Like many or most here, I'm convinced that properly done, bicycling is a very safe activity. I'm convinced in part by my long history of mining data on the issue. I dislike any effort to portray cycling as unusually dangerous, or requiring unusual measures to be acceptably safe.
I'm also skeptical of either the need for, or the safety benefits of, most segregated bike infrastructure. Sure, I can enjoy a nice linear park (AKA rail trail), especially if it's empty of other users. But I know about the many traffic complexities introduced by on-road or parallel bike infrastructure.
But here's my question: What's the honest consensus here on relative risk for riders on a street with a bike lane vs. one without? What's the best data that we have? I'd love actual data, especially on a per-mile-traveled basis - ideally, data comparing streets with facilities to streets without. And in addition to risk ratios, which seem to be the default reports ("20 percent more dangerous!!") I'd love to see actual values, since 20% (or whatever) of a minuscule value is still minuscule.
I'm aware of efforts to exaggerate benefits, e.g. the Lusk paper that purported to find significant benefits to cycletracks in Montreal. That was disputed by others who noted that the "case" vs. "control" streets were very, very different. I'm also aware of papers from Denmark that compared crash data before and after installation of bike lanes and/or cycletracks. Those papers found increased crashes with facilities.
Can anyone point to the best date now available?