Also, Table 4 shows gsk3b inhibitor the ratios of nonmotorized mode, public transportation, and automobile to all modes. Table 4 Travel characteristics of inside and outside commuters. According to Table 4, three obvious differences are found in the comparison of travel characteristics of the two groups. They are as follows: (1) the mean commuting duration of the former group is slightly shorter than the latter group, and the former group of commuters travels less often

than the latter one. But, in the respect of the commute trip number, commuters in the historic district frequently travel for work. (2) Commuters in the district travel 2.88 times per day, while commuters out of the district travel 2.4 times per day. Similarly, the home-based chains of the former are more than that of the latter one. Observing (1) and (2), we can get the explanation that the commute distance of the inside commuters is shorter, so they have more free time and are more likely to travel. (3) The nonmotorized mode is more popular among the inside commuters for their shorter commute distance and lower travel time. Therefore, the public transportation and the automobile, both of which are suitable for long-distance travel,

take lower shares of the total trips in the historic district. 5. Modeling Results 5.1. SEM Model Specification The aim of this paper is to explore the influence of individual and household attributes on travel characteristics of commuters in the historic district, relating to subsistence activity, trip chain, and travel mode. Based on previous research, individual

and household, participated activities are highly related to travelers’ travel characteristics, and it means that individual and household attributes of travelers can not only directly affect their travel (i.e., number of trips and mode choice), but also indirectly influence it by influencing activities which they participate in. In the paper, a model with 7 endogenous variables and 7 exogenous variables relating to commuters travel characteristics is established to obtain the interrelationships among these variables. Figure 2 illustrates the initial conceptual model structure. Using the initial model framework, we developed two models for the two groups, one for commuters in the historic district, and the other for AV-951 commuters out of the district. The following step is to modify the model. The hypotheses can be adjusted and the model can be retested. The model can be adjusted by adding new pathways or removing the original pathways. The final model is decided by the reported statistics. Figure 2 Model structure in the SEM. SEM can be developed in the statistical package software named AMOS, and the estimation can be efficiently achieved by ML (maximum likelihood estimation).