Trust, interaction mechanisms and theoretical foundations in purchase-related social commerce: A PRISMA 2020 systematic review

Social commerce combines social media technologies with digital purchasing platforms and is a significant channel for consumer transactions. Despite growing research, this literature remains fragmented: theoretical frameworks lack an integrating logic, trust is rarely treated as referent-specific, and mediation is more often asserted than formally estimated. This review aims to explore the application and positioning of theoretical frameworks, trust, and social interaction mechanisms in studies modelling purchase-related consumer outcomes in social commerce. A total of 61 studies were identified from the records of Scopus, Web of Science, and IEEE Xplore and screened based on pre-specified eligibility criteria. Reporting was done in accordance with PRISMA 2020 and PRISMA-S; quality was assessed using the Mixed Methods Appraisal Tool 2018; and full texts were coded using a thematic codebook. Forty-eight studies formally name at least one theoretical framework (77 applications). The Stimulus-Organism-Response framework is the most common (17 studies), followed by trust transfer theory (9), social exchange theory (6), the Technology Acceptance Model (5), and the Unified Theory of Acceptance and Use of Technology (1). Trust is directed towards the seller (23 studies), platform (7), online community (6), content or source (5), product (3) and algorithmic recommendation system (2), and 17 studies do not specify the referent. 47 studies report trust as a mediator, and 58 studies report mediation, but only 49 estimate it using a named procedure. In 57 studies, interaction mechanisms are environmental stimuli and in 2 studies, they are behavioural outcomes. There are 54 cross-sectional studies and 54 studies that measure self-reported intention. The evidence thus does not necessarily follow an adoption logic, trust is multi-referent and often unspecified, mediation is more often claimed than calculated, and consumers are not necessarily modelled as producers of the content that affects other consumers.