural model evaluation and hypotheses testing.

We first tested our hypothesized conceptual model without incorporating the moderating variable. The path model revealed an acceptable fit despite of a significant chi-square (χ² = 602.22, df = 264, p < 0.05; χ²/df = 2.28; GFI = 0.90, CFI = 0.90, AGFI = 0.90, TLI = 0.90, NNFI = 0.90, RMSEA = 0.06, SRMR = 0.05).

Figure 2 illustrates the standardized path coefficients of our proposed relationships. The results revealed that altruistic motivation (β = 0.33, p < 0.01), hedonic motivation (β = 0.24, p < 0.01), perceived stress (β = 0.22, p < 0.01), and external rewards (β = 0.15, p < 0.01) had significant positive effects on consumers’ knowledge-sharing intention. H1–H4 were all supported. H5, postulating a positive relationship between knowledge-sharing intention and loyalty to platform, was also supported (β = 0.56, p < 0.01).

To test the hypotheses of the moderating role of feedback, we split the sample to two groups according whether the respondents reported they got feedback or not after they provided answers to user generated questions (N_feedback = 213; N_{no–feedback} = 42). We then compared three multigroup confirmative factor models to ensure the multigroup measurement invariance. For the first model, the same factor structure was imposed on all groups without constraining the factor loadings and intercepts to be equal across groups (χ² = 1,028.57, df = 520); for the second model, not only the factor structure, but also the factor loadings were constrained to be equal across (χ² = 1,052.99, df = 539); for the third model, the factor structure, the factor loadings, and the intercepts were all constrained to be equal across groups (χ² = 1,081.09, df = 558). Chi-square difference test indicated that there was no significant difference between the first model and the second model (Δχ² = 24.42, Δdf = 19, p = 0.18), and there was no significant difference between the second model and the third model (Δχ² = 28.10, Δdf = 19, p = 0.08), indicating strong evidence for measurement invariance across groups in terms of both metric invariance and scalar invariance (Putnick and Bornstein, 2016). Thus, measurement invariance across groups enables us to test the moderating role of feedback using multigroup path analysis.

Following procedures used in Homburg et al. (2005), we tested each of the moderated paths one by one by comparing two models, one model constrained all paths to be equal across groups (no-moderation model), and the other constrained all other paths to be equal except for the one hypothesized moderated path. We relied on chi-square difference test to examine whether the latter model was significantly better than the former model. If the chi-square improvement for the latter model is significant, that is evidence for the moderating hypothesis. As Table 3 shows, H6a and H6c were supported, H6b was not supported.

Q&A reviews are different to traditional online reviews. They are featured with social Q&A. Therefore, we examined Q&A reviews not from the perspective of eWOM, but also from social Q&A and knowledge sharing. Prior research on social Q&A are mostly in the context of online communities (Neshati et al., 2017), whereas knowledge-sharing are discussed in corporate organizations (e.g., Matsuo and Aihara, 2022; Nguyen et al., 2022). Our results extend the research perspectives of eWOM and contribute multidisciplinary to both eWOM and social Q&A.

Firstly, based on self-determined theory and the results of our empirical research, we found that consumers’ motivations for knowledge-sharing intention in Q&A reviews consist of altruistic motivations, hedonic motivations, perceived pressure, and external rewards. Among these, the former two motivations are internal, and the latter two are external. Previous studies seldom examine consumers’ motivations to write eWOM, but generally focus on relationships between specific variables and eWOM, such as satisfaction (Thakur, 2019; Singh and Söderlund, 2020), brand love (Bairrada et al., 2018), brand attitude (Izogo and Mpinganjira, 2021) and financial incentives (Qiao et al., 2020), or merely focus on the motivations to read eWOM (e.g., Teng et al., 2017; Ismagilova et al., 2020; Mainolfi et al., 2022).

Secondly, previous studies discussed the relationship between knowledge-sharing intention and loyalty either in the context of online communities (e.g., Yao et al., 2015) or organizations (e.g., Swanson et al., 2020). However, our results firstly shed light on the influence of consumers’ knowledge-sharing intention on consumers’ loyalty to platform in the context of e-commerce. Since Q&A reviews attribute to social Q&A, consumers may perceive a greater social presence, social interactions, and higher level of ownership from platforms, and thus perform higher level of loyalty.

