The Moritz study is here if anyone is interested:
http://www.bicyclinglife.com/Library/Moritz2.htm
The Moritz study tracks two types of accidents -- "minor" and "serious." Serious is defined as causing more than $50 in damage to a person or equipment. (That doesn't sound necessarily serious to me, but we go to war with the facts we have). Note that the vast majority of these accidents are simple falls -- collisions with motor vehicles only account for 11% of all reported accidents and 24% of serious accidents.
The numbers that Moritz reports -- and Bek echoes -- are for all accidents, minor and serious. Moritz doesn't report rates for serious accidents, and he doesn't make it easy to deduce them from the data he presents, as most of what he gives is percentages, not numbers. However, from the data he presents I was able to deduce that there were 176 serious accidents and 567 total accidents reported. From his percentages I was able to then calculate the number of each type, and from his crash rates I was able to calculate the kilometers traveled for each facility type.
That allowed me to calculate serious accident rate for each facility type. What I got was a very different picture:
Major w/o bike facilities 17.6
Minor w/o bike facilities 17.9
Signed bike route only (BR) 19.9
On-street bike lanes (BL) 16.0
Multiuse trail 24.3
Off road/unpaved 38.9
Other (most often 'sidewalk') 318.4
Now, do these numbers show anything meaningful about relative safety? No.
There were only 7 reported serious accidents in bike lanes. If there had been 8, the accident rate for bike lanes would have been higher than for streets with no facilities. Is 8 significantly different from 7 in a population of this size? With a population of 1956 and 8 observations the standard deviation is 2.8, so we're not even talking half a standard deviation. Not even close to significant.
Now, what is interesting is that bike lanes do have a statistically significant advantage when it comes to
non-serious accidents -- those causing less than $50 in damage. What this mostly reflects, however, is that bike lane accidents are far more likely to be serious. In bike lanes, 64% of the accidents are reported as serious, as opposed to 30% for minor roads and 14% off road. Of course this could all be coincidence since there were only 11 total bike lane accidents reported in the study.
Try again?