With the deep coupling of mobile internet and real-time streaming technologies, live streaming social commerce (LSC), which refers to a business model combining real-time video streaming with e-commerce and social features to enable interactive product promotion and sales, has experienced explosive growth (Liu & Zhang, 2024a; Zheng et al., 2023). In 2024, China’s LSC market reached CNY 5.8 trillion (USD 0.81 trillion), with its user base surpassing CNY 597 million (USD 833 million). Unlike human streamers, virtual streamers are digital avatars created with motion-capture and speech-synthesis technology (Gao et al., 2023). These virtual streamers reduce labor costs and keep brand image perfectly consistent (Liu & Zhang, 2024b), yet the bond users form with these avatars remains parasocial, as users know the streamer is code, but bullet chats, voice calls, and virtual gifts still spark real-time emotion in users (McLaughlin & Wohn, 2021). Many studies have linked human streamers’ cues to product purchase, but their effect on customer citizenship behavior (CCB), which refers to customers’ voluntary, nontransactional behaviors that benefit the streamer or platform (e.g., word of mouth, helping other customers, and providing feedback), is limited (Men et al., 2023), especially for virtual streamers, leaving theory and practice short on nontransactional value guidance.

 

Social exchange theory posits that unexpected relational benefits spur reciprocal extrarole behaviors (Cropanzano & Mitchell, 2005). In LSC, perceived social benefit—an individual’s sense of belonging, being noticed, and community identification (Kuo & Feng, 2013)—is amplified by virtual streamers’ never-reject presence, which lowers social anxiety (Qin & Liu, 2025), and by public, algorithmic microrewards such as name-checks, exclusive emojis, and badges that boost self-presentation (Li et al., 2025). Once the live room is experienced as a second social space, perceived social benefit converts into affective commitment to the streamer and brand, inducing spontaneous recommendations, peer help, and feedback (Jiang et al., 2024) and thereby reducing platform costs while building brand–community social capital (Men et al., 2023).

 

However, it remains unclear how individuals’ parasocial interactions (PSI)—defined here as perceived one-sided, interpersonal-like relationships that viewers develop with media personas such as virtual streamers—translate into CCB through the mediating path of perceived social benefit. To illuminate these relationships, this study proposed a sequential PSI → perceived social benefit → CCB framework. By embedding PSI in the context of virtual streamers, we aimed to not only extend the boundary conditions of classic media–psychology constructs, but to also generate actionable recommendations for brands and platforms to refine streamer persona design, interactive scripts, and community reward systems, thereby converting affective bonds into durable, nontransactional value for users such as voluntary advocacy, content cocreation, and norm enforcement.

The Current Study

After the Ethics Committee of Zhejiang University of Water Resources and Electric Power granted approval, we released the survey through Wenjuanxing, an online platform for questionnaires, tests, and polls (Liu et al., 2022; Liu & Zhang, 2024c). We used a yes/no screening question (“Have you watched virtual streamers in live streaming social commerce before?”) to select qualified participants. Those who answered “no” were excluded. The remaining participants were informed about the aims of the study and provided informed consent. They responded to items based on their latest viewing occurrence. Of 387 surveys we received, we discarded 41 due to missing fields or contradictory replies, yielding 346 usable cases (effective response rate = 89.41%). The final sample comprised 153 (44.22%) men and 193 (55.78%) women, with a mean age of 22.54 years (SD = 3.96, range = 18–36). Each participant received CNY 10 (USD 1.40) after finishing the questionnaire.

All items were rated on a 5-point Likert scale ranging from 1 = strongly disagree to 5 = strongly agree.

 

We measured PSI using three items adapted from Xu et al. (2024). Sample items are “The virtual streamer made me feel comfortable, as if I was with friends” and “The virtual streamer could understand the questions I wanted to ask very well.” Cronbach’s alpha in this study was .79.

 

We measured perceived social benefit using four items adapted from Kuo and Feng (2013). Sample items are “The live stream enhanced my relationships with other consumers” and “The live stream could expand my social network and allow me to make new friends.” Cronbach’s alpha in this study was .82.

 

We measured CCB using six items adapted from Jiang et al. (2024). Sample items are “I assisted other consumers if they needed my help” and “I encouraged friends to watch the virtual streamer.” Cronbach’s alpha in this study was .85.

We used structural equation modeling in Amos 21.0 to analyze the data, test our hypotheses, and assess measurement model reliability and validity, and examined the PSI → perceived social benefit → CCB framework with a bootstrapping analysis (5,000 resamples).

