The Perceived Danger (PD) Scale: Development and Validation

TL;DR

This work defines perceived danger as the anticipation of harmful consequences in robot interactions and fills a gap in HRI measurement by developing a 12-item bifactor PD scale that captures four subdimensions: affective states, physical vulnerability, ominousness, and cognitive readiness. Through four experiments, the authors use exploratory and confirmatory factor analysis to establish a robust bifactor structure, confirm its validity against the Godspeed perceived safety scale, and demonstrate reliability ($\alpha$ up to $0.98$; $\omega_t$ up to $0.99$) and predictive utility for both danger and safety judgments. Experiment 4 further validates the scale in an in-person setting and shows sensitivity to robot speed, aligning with prior findings that faster movement increases perceived danger. The PD scale offers a flexible tool for HRI to assess danger directly, potentially informing safer robot design and deployment across contexts beyond laboratory video stimuli. $\alpha$=0.95, $\omega_t$=0.97 in Experiment 2, and in-person reliability $\alpha$=0.90 illustrate strong psychometric properties across modalities.

Abstract

There are currently no psychometrically valid tools to measure the perceived danger of robots. To fill this gap, we provided a definition of perceived danger and developed and validated a 12-item bifactor scale through four studies. An exploratory factor analysis revealed four subdimensions of perceived danger: affective states, physical vulnerability, ominousness, and cognitive readiness. A confirmatory factor analysis confirmed the bifactor model. We then compared the perceived danger scale to the Godspeed perceived safety scale and found that the perceived danger scale is a better predictor of empirical data. We also validated the scale in an in-person setting and found that the perceived danger scale is sensitive to robot speed manipulations, consistent with previous empirical findings. Results across experiments suggest that the perceived danger scale is reliable, valid, and an adequate predictor of both perceived safety and perceived danger in human-robot interaction contexts.

Figures (2)

• Figure 1: Ordinal regression models for each scale and empirical rankings. Videos were ranked from least (1) to most (3) dangerous. Red diamonds represent the Godspeed PS scale, blue circles represent the scale, and black circles represent empirical data with a 95% CI.
• Figure 2: Ordinal regression models for each scale and empirical rankings. Videos were ranked from least (1) to most (3) safe. Red diamonds represent the Godspeed PS scale, blue circles represent the scale, and black circles represent empirical data with a 95% CI.