English title: The assessment of safety in actual driving developed on Accompanied Driving. In modern cars, driving and operating data are automatically acquired and transmitted to effectors and processors by bus systems (exempli gratia CAN – Controller Area Network). By reading out these data and saving them on an on-board storage device, the car becomes an experimental vehicle for test runs. Measures of driving safety can then be derived from these data. The project at hand aimed at standardizing the use of such an experimental vehicle to assess the driving safety of participants of the program “Accompanied Driving”. The project consisted of three parts: (1) selection and documentation of an appropriate course on public roads, (2) collection of reference data for 40 subjects, and (3) construction of a psychometric driving-safety test. In Part 1, a 78 km course was selected. It covered all three major road types: 15 km (28 min) on urban road, 24 km (25 min) on rural road and 39 km (23 min) on highways. Furthermore, the course was selected to cover a predefined set of driving tasks (e.g left turn on rural roads). The detailed documentation of the course comprised speed limits and road type. In Part 2, experimental runs of 40 subjects provided a reference data set. The sample consisted of two subsamples: 26 younger subjects (aged between 18 and 22 years), 17 of which with and 9 without former “Accompanied Driving”; 14 older subjects (aged between 23 and 50 years), all without experience in “Accompanied Driving”. In Part 3, a psychometric test for traffic safety was developed. Eleven variables (exempli gratia, ‘violation of speed limit’, ‘heading time’) were recorded electronically. Simple statistical calculations of these variables do not suffice as psychometric measures of driving safety. In earlier projects, the authors have developed the system I-TSA (INVENT – Traffic Safety Assessment; W. R. GLASER, WASCHULEWSKI, and SCHMID, 2005). It was taken up, improved and designed more user centred. Through factor analysis, eight scales were found as components of driving safety. They were normalized and standardized by frequency transformation at a mean of 100 and a standard deviation of 10. The scales were ‘1 speed’, ‘2 acceleration’, ‘3 heading time’, ‘4 lane keeping’, ‘5 turn indicator’, ‘6 steering-wheel movements’ (steering entropy), ‘7 mental effort’ (RSME = Rating Scale of Mental Effort, German version by EILERS, NACHREINER, and HÄNECKE, 1986), and ‘8 electronically recorded driving errors’. The electronic measures were complemented by paper-pencil tests for the subjects and the experimenter to compare conventional safety and electronic data. The subjects completed a questionnaire for driving experience which contained 47 items, among them all 20 items of the technology/risk questionnaire by BR. FÄRBER and BE. FÄRBER (2003). By factor analysis, five scales were extracted: ’timidity, fear of overload or excessive demand‘, ’joy of driving‘, ’enthusiasm for technology‘, ’self-confidence as driver‘, and ’observing of norms and rules’. After each driving test, the experimenter completed a form with three rating scales for driver’s behaviour: First, he rated the impairment of driving by 33 items of an error list. These ratings were condensed to a single mean. Second, he rated driving safety of 9 more general driving habits (exempli gratia choice of speed or surpassing). A factor analysis of the 9 general habits provided the two factors “errors” and “lapses” according to the DBQ (Driver Behaviour Questionnaire; REASON, MANSTEAD, STRADLING, BAXTER, and CAMPBELL, 1990). Third, he characterized the driving style by a semantic differential with 11 adjective pairs (exempli gratia considerate – inconsiderate or aggressive – defensive). By factor analysis, the semantic differential yielded two scales, ‘stress’ and ‘adjustment’. Thus, there were 8 I-TSA, 5 subject’s driving-experience, and 5 experimenter scales. Together, these 18 scales form a standardized test of driving safety, which provides a profile for each subject. Most of these scales are highly reliable with Cronbach’s ? > 0.8. The factor analysis of the 18 scales explains 75.8 % of the total variance by two strong and four moderate factors. Thus, the test measures a construct of driving safety with 6 dimensions. Factor 1 comprises all driving errors as measured or rated. Factor 2 represents the driving-experience variables and mental effort. Factor 3 contains the electronically recorded variables of driving dynamics I-TSA 1 to I-TSA 3. The residual 4 variables load on the additional 3 minor factors. There was no experimental design for differences of groups. However, a statistical analysis as an ex-post-facto experiment was possible. Although it showed no difference between the subgroups for sex and age, a logistic regression correctly assigned the subjects to the groups with and without Accompanied Driving. A multivariate F-test yielded single sided significance between these groups. (Author/publisher)
Samenvatting