To avoid bias caused by different levels of information, the participants were known to neither of the interviewers or to both. In addition to the disability and confidence ratings, so- ciodemographic and stroke-related data were recorded for each participant