The interaction between relative vehicle secondary safety and driver demographics.

Vehicle safety rating programs based on crash tests have been criticised that they use crash test dummies that represented a limited range of body shapes. Therefore, such tests may not be addressing the safety needs of occupants whose body shapes do not correspond to that of the test dummies. Similarly, vehicle safety ratings programs based on real world crash data, such as the Used Car Safety Ratings (UCSR) system (Newstead et al, 2006), may not be applicable to occupants whose body shapes do not do not match the body shape of the profile of drivers to which the ratings are standardised. The data compiled for the Used Car Safety Ratings project (Newstead et al, 2006) was used in this project to identify and quantify the difference in relative vehicle market group safety depending on driver age and gender. This was done by introducing terms representing interaction between market group and age and sex of driver into the logistic regression models used to produce the Used Car Safety Ratings of Newstead et al (2006). Data limitations meant that the analysis was limited to studying how driver age and sex affect safety ratings for different market groups and not different models of vehicles. It was found that even though adding the interaction terms to the logistic models significantly affected the models, the addition of these terms did not drastically alter the relative crashworthiness rankings for market groups between different demographic groups. With a few exceptions, types of vehicles that Newstead et al (2006) ranked as being safe for a standardised set of drivers were in general safe for drivers grouped by their age and sex. The same can be said for types of vehicles that Newstead et al (2006) rated as performing poorly for a standardised set of drivers. This suggests that the vehicle safety ratings published by Newstead et al (2006) are on the whole applicable for drivers of all demographics. Incidences where market rankings did vary depending on driver demographics are identified and directions for further research have been suggested. (Author/publisher)