The following was put to me recently: How would you go about researching which areas/cities have the worst drivers?
Here's my take, note that I'm not really answering the question, and I'm sure Justin can improve upon it in half the words in the comment section:
1. It's tough to capture bad driving because many things that we consider to constitute "bad driving" aren't captured in any sort of statistics. I've yet to see the state-by-state time series of "number of instances of cutting someone off."
2. Surveys wouldn't work; I'd expect a similar result to 70%-of-people-consider-themselves-better-than-average-drivers. People systematically underestimate the quality of drivers around them. I am Exhibit 1A for that one.
3. Bad driving outcomes could very well be a function of mixing different types of drivers. Let's assume we've got two types of drivers, aggressive and passive. A city with nothing but aggressive drivers may well have more reported accidents than a city with nothing but passive drivers, but a city with a mix of the two may have the most of all. To that end, quantifying the quality of driver is only useful insofar that you can define the relationship between a spectrum of drivers operating within a driving environment.
4. With regards to age, I will say that more younger drivers (read: first few years of driving) make for more dangerous driving situations, if only because experience matters. It's unclear to me at this point how the upper end of the age spectrum looks.
What else is there?
Here's a post from the summer on courteous driving.