<![CDATA[Live-streaming commerce (LSC) has emerged as a prominent digital retail format by integrating real-time interaction, information exchange, and social engagement. While prior studies largely conceptualize interactive features as behavioral stimuli, limited attention has been paid to their role as institutional mechanisms that coordinate trust and purchasing behavior under uncertainty. This study examines how interactive features—particularly Perceived Streamer–Viewer Interaction (PSVI)—influence purchase intention in LSC featuring live comment functionality. Using a quantitative survey approach, data were collected from 108 users who had participated in LSC sessions within the past three months. The study employs Robust Least Squares (RLS) regression to account for heteroscedasticity and potential outliers. The model includes PSVI, User-to-User Engagement (UUE), Comment Availability (CA), Streamer Quality (SQ), Screen Time Intensity, and Gender as explanatory variables. The results reveal that PSVI has a strong, positive, and highly robust effect on purchase intention across all specifications, confirming its role as the primary driver of transactional behavior. In contrast, UUE does not exhibit a significant direct effect, suggesting that peer interaction primarily shapes social atmosphere rather than purchase decisions. Notably, CA demonstrates a negative and statistically significant relationship with purchase intention, indicating information overload in high-velocity comment environments. SQ shows a conditional negative effect when interaction and information variables are jointly controlled, highlighting a potential trade-off between professionalism and perceived authenticity. Robustness checks excluding UUE further confirm the stability of the main findings.]]>
