The Freakonomics blog had a post last week on "Who causes cyclists' deaths". They mis-cite a study out of Toronto as concluding that in 90% of cyclist fatalities, the driver is primarily at fault. This isn't correct. First of all, the study deals with collisions, not fatalities. The summary report lists the top 10 causes which constitute ~80% of all the collisions. Of those more than 90% of the collisions are driver labeled. But the full report lists all causes. Driver labeled incidents account for 75% of collisions and cyclist labeled ones account for 22.3% with the rest (4.3%) being unknown. In addition these labels are not meant to assign blame. Not all doorings, for example, are the result of driver inattention.
The 90% number comes from several studies conducted by Charles Komanoff and members of the Right of Way organization in New York that concluded that cyclists and pedestrians were strictly culpable for less than 10 per cent of bike/pedestrian-car accidents. And I suspect the pedestrian numbers swamp the cycling ones. According to a cyclist only report from the same organization, drivers are highly culpable between 66% of the time (if DWI and driving without a license always makes the driver culpable) and 44% (if they ignore those factors and accept driver's uncorroborated statements). They are not culpable between 15% and 40% of the time respectively.
An Orlando study found very different results. In that study cyclist traffic violations were involved in 39.1% of all crashes and the driver was at fault in 16.7%. But I suspect in bike-unfriendly cities you get a much higher percentage of "cyclists by necessity" which includes poorly trained cyclists, people who've lost their license due to drunk driving (about 9% of the crashes involved alcohol or drug use by cyclists) and people who can't afford lights and reflectors.
Still, the NYC study does seem to state that bad driving, not bad cycling is more often to blame in an urban environment. That doesn't really help us much. Most cyclists I talk to already suspect that. What is important to learn from these studies is that biking sober, biking with lights and reflectors at night, biking with traffic - even on the sidewalk and signaling can help you to avoid a collision.
But it is good ammunition when the "cyclists are scofflaws" talk comes up.
The Freakonomics blog also wrote about the desire of one Oregon lawmaker to require cyclists to pay $54 every two years to help pay for the roads they use. He's under the misconception that cyclists are free riders. Commenters do a good job of shooting that down (road funding comes from many additional sources than just automotive taxes, pedestrians don't pay for sidewalks, cars do much more damage and need much more infrastructure, etc...)