The Measurement of the Propensity to Trust Technology

Abstract

In recent years, there has been a focus on not just how people work with automation, but how humans interact with and rely on automation in human- automation teams. Few studies have examined how propensity to trust in auto- mation influences trust behaviors. Of the published studies, there are inconsist- encies in how propensity to trust automation is conceptualized and thus meas- ured. Research on attitudes and intentions has discerned that reliability and va- lidity of measures can be increased by using language that is more specific for a particular context, which reduces respondent ambiguity and increases the ability to predict behavior. This study examined traditional measures of propensity to trust automation, and whether adapting measures could enhance our ability to predict beliefs about automation trustworthiness (perceived trustworthiness) and behaving in a trusting manner when interacting with automation (behavioral trust). Participants (N = 55) completed three propensity to trust in automation surveys including Propensity to Trust Technology, an adapted propensity meas- ure, and the Complacency-Potential Rating Scale. Participants played a modified investor/dictator game, where people thought they were teaming with a NAO ro- bot. This study demonstrated that compared to a more generally-worded meas- ure, the context-specific measure of propensity to trust automation was more re- liable and better predicted perceived trustworthiness and behavioral trust. Fur- thermore, the adapted measure was the only significant predictor of both beliefs about the trustworthiness of the automation and the actual trusting behaviors of participants. By decreasing the ambiguity of measures of propensity to trust au- tomation, the reliability and predictive validity are increased.