Thirdly, we also highlight the moderation role of feedback. We define feedback as a post-replying behavior, i.e., behavior of responding to the answers. Feedback is a key mechanism of Q&A reviews. Our empirical results show that when feedback is received, consumers’ altruistic motivations have a stronger influence on knowledge-sharing intention and their knowledge-sharing intentions have a stronger influence on loyalty to platform. Although feedback did not significantly moderate the relationship between hedonic motivations and knowledge sharing intention, the evidence presented in this research still provides a new perspective of feedback, since seldom studies have examined it (Fang et al., 2018).

According to the results of our research, we found altruistic motivations, hedonic motivations, perceived pressure, and external rewards significantly influence consumers’ knowledge sharing intentions. Due to the fact that altruistic motivation is more related to personality characteristics, we suggest managers to stimulate other three motivations. For instance, practitioners can stimulate consumers’ hedonic motivations through gamify Q&A reviews. Hamari and Koivisto (2015) has pointed out that the hedonic value is the key value that consumers can perceive from gamified products or services. Gamification is more often used in a marketing context. Therefore, adding gamification elements may improve consumers’ hedonic motivations to participate in knowledge-sharing.

Besides, as perceived pressure derived from social norms, managers need to present consumers’ values and others’ active participation intentions in knowledge sharing (Zhou, 2021). Moreover, past studies suggested that external motivations can be internalized via gamification (Bittner and Schipper, 2014; Kuo and Chuang, 2016). Managers may thus gamify external rewards to realize internalization and increase the impact of extant motivations. For example, Kim and Ahn (2017) found that internal motivation has a strong positive relationship with consumer engagement and external motivation can be internalized by applying game elements to increase consumer engagement. Therefore, after consumers share their knowledge, practitioners may present some games (e.g., lottery, competitions, etc.) not only to make them gain some rewards, but also to satisfy their hedonic needs.

On the other hand, we found that feedback has a significant moderating effect. When receiving feedback, consumers’ altruistic motivations may have stronger influences on knowledge-sharing and in turn has a stronger influence on their loyalty to platform. Managers may set notification function of feedback, or use virtual external rewards (e.g., points, badges) to encourage consumers to provide more feedback and thus enhance social dynamism of e-commerce platforms.

Due to the limitation of time, we only collected data of 42 respondents who did not get feedback. The sample size problem might be the reason why some of the paths were not significant. Future studies may collect more data of people who did not get feedback and test the moderation effect of the relationship between internal motivation and knowledge sharing intentions.

Moreover, Future studies can further explore other motivations behind consumers’ knowledge-sharing intention. For example, social recognition is another possible motivation. Research in organizations has shown that the purpose of employees in contributing knowledge is to gain recognition from other colleagues (Constant et al., 1996); furthermore, in the context of social Q&A, research (e.g., Rui and Whinston, 2012) indicates that users expect to gain the attention of others by sharing knowledge, and they may even write comments that cater to other users’ interests to gain more social recognition (Fang et al., 2018). Therefore, in the context of e-commerce, consumers may also have expectations of winning social recognition from others, and this can be regarded as one motivation.

Finally, future research may place more of a focus on feedback. In this study, feedback referred to the behavior of providing comments or likes. Although responding behavior is an important social interaction indicator and feedback is also an important mechanism in Q&A reviews, there is still a lack of studies discussing this topic (Fang et al., 2018). In this research, we highlight the moderation role of feedback. However, in Q&A reviews, feedback from consumers is also spontaneous and voluntary. Hence, to exploit the critical impacts of feedback, future studies may further discuss its antecedents. Finally, in terms of outcome variables, as loyalty to platform reflects consumers’ willingness to repurchase on e-commerce platforms, second-hand data can be used in future studies to further discuss the relationship between consumer knowledge-sharing intention and repurchase frequency or other related variables.