Too early. "We need at least 12 months of collision reports after the work ended, and police reports arrive months late and keep changing for a year. Until then, any change we could show would be too small to tell apart from normal ups and downs."

No measurable change. "Cyclist collisions reported to police on this street are within the range we would expect if nothing had changed, compared with similar streets. Our best estimate is a change of X%, but the true change could be anywhere from Y% to Z%. This does not prove there was no effect. It means we cannot measure one yet."

A change. "Cyclist collisions reported to police on this street went up (or down) by about X% compared with similar streets that did not change. The true change could be anywhere from Y% to Z%. This shows what happened on the street, not why."