As shown in Table 1, composite reliability values surpassed the .70 benchmark, while average variance extracted values remained comfortably above .50. Furthermore, the overall fit indices corroborated the model’s alignment with the observed data. Collectively, these metrics established that the measurement instrument achieved robust reliability and solid validity. Harman’s single-factor test showed that the first extracted factor with an eigenvalue greater than 1 explained 37.85% of variance, which is lower than the critical criterion of 40%. This suggested no significant common method bias was present in our study (Podsakoff et al., 2003).

We evaluated the hypotheses using structural equation modeling with 95% confidence intervals (CIs) in Amos 21.0. The full pattern of these results is shown in Figure 2. Hypothesis 1 was supported, in that PSI with virtual streamers during within LSC had a positive relationship with CCB, β = .22, p < .01. Further, PSI with virtual streamers was positively associated with perceived social benefit, β = .46, p < .001, 95% CI [0.33, 0.57], and perceived social benefit, in turn, was associated with CCB, β = .39, p < .01, 95% CI [0.25, 0.52]. A bootstrapping analysis with 5,000 resamples revealed a significant indirect path from PSI with virtual streamers to CCB via perceived social benefit, β = .18, p < .001, 95% CI [0.11, 0.27], and that the proportion of the indirect effect to the total effect was 45%, thereby supporting Hypothesis 2.

By situating virtual streamers within the LSC context, this study clarified how PSI with virtual streamers influences CCB among viewers. We found that PSI with virtual streamers was positively related to CCB, and this effect was partially mediated by perceived social benefit. The PSI → perceived social benefit → CCB chain revealed that even nonhuman interactants can activate the reciprocity norm among viewers of LSC, provided they deliver reliable, emotionally charged social cues. Specifically, algorithm-driven real-time feedback—such as name-checking, exclusive emojis, and virtual badges—supply public symbolic rewards that elevate users’ self-presentation and community status (Li et al., 2025), encouraging them to perceive the live room as a second social space. Within this space, accumulated perceived social benefits are converted into affective commitment toward both the virtual streamer and the sponsoring brand, ultimately triggering spontaneous CCB such as making recommendations, helping other consumers, and giving proactive feedback (Jiang et al., 2024).

Theoretically, this study relocates PSI from its traditional fixation on human influencers and one-way mass-media personas into the dynamic, real-time domain of artificial-intelligence-rendered streamers. Our findings demonstrated that algorithmically orchestrated recognition moments such as time-stamped badges and synchronized emoji showers emitted a hypersocial cue density that viewers interpreted as authentic interpersonal feedback, thereby satisfying the symbolic reward precondition for social exchange activation (Li et al., 2025) and eliciting genuine extrarole labor even when users were fully aware that the interaction partner lacked biological authenticity. Simultaneously, by inserting perceived social benefit as an affective mediator, the study linked previously disconnected streams of brand community research, which foregrounds belonging and community identification as antecedents of commitment, and customer citizenship research, which has rarely examined nonhuman sources of prosocial motivation (Men et al., 2023). The resulting granular PSI → perceived social benefit → CCB pathway explains why consumers willingly work for the brand in the absence of monetary or loyalty program incentives, and positions perceived social benefits as an emotional ledger that converts algorithmic attention into voluntary value-creation behaviors.

The findings demonstrate that PSI with virtual streamers drives CCB through the mediating role of perceived social benefit. For managers operating in LSC, two actionable strategies emerge directly from this evidence. First, virtual streamers should be designed to deliver algorithmically orchestrated social cues that viewers interpret as authentic interpersonal feedback. Real-time name-checking, exclusive emojis, and virtual badges serve as public symbolic rewards that elevate users’ self-presentation and community status, satisfying the precondition for social-exchange activation. These recognition moments—deployed consistently throughout the stream—trigger the reciprocity norm even when users understand the interactant is nonhuman. Second, perceived social belonging should be cultivated as a critical affective bridge between interaction and CCB. Managers should structure the live streaming room environment as a second social space, where accumulated perceived social benefits convert into affective commitment toward both the virtual streamer and the sponsoring brand. This entails maintaining conversational continuity across sessions and visibly acknowledging returning viewers to reinforce their community standing. These strategies leverage the specific PSI → perceived social benefit → CCB mechanism identified in this study, offering a targeted framework for transforming algorithmic interaction into sustainable brand advocacy through virtual streamer deployment.

First, self-reported survey data are vulnerable to temporal ambiguity, and their single-timepoint nature precludes the establishment of causal relationships. Thus, longitudinal or experimental designs that track viewers across multiple streaming sessions are needed to firmly establish causality. Second, the virtual streamer was treated as a homogeneous stimulus, and avatar realism, voice gender, interactivity granularity, and reinforcement schedules were not manipulated. Future research could isolate which of these design features amplify or dampen the link between PSI and perceived social benefit.

