There are a few hundreds of institutes worldwide that do research into driver behaviour on a regular basis. A large number of universities have faculties in mechanical engineering, human factors research or experimental psychology where driver behaviour research is being done. For experimental research, a research driving simulator has a lot advantages.
Driving in the real world is a complex task and it’s very difficult to control that environment. For human behaviour experiments, control of the circumstances is of utmost importance. When you want to test a hypothesis, the best experimental design is a within-subjects design. This means that each test subject performs a number of tests where a condition is manipulated between the tests. The differences on a number of dependent variables are tested as measured during the different test conditions. Because each subject received the same tests, differences in results betwen test conditions are tested within the same subjects, therefore the name within-subjects design. The most important reason for this design is the reduction of unexplained variance. In any experiment, you want to keep the things that are not relevant for the experiment the same, because irrelevant variation increases the unexplained varianed. The ratio explained variance divided by unexplained variance must be as large as possible in order to find significant results.
The explained variance increases when the difference on a number of dependent variables between test conditions is larger, while the power of the results increses when unexplained variance decreases.
In a driving simulator, the noise in the data, and thus unexplaned variance, can be reduced a lot compared to testing in the real world, because the test conditions can be controlled completely in a car simulator. Traffic can be controlled, the weather can be controlled, the virtual environment can be designed precisely as the researcher wants it. The experimental manipulations are fully under the experimenter’s control. If done the right way, this all helps to improve your experimental design and get better results with your experiments.