Carl Westin

68 Source bias effects The majority of studies considered in the source bias literature review have re- lied on samples of students for their experiment. Although much previous research has involved simple tasks that do not require much familiarization and training, samples can in general be considered to represent novices. Few studies have used experienced participants conducting tasks in contexts with which they are familiar. This is notable since trust is considered a comparison process between one’s self- confidence and the perceived trust in an aid. As such, highly trained, skilled, and experienced professionals would be expected to have high self-confidence which may influence reliance on an external source differently from novices. In addition, experienced people may have lower expectations from an aid. This study investigate source bias effects on trust and acceptance in relation to skilled and experienced professionals in their own field of expertise. Perceptions of trust and advisory acceptance are investigated in light of advisory conformance that is either conformal or nonconformal to a controller’s own conflict-solution style. It is hypothesized that trust will be higher for the automated source, measured after the simulations. Furthermore, it is expected that the acceptance of and agreement with advisories will be higher for the automated source, particularly for conformal advi- sories. Finally, it is expected that conformal advisories will benefit response time (i.e., faster response times), particularly when provided by the automated source. 4-3 Method 4-3-1 Participants The sample consisted of five experienced terminal approach controllers. Four were working at the ¨Ostg¨ota Terminal Control Center in Norrk¨oping, Sweden, and one at the Air Traffic Control Center (ATCC) in Arlanda, Sweden. One female and four males participated in the study. Age varied between 26 and 47 (mean = 32.8 years) and experience between 15 months and 24 years (mean = 8.7 years). Participation was voluntary and no remuneration was offered. Simulation data was partially lost for one participant due to technical issues. As a consequence, all simulation data is based on four participants, except the accept/reject count of advisories that was recorded manually. However, questionnaire responses from all participants were retained. 4-3-2 Simulator A Java-based ATC simulator from the initial study was used. It ran on a laptop connected to an external 23” TFT monitor with a resolution of 1600x1200 pixels. Aircraft plots updated every second to simulate a 1 Hz radar frequency. The orig-