Abstract
The act of driving a car is a complicated multi-functional task, which gives rise to a broad range of reactions and experiences, both across a sample of drivers and for an individual driver over time. In the attitude-forming process the driver is influenced by a number of factors, these may include their previous driving experience, their exposure to the car under test and of course, the numerous sensory inputs during the driving experience itself. Any rating extracted from the subject is made up of an unknown balance of these factors and the actual rating given may be further influenced by the method of recording. To achieve an acceptable level of accuracy, repeatability and therefore confidence in subjective ratings a test method tailored to whole vehicle subjective assessment must be employed.
This paper reports on three aspects of test technique essential to the detailed analysis of subjective performance. i) The application of a fully mapped electronic throttle system. ii) The analysis of the suitability of different rating scales to meet the requirements of accuracy, repeatability and the ability to record small changes. Three scales are included, these are the:
Absolute Likert, Comparative Likert, and the Semantic differential. iii) To investigate the time taken to form a steady state opinion of a change to vehicle performance.
(attitude formation.)
During assessments performed in normal driving situations, a single vehicle, equipped with an electronic throttle system is used in a series of experiments to prompt attitude changes related to the performance of the test vehicle. This method provides a considerable measure of control over the attitude forming process, allowing the rating methods to be investigated. Finally, some example results are shown from a factorial study of subjective performance to investigate the influence of the level of wide open throttle acceleartion, the throttle progression, and the rate of change of acceleration with engine speed. The results highlight the type of data that can be generated using these robust test techniques.