ABSTRACT
Mislevy´s evidence centered design model for assessment was used in order to develop a procedure by which driving task scenarios can be derived as a basis for developmentally tailored driving assessments. Borrowing from recent theories on driving and driving errors, task environment attributes were defined which may complicate the sub processes of driving and thus may result in varying task difficulty. A universe of assessment tasks was defined by combining basic driving tasks and critical task environment attributes. A collection of 55 critical driving task scenarios was selected from 39 video recorded driving lessons, throughout different stages of driving education. Results of a difficulty rating study pertaining to these scenarios including experienced driving instructors show that the scenarios discriminate well between beginning and advanced learner drivers. A contrasting comparison between easy and difficult scenarios revealed that the degree of time pressure, the number of necessary maneuvers to be carried out consciously in a given time-span may have been a discriminating factor in addition to the specified task environment attributes.
Successful scenario solution can be predicted by using an IRT function, where solution probability is a function of driver ability and task difficulty. Implications for assessment design activities are discussed.
