Table 1 Experiments overview Study... | Emerald Publishing

Table 1

Experiments overview

Study	Method and design	Variables	Forms of interactions	Scenario
Study 1a	Experimental vignette method 2 (AI vs. human) x 2 (successful vs. unsuccessful performance)	DV: Locus of causality (3-item scale) Controls: age, gender, managerial experience, number of subordinates	Participant witnessed successful (vs. unsuccessful) interaction between: AI agent and human manager; vs. human agent and human manager	Marketing managers viewed an email exchange between two companies and imagined themselves as a food company manager making a marketing decision about a trade show. After observing the success or failure of the interaction, they reflected on the reasons and provided locus of causality ratings
Study 1b	Real interactive experiment using ChatGPT-based AI Successful vs. unsuccessful conditions were provided by median split technique	DV: Locus of causality (1-item semantic differential scale) Grouping variable: Satisfaction with business interaction (4-items scale) Controls: age, gender, managerial experience, number of subordinates	Participants engaged in a real interaction with the AI agent	Marketing managers, recruited on Prolific, were directed to a specially designed website where they could engage in an open conversation with an AI agent (that had access to data on different ads effectiveness). Their task was to use the AI’s assistance to determine which advertisement they should choose. They then made this choice and rated their satisfaction with the interaction, followed by an assessment of whether the reasons for their satisfaction or dissatisfaction were due to internal vs. external factors (locus of causality)
Study 2	Experimental vignette method: AI vs. human Agent	DV: Satisfaction with business interaction (4-items scale) Mediator: Locus of causality (3-item scale)	Participant witnessed successful interaction between: AI agent and human manager; vs. human agent and human manager	The same as in Study 1a

Study	Method and design	Variables	Forms of interactions	Scenario
Study 1a	Experimental vignette method2 (AI vs. human) x 2 (successful vs. unsuccessful performance)	DV: Locus of causality (3-item scale)Controls: age, gender, managerial experience, number of subordinates	Participant witnessed successful (vs. unsuccessful) interaction between: AI agent and human manager; vs. human agent and human manager	Marketing managers viewed an email exchange between two companies and imagined themselves as a food company manager making a marketing decision about a trade show. After observing the success or failure of the interaction, they reflected on the reasons and provided locus of causality ratings
Study 1b	Real interactive experiment using ChatGPT-based AISuccessful vs. unsuccessful conditions were provided by median split technique	DV: Locus of causality (1-item semantic differential scale)Grouping variable: Satisfaction with business interaction (4-items scale)Controls: age, gender, managerial experience, number of subordinates	Participants engaged in a real interaction with the AI agent	Marketing managers, recruited on Prolific, were directed to a specially designed website where they could engage in an open conversation with an AI agent (that had access to data on different ads effectiveness). Their task was to use the AI’s assistance to determine which advertisement they should choose. They then made this choice and rated their satisfaction with the interaction, followed by an assessment of whether the reasons for their satisfaction or dissatisfaction were due to internal vs. external factors (locus of causality)
Study 2	Experimental vignette method: AI vs. human Agent	DV: Satisfaction with business interaction (4-items scale)Mediator: Locus of causality (3-item scale)	Participant witnessed successful interaction between: AI agent and human manager; vs. human agent and human manager	The same as in Study 1a

Source(s): Table by authors