A corridor specific regression model of modal split behavior in Columbus, Ohio.

The relationship between modal choice for the work trip and a series of socioeconomic and spatial variables is examined using census data for the Broad and High Street corridors in Columbus. The model distinguishes between passenger (bus and car-passenger) and non-passenger modes. 7 variables account for some 80% of variation in modal split with race, population density, and automobile ownership accounting for over 70%. High levels of multicollinearity lead to an underestimation of the influence of income and other variables on modal split. The Kain Model and a principal components regression model are employed to gain further insights into these relationships. -Author