Thank you very much. Walking distance is actually (fairly) normally distributed.
I actually deliberately coded visits as 0, 1 and 8 on a continuous scale to account for the different time differences between visits. I do not want it to be symmetric as you suggest with 0, 1, 2 as that would...