And Mavis - The relative numbers of cycles and cars in the Westminster study is irrelevant to the argument - 100% of the accidents studies involved 1 bike and 1 motor vehicle. In 68% of these accidents, the motorist was found to be at fault.
If you can somehow interpret this as showing that cyclists are more dangerous than cars, then I suggest you apply for the Fields Medal as you have clearly invented a whole new branch of statistics, previously unknown to the mathematical world.
So in 68% of accidents, the car driver is at fault. If the ratio of cars to bicycles is exactly 68%, then an individual car driver is as likely to be at fault in an accident as an individual cyclist. (Number of "at fault" accidents divided by number of "vehicles.") If the ratio is much higher, then each individual car driver is much less likely to be involved in an accident than each individual cyclist.
To illustrate, assume 68000 drivers, 32000 cyclists, 100 accidents, car driver at fault 68 times and cyclist 32 times. 68/68000 = 1/1000 car drivers at fault, 32/32000 = 1/1000 cyclists at fault. If in fact there are (say) 84,000 drivers and 16,000 cyclists, the 68/84000 car drivers at fault = 0.08% compared to 32/16000 = 0.20% cyclists, so each cyclist more than twice as likely to be at fault.
(I don't really care about this issue - but I want that medal.)