Digital Twin Robo-Advisor Experiment Dataset

This dataset contains respondent-level data from a randomized online experiment on digital twin robo-advisors and conversational AI in investment services. The study examines how different robo-advisor designs influence investor trust, privacy concern, perceived personalization, adoption intention, willingness to follow advice, and hypothetical allocation behavior.

The experiment uses a 2 × 2 between-subjects design. Respondents were randomly assigned to one of four fictional robo-advisor conditions: (1) standard robo-advisor with plain dashboard interface, (2) standard robo-advisor with conversational AI interface, (3) digital twin robo-advisor with plain dashboard interface, and (4) digital twin robo-advisor with conversational AI interface. In all conditions, the recommended portfolio remained constant, allowing the study to isolate the effects of personalization architecture and interface style.

The digital twin conditions describe a continuously updated financial profile built, with user permission, from factors such as goals, risk tolerance, income pattern, spending behavior, savings rate, liabilities, household needs, tax context, and projected life events. The conversational AI conditions present the recommendation through a chatbot-like advisory interface.

The dataset includes responses to manipulation checks, trust measures, privacy concern measures, perceived usefulness and personalization items, adoption-related items, hypothetical investment allocation responses, and demographic controls. It is intended for research on fintech adoption, robo-advisors, AI-enabled financial services, trust in automated advice, privacy trade-offs, and digital personalization in wealth management.

The dataset supports the study: “From Robo-Advisors to Financial Digital Twins: A Randomized Experiment on Conversational AI, Trust, Privacy, and Adoption.”