from your source:
"In 62% of cases (64 of 104) in which the cyclist collided with a vehicle (defined as a motor vehicle, streetcar or train), one or more modifiable actions on the part of the driver were identified which were felt to have contributed to the death."
those modifiable actions include speeding. the source you use finds causality attached, but you don't. lol, nice cherry-picking. dismiss it all you want, but if one can arbitrarily discount findings then what value does your source actually have? apparently none, which means your argument has the same value.
. . .good work
Speed generally doesn't cause incidents. It results in an inability to avoid an incident. That invalidates 30% of the data, outright. For this reason my position is that the cyclists' behaviour is what should be modified.
And I've already stated what I think of the source based on the recommendations, which are not actually supported in the data.