Quote:
Originally Posted by chiefneil
10x more likely per what measure? Per vehicle mile? Per reported accident?
Without knowing the measure and methodology it's impossible to form a logical conclusion. If there are 10x more BMW's than Toyotas in Sweden, then it would be reasonable to expect 10x more accidents involving BMW's. You need to normalize the results for the distribution of vehicles and control for variables like driver age and location to get a useful result.
The chart may well be normalized, but unless it explicitly says so then interpretations are just based on assumptions.
|
I agree that what you are saying is the correct way to do it. Considering this came from an insurance study, they must have done the normalizations.