Predicting Mode Share

Renne, John L., Hamidi, Shima and Ewing, Reid (2017) "Transit commuting, the network accessibility effect, and the built environment in station areas across the United States", forthcoming in 'Research in Transportation Economics' is a great article. I've taken the liberty of reproducing some of the relevant results here. (I've removed the discussion of significance, which is more relevant to academics rather than practicioners)

At the Neighborhood Level

• A doubling of the average household income across the station area results in a 17.7% decline in transit commuting 
• A doubling of the intensity of jobs and people yields a 17.5% increase in transit commuting. 
• A doubling in the share of nonwhites across the station area yields a 14.4% premium for transit commuting. 
• Light rail/streetscar (vs bus) increases ridership by 22.5%
• Heavy rail/metro increases ridership by 30.6% over a bus. 
• land-use mix is captured by the jobs-population balance index, which indicates that neighborhoods with more balanced land uses exhibit higher shares of transit commuting. A doubling of the values in this index yields a 23.2% increase in the mode share for transit commuting. 
• Doubling the number of 4-way intersections increases transit commuting by 9%
• Doubling the walk-score increases transit commuting by 27.6%
• Doubling in transit service frequency yields an 18.6% bonus for the mode share of transit commuting.
• For a doubling in the share of Hispanic residents, transit commuting decreases by approximately 12%.

I feel that, given these numbers, it should be possible to do a benefit-cost analysis for all of these things....

At the Regional level:

• A doubling in the share of regional jobs and population within a half-mile of all stations in the region, also known as the network effect, yields a 38.6% increase in the mode share of transit commuting in station areas. 
• For each doubling in annual travel delay, the average share of transit commuting increased by 91%-124%,