*** Interdependence and Communication Style in Human-Agent/Robot Teamwork ***
Authors: R.S. Verhagen, M.A. Neerincx, M.L. Tielman
Interactive Intelligence, Faculty of Electrial Engineering, Mathematics, and Computer Science, Delft University of Technology

Corresponding author: R.S. Verhagen

Contact information:

r.s.verhagen@tudelft.nl

Delft University of Technology - Faculty of Electrical Engineering, Mathematics, and Computer Science
Van Mourik Broekmanweg 6
2628 XE Delft
The Netherlands

*** General Introduction ***
This dataset was gathered in an experiment with the aim of studying the effects of interdependence and robot communication style on human-robot teamwork. Human participants were presented with a virtual (urban search and rescue inspired) task and robot. Robot communication style varied between participants while interdependence between human and robot varied within participants. The goal for the human-robot team was to search and rescue all victims as soon as possible. We wanted to gain insights into the following:
- How does robot communication style influence human-robot teamwork?
- How does interdependence between human and robot influence human-robot teamwork?
- How does interdependence influence the relationship between robot communication style and human-robot teamwork?

*** Description of the data in this data set ***
The excel file contains the following data:
- Gender: gender of the participant with answer options 1: Female, 2: Male, 3: Other, 4: Prefer not to say.
- Age: age of the participant with answer options 1: 12-17, 2: 18-24, 3: 25-34, 4: 35-44, 5: 45-54, 6: 55-64, 7: 65+, 8: Prefer not to say.
- Education: highest degree or level of education completed by the participant with answer options 1: No schooling completed, 2: Some high school, no diploma, 3: High school graduate, 4: Some college credit, no degree, 5: Associate degree, 6: Bachelor's degree, 7: Master's degree, 8: Ph.D. degree or higher.
- Gaming: how often the participant played video games with answer options 1: Several times a year, 2: Several times a month, 3: Several times a week, 4: Daily.
- Version: since the experiment consisted of two search and rescue tasks, one for each interdependence condition, we created two experiment order versions. Version 1 started with the low interdependence condition followed by the high interdependence condition. Version 2 started with the high interdependence condition followed by the low interdependence condition.
- Communication: participants teamed up with one of four versions of the robot, varying in communication style. They collaborated with either the silent, transparent, explainable, or adaptive version.
- Interdependence: participants completed two search and rescue tasks, varying in the level of interdependence between participant and robot (low vs. high).
- Trust: we measured subjective trust in the robot after completing each task using eight questions on a 5-point Likert scale and calculated the mean of the eight items as the final trust score.
- Workload: we measured subjective workload during the task after completing each task using six questions on a 100-point slider scale and calculated the mean of the six items as the final workload score.
- Understanding: we measured subjective robot understanding after completing both tasks using eight questions on a 7-point Likert scale and calculated the mean of the eight items as the final understanding score.
- SA: we measured objective situation awareness twice during each task using eight questions and calculated the mean percentage of correctly answered questions as the final situation awareness score. 
- Completeness: we logged completeness of the task after completing each task as the percentage of victims rescued.
- Accuracy: we logged accuracy of the task after completing each task as the percentage of victims collected in the correct order.
- Ticks: we logged the time taken to finish the task, expressed in ticks and with one tick representing 0.05 seconds, after completing each task. 
- HumanReliance: we logged the percentage of victims found by the participant but rescued by the robot after completing each task.
- HumanMessages: we logged the number of messages sent from the participant to the robot after completing each task.
- HumanContribution: we logged the percentage of total victims rescued by the participant after completing each task.
- TimeLeft: we transformed the time taken to finish the task, expressed in ticks, into the percentage of time left to finish the task.
- Performance: we calculated team performance as the mean of TimeLeft, Completeness, and Accuracy. 
