The concept of a "walk score," a metric for how easy it is to not use a car to do things, is fun. But so far as I can tell, the walk score presented on the web site is somewhat arbitrary; I have no idea how it was calibrated (although they do say "Street Smart Walk Score gives more weight to amenities that are highly correlated with walking. In addition, multiple amenities in each category count towards your score—for example, we count 10 restaurants to reflect the depth of choice that walkable neighborhoods offer.")
What I can say is this: my house in Bethesda received a much lower walk score than my house in Pasadena, and yet I almost never used my car during the week in Bethesda, because I could walk the .85 miles to the Red Line metro stop to get to work, and because for me driving in Washington was a much worse option than driving in LA (believe it or not!). Both places are comparable in terms to walking to amenities in the evening.
So if we are going to do walk scores, we need to look at how often people in different neighborhoods, well, walk. I am guessing my USC colleagues Gen Giuliano, Lisa Schweitzer and Peter Gordon, have done some work along these lines, and there seems to be a trove of data at Minnesota.