Originally Posted by acidfast7_2

I'd argue that the difference is that cycling to work requires a different skill set than every other commuting type, including pedestrian commuting, and that cohort is by default more healthy (which is nearly impossible to control for).

Thus, instead of phrasing that argument that those that bike commute have a "47% less chance of mortality," I'd argue that those with lower mortality are more likely to bike commute.

If the study broke all participants into deciles by VO2 max, we'd likely see more bicycle commuters in the more-fit deciles.

I guess that it's the classic correlation not causation argument based on the way the data was collected for the longitudinal study.

I also doubt if someone became a bike commuter that their mortality would significantly go down, but that's a different study entirely with an even smaller sample size.