Investigating the interplay of different recommendation sources and social closeness on consumer behaviour in the context of gift purchase

Gift giving is a socially and psychologically rich activity influenced by relational closeness and the credibility of recommendation sources. This thesis explores how recommendation source (AI vs. human) and social closeness (close vs. distant relationships) affect recommendation acceptance and gift choice behaviour.

Using a sequential explanatory mixed methods design, the study combines quantitative lab experiments (n = 459) with qualitative interviews (n = 12). Grounded in Social Exchange Theory (SET), the Stimulus-Organism-Response (SOR) model, and the Elaboration Likelihood Model (ELM), it investigates mediators such as trust and emotional resonance, and moderators including age, gender, and occupation.

Quantitative findings revealed no significant main effects due to contrasting individual preferences—some consistently favoured AI, while others preferred human input. These opposing tendencies cancelled out the anticipated effects.

Qualitative insights added depth, showing that preferences were shaped by context. Participants tended to favour AI for functional and human advice for symbolic or emotionally significant ones. Influencing factors included decision-making autonomy, emotional value, cultural expectations, and individual differences such as personality and cognitive style. Gender showed no consistent pattern.

Theoretically, the study extends SET, SOR, and ELM into emotionally and relationally complex contexts, and reconceptualises social closeness as a multidimensional construct. Practically, it supports the development of hybrid recommendation systems that combine AI’s analytical power with the emotional intelligence of human input, advocating for inclusive, transparent, and human-centric digital decision-making.