How does online streaming reform e-commerce? An empirical assessment of immersive experience and social interaction in China

The study addresses why live-stream commerce has outperformed traditional e-commerce in China, where live streaming integrates social interaction and dynamic product presentation. Traditional e-commerce has been criticized for information-unilateral experiences lacking social and visual engagement. Given rapid user adoption of live streaming in China, the research investigates a gap in the literature: few models jointly consider both individual (perceived interactivity) and situational (dynamic characteristics and atmosphere cues) influences on consumers’ internal states and purchase intentions. The purpose is to integrate information processing theory (IPT) and flow theory within an S-O-R framework to explain how live-streaming attributes drive immersive experience and social interaction, which in turn shape online purchase intention.

The paper adopts the Stimulus-Organism-Response (S-O-R) framework, integrating IPT and flow theory. IPT posits that individuals selectively process stimuli from both individual and situational sources; flow theory explains optimal engagement states that focus attention and filter extraneous stimuli. The model proposes three live-streaming stimuli: (1) dynamic characteristics (convenience, interest), enabling participation anytime/anywhere and elevating pleasure and control; (2) atmosphere clues (information content, navigation, interface/appearance), akin to store atmospherics that shape cognition and emotion online; and (3) perceived interactivity (responsiveness, personalization), enabling two-way, tailored communication. These stimuli are expected to elicit organismic states: immersive experience (valence, concentration, control) and social interaction (streamer–customer, customer–customer), culminating in the response: purchase intention. Hypotheses posit that immersive experience mediates effects of dynamic characteristics, atmosphere clues, and interactivity on purchase intention (H1–H3), and that social interaction similarly mediates these relationships (H4–H6). Prior work on online atmospherics, interactivity, and flow, as well as social presence/interaction literatures, underpin the hypothesized paths.

Design: Cross-sectional survey with a two-stage process (pre-test and formal survey) conducted in 2020 among consumers in China who had live-stream shopping experience. Sample: Voluntary online sampling with a filter question; 408 valid responses. Demographics: 65.0% female; 66.4% under 25; 71.3% undergraduate degree; 78.2% monthly income up to US$940. Sample aligns with market distributions of live-stream users in China. Measures: Six constructs measured on 5-point Likert scales (1=strongly disagree; 5=strongly agree). Dynamic characteristics (convenience, interest; 8 items from Lu & Su, 2009). Atmosphere clues (information content, navigation system, appearance design; 9 items from Floh & Madlberger, 2013). Perceived interactivity (responsiveness, personalization; 6 items from Wu et al., 2014). Immersive experience (valence, concentration, control; 9 items from Koufaris, 2002). Social interaction (streamer–customer and customer–customer interaction; 6 items from Chou et al., 2022). Purchase intention (3 items from Kolesar & Galbraith, 2000). Cronbach’s alphas ranged 0.74–0.91; 41-item scale alpha=0.96. One item with SFL 0.562 (VA3) was retained for content validity; AVE for IE remained above 0.5. Analysis: Confirmatory factor analysis (CFA) and structural equation modeling (SEM) using AMOS 22. Reliability assessed via standardized factor loadings (SFLs) and composite reliability (CR); validity via average variance extracted (AVE) and discriminant validity. Model fit evaluated using χ²/df, GFI, CFI, IFI, SRMR, and RMSEA. Collinearity diagnostics used VIF and Tolerance; all within acceptable ranges (VIF 2.55–3.50; Tolerance 0.27–0.39).

Measurement model: Good reliability and validity. CRs: DC≈0.95; AC≈0.97; PIA≈0.95; IE≈0.95; SI≈0.93; PI≈0.85. AVEs all >0.50; discriminant validity satisfied (square roots of AVE exceed inter-construct correlations). Inter-construct correlations r≈0.60–0.75 (p<0.01). Model fit: Measurement model—χ²/df=1.772; GFI=0.858; CFI=0.943; IFI=0.943; SRMR=0.041; RMSEA=0.044. Structural model—χ²/df=1.826; GFI=0.854; CFI=0.939; IFI=0.939; SRMR=0.041; RMSEA=0.045. Structural paths (standardized): - DC → IE: 0.148, p<0.001 (significant) - DC → SI: 0.017, p=0.643 (ns) - AC → IE: 0.577, p<0.001 - AC → SI: 0.476, p<0.001 - PIA → IE: 0.252, p<0.001 - PIA → SI: 0.504, p<0.001 - IE → PI: 0.385, p<0.001 - SI → PI: 0.218, p<0.001 - Direct to PI: DC → PI: 0.078, p=0.033; AC → PI: 0.206, p<0.001; PIA → PI: 0.043, p=0.344 (ns) Explained variance (R²): IE=0.418; SI=0.480; PI=0.474. Mediation results: - H1: IE partially mediates DC → PI (DC→PI remains significant). - H2: IE partially mediates AC → PI; additionally, SI partially mediates AC → PI (H5 partially supported). - H3: IE fully mediates PIA → PI (no direct PIA→PI effect). - H4: Not supported (DC does not significantly affect SI). - H6: SI fully mediates PIA → PI.

Findings substantiate that live-streaming attributes operate through cognitive-affective organismic states to influence purchase intention. Dynamic characteristics enhance immersive experience and have a small direct effect on purchase intention but do not strengthen social interaction, indicating that convenience/interest alone do not build interactive social bonds online. Atmosphere cues strongly bolster both immersive experience and social interaction, yielding both mediated and direct effects on purchase intention. Perceived interactivity drives purchase intention entirely via immersive experience and social interaction, underscoring the centrality of responsive, personalized, two-way communication in live-stream commerce. These results address the research question by demonstrating that live-stream commerce outperforms traditional e-commerce because it better triggers flow-like immersive states and real-time social exchanges, mechanisms absent or weaker in traditional formats. Practically in China, while platforms deliver strong atmospherics and interactivity, dynamic features currently elevate immersion more than social interaction, signaling a need to enhance infrastructure and design for richer, trustworthy social engagement.

This study integrates IPT and flow theory within an S-O-R framework to explain how live-streaming attributes—dynamic characteristics, atmosphere clues, and perceived interactivity—shape immersive experience and social interaction and, in turn, purchase intention. The model shows strong explanatory power (R²≈0.42–0.47) and validates key mediated pathways: atmosphere cues and interactivity influence purchase intention primarily via immersive and social experiences, while dynamic characteristics affect immersion and immediate purchase but not social interaction. Contributions include a unified account of individual and situational stimuli, identification of distinct mediating mechanisms, and guidance for live-stream platform design. Future research should validate the model across countries and platforms; examine the role of immersive technologies (AR/VR/AI) in enhancing presence and empathy; test category contingencies (e.g., involvement level, search vs. experience goods); and further explore how perceived control and trust shape flow states and downstream behavior.

The study is cross-sectional and limited to China’s live-stream commerce context, constraining causal inference and generalizability. The voluntary online sample, while aligned with market demographics, may introduce self-selection bias. Measurement relies on self-reports. The model does not explicitly incorporate technological affordances such as AR/VR/AI, nor does it test category differences (e.g., high/low involvement). Social interaction’s non-significant link with dynamic characteristics suggests unmeasured moderators (e.g., platform norms, community maturity) warrant exploration.