Carl Westin

14 Introduction Factors affecting acceptance. Results from the study in Chapter 3 not only indicated that the advisory conformance played a significant role in the acceptance of those advisories, but also that controllers sometimes rejected their own conformal advisories (i.e., their own solutions). To investigate this, the following three chapters detail follow-up studies investigating three research questions associated to different factors that may interact with conformance and affect acceptance. The three studies are: • Source bias (Chapter 4): To what extent are controllers more biased against advice from a machine than from a human? Research has shown that peo- ples’ trust and reliance behavior varies with the (perceived) source (human or machine) they are interacting with. • Automation transparency (Chapter 5): To what extent were rejections driven by a lack of understanding conformal advisories? Research has indicated that automation transparency is a critical quality of automation for facilitating understanding of its behavior. • Conflict solving consistency (Chapter 6): To what extent are controllers inter- nally consistent in their resolution strategies over time? Internal consistency is a requirement for strategic conformance in that a controller’s conflict solu- tions need to be stable over time. Note that both the source bias and automation transparency research questions were empirically investigated together with strategic conformance. For this purpose, two separate real-time simulations were conducted. The consistency research ques- tion, however, was addressed by post hoc analysis of prequel data and controllers’ solutions to repeated conflicts. Thus, the consistency study used prequel data from the previous three real-time simulations (reported in Chapter 3 through 5). Chapter 4: Source bias effects. Chapter 4 investigates how strategic confor- mance, together with the perceived source of an advisory (human or automation) affects the acceptance of such advice. Automation trust research is reviewed to ex- plore whether people have a dispositional bias against the use of automated decision aids and to what extent, if any, strategic conformance can mitigate the negative ef- fects of such bias. To empirically investigate this, advisory source and advisory conformance were varied in a human-in-the-loop simulation involving experienced controllers. Chapter 5: Automation transparency effects. Chapter 5 investigates strate- gic conformance effects on advisory acceptance in light of the automation’s trans- parency. Transparency was manipulated by means of varying the amount of meta-