Safety at Signalized Intersection with Permitted Left Turns; Another One Way vs. Two Way Debate Item

I got a call today from a colleague who was working for Vancouver BC. They were doing research on safety of pedestrian and bicycle travel and one of the things that BC has experienced is a higher crash rate of pedestrians and left turns than cities such as Portland.
There other peer cities were Toronto, Calgary, Seattle, and San Francisco.

The answer to this was easy for me. Downtown Vancouver BC has quite a few two way streets (one is shown in the picture to the left) which results in permissive operations. When operating a motor vehicle and navigating across two lanes of approaching traffic and pedestrians from both directions on the crossings it is difficult to gauge the relative safety of accepting a gap. This is also confounded when we throw people on bicycles into this where vehicles may not see the oncoming person (low light visibility, etc).

Which brings me back to a post I had on twitter awhile back, with some research that supported this perspective. It read as follows:

"It was clear from the results that converting from protected phasing to FYA operation (third scenario) leads to a dramatic increase in left-turn crashes"
http://www.trb.org/Main/Blurbs/Evaluation_of_Safety_Strategies_at_Signalized_Inte_165938.aspx

The research suggests increasing the yellow and all red times at the intersections (not a good idea in my opinion), but also states the following:

Specifically, the analysis indicated a statistically significant reduction (at the 0.05 level) in 34 total crashes as a result of (1) increasing the all red phase only and (2) increasing the total change interval to be less than the ITE recommended practice. Injury crashes were significantly reduced as a result of increasing the total change interval to be less than the ITE recommended
practice. Rear-end crashes were significantly reduced as a result of increasing the total change interval to be greater than the ITE recommended practice. The change in angle crashes was statistically insignificant under all scenarios investigated.
Table 6.3 also has the caveat that " The sample of sites used in this evaluation is limited. So these results should be used with due caution."