46 Transportation Research Record: Journal of the Transportation Research Board, No. 2606, 2017, pp. 46–53. http://dx.doi.org/10.3141/2606-06 An analysis of factors associated with nonwork trips was conducted. The key variable of interest was proximity to transit-oriented develop- ment areas around train stations in New Jersey. The data were derived from a questionnaire of households around eight train stations and stratified by distance from a station. The focus was on the mode used for trips to restaurants–coffee shops and grocery–food stores, two of the most common and frequent nonwork trips. The analysis focused on the choice of walking versus driving for these trips by using an instru- mental variable probit model aimed at controlling for the endogeneity associated with self-selection bias (i.e., people who prefer not to drive choose more walkable neighborhoods). Results showed strong asso- ciations between proximity to transit-oriented development areas and walking trips, and these results hold when endogeneity is controlled for. Local street density and per capita vehicle ownership also had the expected effects on walking trips. New Jersey began the transit village initiative in 1999 with the aim of encouraging municipalities with train stations to focus new devel- opment around and near their station. As of 2016, 32 municipalities in the state have been designated as transit villages (1). Previous research has evaluated a number of benefits associated with transit- oriented development (TOD) in New Jersey (2). (“Transit-oriented development” can refer to a specific development located near a train station or, more broadly, to the overall development surround- ing a station area. The broader definition is used in this paper.) This research included evaluation of resident satisfaction (3), property value impacts (4), and travel behavior (5). In general, these studies found that residents are generally happy with the development that has occurred, property valuations increase closer to transit stations, and those who live in closer proximity to TOD and transit tend to walk more, use more transit, and drive less compared with those more distant. While TOD is often focused on transit and is seen as a way to boost transit usage, the physical features surrounding a train station may be more effective at encouraging walk trips. While a TOD may increase residential density and thus increase the proximity of activities, other features may be more important. For example, better pedestrian infrastructure makes walking safer and more pleasant while simultaneously reducing the mobility of vehicles. Many of the transit villages in New Jersey have combined new development with additional pedestrian features in their community. Most studies have also typically examined commute trips. One study used a previous survey of TOD areas in New Jersey and found that those living closer to train stations made fewer grocery shopping trips (6). This paper also examines nonwork travel modes, specifi- cally trips for grocery shopping and to restaurants or cafes. The focus is on how these trips are affected by residence location while control- ling for other features of the TOD area. The modeling approach used also provides a way to control for endogeneity (i.e., self-selection of residents into a walkable area). PREVIOUS RESEARCH AND THEORETICAL FRAMEWORK Previous research has shown that one of the major achievements of TODs is a reduction in driving while increasing public transportation use, walking, and bicycling relative to other locations. A comprehen- sive review of the literature found that proximity to transit is associ- ated with more transit use as well as various characteristics of how the streets are designed, consistent with TOD characteristics (7 ). Compact and mixed-use neighborhoods allow residents and workers to get around without a car by maximizing their access to work and other destinations. Increased transit ridership and active transportation are both per- ceived as a benefit to society, as reducing auto usage helps to relieve traffic congestion, improve air quality, and reduce tailpipe emis- sions. Yet such benefits accumulate only if TODs manage to affect travel behavior and if people who formerly drove switch to transit or nonmotorized modes. Evidence indicates that TOD effects on travel behavior can be at least partially explained by residential self- selection. Individuals with an affinity for a certain mode of travel often choose a residential location that enables them to travel as much as possible with their preferred travel mode (8–10). If this is the case, then residential self-selection could potentially bias statistical results. Several studies have found that, in general, self-selection bias plays a small role in influencing an individual’s travel decisions and is likely not a major problem in most statistical analyses, but any analyses should seek to control for it, as best as possible (11, 12). A potential impact on travel behavior is the effect of TODs on vehicle miles traveled (VMT). One study found that mixed-use developments, higher residential and employment density, and pedestrian-oriented designs decreased VMT in Baltimore, Maryland; Washington, D.C.; Seattle, Washington; and Virginia (13). The strongest predictor was increased residential density, which led to a 20% VMT reduction in Washington, D.C., compared with previ- ous levels. An analysis of trips generated from 17 TODs (specific developments, not areas) found that on average 44% fewer vehicle Nonwork Trips Associated with Transit-Oriented Development Maria Luz Laham and Robert B. Noland Alan M. Voorhees Transportation Center, Edward J. Bloustein School of Plan- ning and Public Policy, Rutgers University, The State University of New Jersey, 33 Livingston Avenue, New Brunswick, NJ 08901. Corresponding author: R. B. Noland, rnoland@rutgers.edu.