A Meta-Analysis of Factors Influencing the Development of Trust in Automation: Implications for Human-Robot Interaction

Report No. ARL-TR-6984
Authors: Kristin E. Schaefer; Deborah R. Billings; James L. Szalma; Jeffrey K. Adams; Tracy L. Sanders; Jessie Y. C. Chen; Peter A. Hancock;
Date/Pages: July 2014; 70 pages
Abstract: Trust has become increasingly important because of the increasing need for synergistic human-machine teaming. Here, we expand on our previous meta-analytic work in the field of human-robot interaction (HRI) to include all of human-automation interaction. We used meta-analytic procedures to assess trust. A total of 343 articles were reviewed, of which 30 studies, providing 164 effect sizes, and 16 studies, providing 63 correlational effect sizes, met the selection criteria for analysis. The overall experimental effect on trust was g= +0.48, and the correlational effect was r = +0.34, each of which represented a medium to strong effect. Moderator effects were examined for the human-related (g = +0.49; r = +0.16) and automation-related (g= +0.53; r= +0.41) factors. However, moderator effects specific to the environmental factors were not calculated because of a lack of presently qualified studies. Submoderating factors were examined and reported for human-related (i.e., age, cognitive factors, emotive factors) and automation-related (i.e., features and capabilities) effects. Analyses were also conducted for type of automated aid: cognitive, control, and perceptual automation aids. Automated cognitive aids provide recommendations to users about the current and potential future states of systems. Automated control aids replace varying levels of human (operator, user) action. Perceptual aids are used to assist the operator or user by providing warnings or to assist with pattern recognition. All three types of aids–cognitive (g= +0.41; r= +0.39), control (g= +0.51; r= +0.12), and perceptual (g= +0.62; r= +0.37)— had a moderate effect on trust development. These findings provide a quantitative picture of human-, automation-, and environment-related factors influencing trust in automation. Findings from this work were used to develop design and training guidelines that are also applicable to HRI. Future research needs were identified based on the results of the foregoing analyses.
Distribution: Approved for public release
  Download Report ( 0.722 MBytes )
If you are visually impaired or need a physical copy of this report, please visit and contact DTIC.
 

Last Update / Reviewed: July 1, 2014