Publication Detail
Challenges to the Design of Driving Simulator Experiments
|
UCD-ITS-RP-95-48 Journal Article Download PDF |
Suggested Citation:
Jovanis, Paul P. (1995) Challenges to the Design of Driving Simulator Experiments. Institute of Transportation Studies, University of California, Davis, Journal Article UCD-ITS-RP-95-48
Proceedings of Driving Simulation Conference, Sophia Antipolis, France
In every study referenced in this paper, ANOVA or regression was used to study a single dependent variable. Several studies used more than one measure, but the analysis of each was separate.
It is the author's contention that this comes, in part, from the strong experimental training that is provided to most researchers in the field. Control of experiments is emphasized. Right independent variables are developed and used in the experimental design with a simple dependent measure. Simulator studies will be much more effective, useful, and powerful if multiple measures are linked.
Two examples of linked dependent measures are given in the paper. The dependency of speed and headway, (or speed and reaction time), eye movement, and lateral position. Statistical procedures such as variable instrumentation have been used in survey data and are directly applicable here. What is more difficult is to design experiments to take advantage of these methods. This requires training of the experimenters and some courage in conducting more flexible experiments.
In every study referenced in this paper, ANOVA or regression was used to study a single dependent variable. Several studies used more than one measure, but the analysis of each was separate.
It is the author's contention that this comes, in part, from the strong experimental training that is provided to most researchers in the field. Control of experiments is emphasized. Right independent variables are developed and used in the experimental design with a simple dependent measure. Simulator studies will be much more effective, useful, and powerful if multiple measures are linked.
Two examples of linked dependent measures are given in the paper. The dependency of speed and headway, (or speed and reaction time), eye movement, and lateral position. Statistical procedures such as variable instrumentation have been used in survey data and are directly applicable here. What is more difficult is to design experiments to take advantage of these methods. This requires training of the experimenters and some courage in conducting more flexible experiments.