A Panel-Based Evaluation of the San Diego I-15 Carpool Lanes Project

Abstract

The operation of reversible high-occupancy vehicle (HOV) lanes on Interstate 15 north of San Diego, California, has been monitored by an annual panel survey of commuters in the region and by traffic flow observations. The panel survey, initiated in 1988, collected mode choice, travel time, and attitudinal data in one wave prior to the opening of the lanes and in two waves after the opening of the lanes. These data are used to model the causes of changes in four variables at three points in time: (1) choice of ride-sharing versus solo driving, (2) travel time, (3) perceptions of traffic conditions on the I-15 mixed-flow lanes, and (4) attitudes concerning the HOV lanes. The model involves simultaneous equations with mixed discrete-choice, ordinal-scale and continuous variables, estimated by probit sub-models and distribution-free generalized least squares. An important feature is the use of individual-specific constant terms, which take advantage of repeated measurements on the same individuals to account for population heterogeneity. Results show mutual cause and effects among mode choice, travel times, and attitudes.