The point is you're using irrelevant statistics to "prove" your argument. If you want to compare the relative risks of head injury for walking and for cycling, you shouldn't be using statistics that include irrelevant injuries (stubbed toes, for instance) or dissimilar sample sizes.