LDA-SEM-fsQCA analysis of AI digital human anchor traits on consumers’ purchase intention in agricultural e-commerce: A mixed-methods study

The prominent contradictions between the pain points of difficult verification of agricultural product regional authenticity and large cross-regional trust gaps in agriculture-supporting livestreaming, and between the “de-regionalized expression†and “template-based emotional interaction†of artificial intelligence (AI) digital humans, have become increasingly prominent. Focusing on the vertical scenario of AI digital humans in agriculture-supporting livestreaming, this study constructs an integrated SOR-PAD-T analysis framework based on the Stimulus-Organism-Response (SOR) model and the Pleasure-Arousal-Dominance (PAD) emotional theory. Adopting a mixed research method combining Latent Dirichlet Allocation topic modeling, Partial Least Squares Structural Equation Modeling, and Fuzzy Set Qualitative Comparative Analysis, it reveals the complete mechanism of “trait-emotion-cognition-purchase intention†through which AI digital human streamers drive consumers' purchase intention. The findings are as follows: (1) Four core traits—namely, interactivity, persona matching, dialect use, and story-based narration—constitute the stimulus system in the agriculture-supporting scenario, among which interactivity and persona matching exert a direct positive impact on purchase intention, while dialect use and story-based narration act indirectly through the emotion-cognition chain. (2) Consumers’ emotions and cognition form a progressive chain of “arousal→pleasure→dominance→trust†with trust being the strongest driving factor for purchase intention. (3) Six paths to high purchase intention are identified and categorized into two configurations, which are suitable for agricultural business entities with different resource endowments. This study constructs a scenario-specific SOR-PAD-T framework for agriculture-supporting livestreaming, enriches the scenario-based theoretical system of research on AI digital human streamers, and provides practical guidance for the digital upward flow of agricultural products and the digital transmission of rural trust in the context of rural revitalization.