Cropanzano, R., & Mitchell, M. S. (2005). Social exchange theory: An interdisciplinary review. Journal of Management, 31(6), 874–900. https://doi.org/10.1177/0149206305279602

 

Gao, W., Jiang, N., & Guo, Q. (2023). How do virtual streamers affect purchase intention in the live streaming context? A presence perspective. Journal of Retailing and Consumer Services, 73, Article 103356. https://doi.org/10.1016/j.jretconser.2023.103356

 

Jiang, C., He, L., & Xu, S. (2024). Relationships among para-social interaction, perceived benefits, community commitment, and customer citizenship behavior: Evidence from a social live-streaming platform. Acta Psychologica, 250, Article 104534. https://doi.org/10.1016/j.actpsy.2024.104534

 

Kuo, Y.-F., & Feng, L.-H. (2013). Relationships among community interaction characteristics, perceived benefits, community commitment, and oppositional brand loyalty in online brand communities. International Journal of Information Management, 33(6), 948–962. https://doi.org/10.1016/j.ijinfomgt.2013.08.005

 

Li, Y., Lin, Q., Luo, X., & Liu, B. (2025). Investigating the impact of virtual streamers on the engagement of Generation Z consumers: The moderating effect of congruence. Journal of Organizational and End User Computing, 37(1), 1–23. https://doi.org/10.4018/JOEUC.380656

 

Liu, X., & Zhang, L. (2024a). Impacts of different interactive elements on consumers’ purchase intention in live streaming e-commerce. PLOS One, 19(12), Article e0315731. https://doi.org/10.1371/journal.pone.0315731

 

Liu, X., & Zhang, L. (2024b). Virtual streamer responsiveness and consumer purchase intention in live streaming commerce: Social presence as a mediator. Social Behavior and Personality: An international journal, 52(11), Article e13708. https://doi.org/10.2224/sbp.13708

 

Liu, X., & Zhang, L. (2024c). Entrepreneurial bricolage, business model innovation, and sustainable entrepreneurial performance of digital entrepreneurial ventures: The moderating effect of digital entrepreneurial ecosystem empowerment. Sustainability, 16(18), Article 8168. https://doi.org/10.3390/su16188168

 

Liu, X., & Zhang, L. (2025). Virtual streamer anthropomorphism affects consumers’ purchase intention: The mediating role of social presence. Social Behavior and Personality: An international journal, 53(7), Article e14517. https://doi.org/10.2224/sbp.14517

 

Liu, X., & Zhang, L. (2026). Virtual streamers’ social presence and consumer engagement in live streaming social commerce: Immersive experience as a mediator. Social Behavior and Personality: An international journal, 54(8), Article e16148. https://doi.org/10.2224/sbp.16148

 

Liu, X., Zhang, L., & Chen, Q. (2022). The effects of tourism e-commerce live streaming features on consumer purchase intention: The mediating roles of flow experience and trust. Frontiers in Psychology, 13, Article 995129. https://doi.org/10.3389/fpsyg.2022.995129

 

McLaughlin, C., & Wohn, D. Y. (2021). Predictors of parasocial interaction and relationships in live streaming. Convergence: The International Journal of Research into New Media Technologies, 27(6), 1714–1734. https://doi.org/10.1177/13548565211027807

 

Men, J., Zheng, X., Shan, F., Shi, X., & Yang, F. (2023). Exploring customer citizenship behavior in social commerce from the parasocial perspective. Social Science Computer Review, 42(3), 681–699. https://doi.org/10.1177/08944393231214026

 

Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88(5), 879–903. https://doi.org/10.1037/0021-9010.88.5.879.

 

Qin, M., & Liu, M. (2025). How consumer perceptions of virtual streamers affect purchase intention in live streaming commerce: Based on cognition-affect-behavior model. Asia Pacific Journal of Marketing and Logistics, 37(6), 1679–1695. https://doi.org/10.1108/APJML-06-2024-0736

 

Xu, H., Law, R., Lovett, J., Luo, J. M., & Liu, L. (2024). Tourist acceptance of ChatGPT in travel services: The mediating role of parasocial interaction. Journal of Travel & Tourism Marketing, 41(7), 955–972. https://doi.org/10.1080/10548408.2024.2364336

 

Zheng, S., Chen, J., Liao, J., & Hu, H.-L. (2023). What motivates users’ viewing and purchasing behavior motivations in live streaming: A stream-streamer-viewer perspective. Journal of Retailing and Consumer Services, 72, Article 103240. https://doi.org/10.1016/j.jretconser.2022.103240

Xiaoli Liu, School of Economics and Management, Zhejiang University of Water Resources and Electric Power, No. 583 Xuelin Street, Xiasha, Hangzhou 310018, People’s Republic of China. Email: [email protected]