Actually average effective speed is much closer then that. I forget the actual, but I think it on the order of 1.5 to 2x... We speed a heck of a lot of time stuck in traffic. I'm sure someone in commuting forum can get you the actual numbers with references.

But even though my google-fu is quite weak. Here's a reference with citations.
http://www.ohiobike.org/misc/CyclingIsSafeTLK.pdf

It seems that the activity's decrease in mortality vastly offsets the crash risk.

And this where the bicycle fatalities there are, well they heavily skew to people NOT on this board. Sidewalk riders, Salmon, Ninja Salmon, and drunks make up the vast majority of fatalities. Heck, just drunks alone are almost a majority by themselves...
So DO YOU DRINK AND RIDE?
Sidewalk surf?
Are you a Salmon?
or a Ninja?

We have many built-in errors in the way we asses risk. Think of it as flying from NY to LA... Far more like to die in car on the short ride to the airport then the long flight by many orders of magnitude. However, we vastly over estimate the risk of flying for many reasons.