Calibrating Trust in Automation
Author Information
Author(s): Christopher Holland, Grace Perry, Heather F. Neyedli
Primary Institution: Dalhousie University
Hypothesis
How do changes in automation reliability affect users' trust and dependence on automated systems?
Conclusion
Trust does not always correlate with system reliability, and dependence behavior follows system reliability changes.
Supporting Evidence
- Participants' dependence behavior followed system reliability changes.
- Trust levels did not significantly change with varying reliability.
- Self-confidence remained stable regardless of automation reliability.
Takeaway
This study looked at how people trust machines that help them with tasks. It found that even if people don't trust the machine more when it gets better, they still depend on it more when it works well.
Methodology
Participants completed a color identification task with an automated aid, measuring trust, self-confidence, and dependence throughout the experiment.
Limitations
The use of MTurk participants may limit generalizability, and the task context may not fully capture real-world complexities.
Participant Demographics
Participants were 139 individuals recruited from Amazon Mechanical Turk, with normal or corrected-to-normal vision.
Statistical Information
P-Value
p=0.05
Statistical Significance
p<0.001
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website