Using warning systems with adaptable thresholds: Choice of security level, compliance and performance in a simulated computer security task. Issue 125 (May 2019)

• Record Type: Journal Article
• Authors: Wiczorek, Rebecca; Zirk, Anna
• Is Part Of: International journal of human-computer studies. Issue 125(2019)
• Page Start: 32
• Page End: 40
• Publication Date: 2019-05
• Subjects: Alarm systems -- Warnings -- Compliance -- Adaptable automation -- Threshold-setting -- Personality; Human-machine systems -- Periodicals; Systems engineering -- Periodicals; Human engineering -- Periodicals; Human engineering; Human-machine systems; Systems engineering; Periodicals; Electronic journals; 004.019
• DOI: 10.1016/j.ijhcs.2018.12.006
• Languages: English
• ISSNs: 1071-5819

Abstract

Highlights: Warning systems with adaptable thresholds increase trust and compliance. Acceptance of such systems is high and perceived workload low. Adaptable systems raise performance and have the potential to increase security. Users prefer high (i.e. conservative) over low (i.e. liberal) thresholds. Certain personality traits can have an impact on users' choice of threshold. Abstract: The aim of the study was to examine the potential benefits of warning systems with adaptable thresholds regarding user compliance and performance, and whether certain aspects of personality affect users' choice of threshold level. Most warning systems have low thresholds to reduce the number of misses. However, low thresholds produce large numbers of false alarms, which reduce users' trust in the system and lower their compliance with alarms, and thus can harm performance. Adaptable alarm systems that allow users to choose their preferred threshold could be a promising approach to mitigating problems associated with conventional warning systems. Within a simulation environment from the context of computer security, a fixed warning system with a conventional low threshold was compared to an adaptable system. Trust, compliance, performance, workload, and acceptance were measured. Additionally, various personality aspects were assessed, and their potential impact on the choice of threshold level was investigated. Participants working with the adaptable system reported higher levels of trust and acceptance as well as a lower workload. They complied more often with alarms, which led to better performance in comparison with the control group. Financial risk-taking was the only personality aspect related to the choice of threshold. The higher a participant's risk-taking tendency, the higher threshold they chose. Adaptable thresholds have the potential to improve users' interactions with warning systems and thus to increase security. In less safety-related cases such as computer security, adaptable thresholds could be a valuable alternative to the commonly applied fail-safe approach conventionally using low thresholds. Graphical abstract: Image, graphical abstract