Unless they make the same people bike in a randomly decided way during the study, it's likely selection bias to a large extent - more experienced cyclists would ride on roads, less experienced (and on worse bikes without disc brakes) - on sidewalks.