Is Mode Share Related to Walkability?

Based on a suggestions from a friend, in this post I will connect my two previous posts and evaluate if people who live in more walkable areas actually do in fact walk more? I should begin by saying there have been many comprehensive studies published that clearly establish this relationship, see for example Active Transportation Benefits of Walkable Approaches to Community Design in BC  produced by Dr. Larry Frank and his team at the UBC Health and Community Design Lab or the work of Walkonomics and Walkscore. However, my goal in this project is to test this relationship using two new spatial data sets that have not been  previously linked.  I conducted this analysis by spatially joining the home postal code locations for 3,657 BC Commuter Challenge participants to my WalkMap data set for the Metro Vancouver area. These data are shown on the map below:

To provide I grouped the WalkMap scores for each home location into five quintiles with about 720 participants in each category. Finally, I summarized the number of trips by mode within each category as shown below - it is important to emphasize that this breakdown is not representative sample of the overall mode share for Metro Vancouver because these data are based on trips during the week-long BC Commuter Challenge.

Mode Share by Walkability Quintile          

I was happy to confirm that people who live in more walkable areas did in fact walk more. However, I was somewhat surprised to see that cycling and transit mode share was nearly constant between scenarios. This brings up the issue of whether I should add a transit component score to improve the practicality of my WalkMap model. What do you think: should a measure of access to transit be included in my WalkMap model or should this be a separate TransitMap-type score?