I have put the conclusion of my reading first in case you don't have the time to read through all of this.
Here's the bottom line-- I think that even bike lanes that are well designed often create more problems than they solve. Thus, at this point, given the dozens of research articles and reports I have read from around the world, I strongly recommend the development of wide outside lanes with effective road signs and education first rather than blindly creating separate bike lanes in most cases.
However, there are some well-designed bike lanes that actually solve safety problems for cyclists. Certain traffic situations are much improved when a bike lane is thoughtfully integrated into the traffic planning. The important words here are 'thoughtfully" and integrated. High quality facilities should be encouraged.
Here's a paradox- if a road is wide enough to add a bike lane without subtracting any regular motor vehicle lanes, the road probably doesn't need a bike lane. Example: There is a road in Memphis
Would you want to ride on a highway in your car if it were designed by someone who didn't understand cars? That's one of the major problems with bike lanes-- many bike lanes are designed by planners who do not ride bikes and who do not understand the complexities of bicycling in urban traffic. Just lay down a stripe within the legal limits and it is done-- a bike lane! It is a lot more complicated than that. So I am strongly against blindly rushing to paint stripes on roads when it will do no real good and might even make things worse.