I think one part of this discussion is related to the cars seeing you when you are in the cars line of vision in the same lane. As opposed to being over to the side and the driver not noticing you and keeping right on going without slowing. Somewhat similar to moving right, into a space between parked cars, thus disappearing. Then swerving out into the lane and the driver is now surprised and has to react. If you keep going straight without moving over the drivers see you early on, and are not surprised, and have more time to notice you and decide what to do.

I do feel that having the car notice you early, and being predictable (going straight for example) gives confidence to the driver and makes it safer for the cyclist. I think this part of VC is right on the money. You have to imagine the drivers point of view. Make it easy for him, that's safer for you.

I ride with a "Take a look" mirror all the time. I can see ahead and behind almost at the same time. I am usually scanning back and forth, especially when I hear something approaching. I see cars coming from behind a long way away. I see almost all of them. I'm never surprised by an oncoming vehicle from behind. When I am in the lane they do slow down when they approach. They move over and accelerate past also. As to exactly what this does to the odds of being safe as a cyclist I am not sure. But the cars do slow and move over.

I ride a century every week until it gets below freezing plus about another 75 miles on another day in the week. I often get home at midnight from a century. I use a blinding taillight, much brighter than any car taillight, and the same for the headlight.
I find at night I am noticed very far away and the cars approach and pass with even more care than during the day. I assume this is because they don't always know what I am, the light seems to bright to be a bike. But they slow way down if the lane is narrow and give me a huge clearance when passing.

I repeat I don't know if it's safer, but it sure does happen.
But all I know is what happens to me. That is NOT statistically important.

I feel that one persons experience, or even a study in just one town or location is NOT enough of a base to establish statistics as to what is better for traffic rules or behavior. To convince me It would have to be a wide geographic study in varying locations, urban, suburban Northeast, Southwest etc. I have no idea if this has been widely studied across the country or not. What I experience is too small of a sampling for any kind of a conclusion. Same for any one rider. Maybe in a huge city this is the exception? That's not where I typically ride.