With the rapid development of live-streaming e-commerce, real-time interaction has emerged as a salient situational factor that co-varies with consumer behavior. Existing research, however, has largely treated live-stream interactions as a homogeneous...
With the rapid development of live-streaming e-commerce, real-time interaction has emerged as a salient situational factor that co-varies with consumer behavior. Existing research, however, has largely treated live-stream interactions as a homogeneous construct and has paid insufficient attention to the functional heterogeneity among different interactive signals during the consumer decision process. As a result, how distinct types of interactions co-occur with viewing and purchasing behavior remains insufficiently characterized. To address this gap, the present study focuses on different types of interactive signals in live-streaming e-commerce and examines their differentiated associations with consumer behavior. Because the analysis relies on cross-sectional observational data, all reported findings should be read as associations rather than as evidence of causal effects.
Drawing on Social Proof Theory, Social Presence Theory, and the literature on information processing, this study treats likes as a proxy for the emotional dimension of social proof and treats comments and bullet chats as a proxy for the informational dimension of social proof—i.e., as signals that, relative to likes, exhibit a higher information affordance. The study examines how these two types of signals are differentially associated with viewing behavior and purchasing behavior, and further introduces virtual-streamer livestreams as a comparative setting to explore potential boundary differences across streamer types.
A multimodal data-analytic strategy is adopted. The analytical framework integrates interaction metrics, textual semantics of comments and bullet chats, streamer characteristics, and visual presentation variables. Drawing on data from 434 real-streamer livestreams and a smaller sample of virtual-streamer livestreams from Taobao Live, multiple regression models are estimated to assess the associations between interactive signals and consumer behavior. The empirical patterns suggest that: (1) different types of interactive signals are associated with consumer behavior in differentiated ways; (2) likes, as a proxy for emotional social proof, exhibit stronger associations with viewing-stage behaviors; (3) comments and bullet chats, as a proxy for the informational dimension of social proof, exhibit relatively stronger associations with purchasing behavior; and (4) the strength and the differentiated pattern of these associations appear to vary between real-streamer and virtual-streamer contexts, though the latter finding remains exploratory given the limited virtual-streamer sample.
Theoretically, this study extends the explanatory framework of live-stream interaction research by foregrounding the functional differentiation among types of interactive signals, and provides new empirical evidence for understanding how different kinds of interactions co-vary with consumer decision making. Practically, the findings offer implications for the design of interactive features on live-streaming platforms, for streamer-side operational strategies, and for the development of virtual-streamer livestreams.