Actually, the objective of employing the SEM in the study is to estimate the whole set

of coefficients contained in the above 8 matrixes, which could be set as a fixed one or Bortezomib a free one. A complete SEM consists of 8 coefficient matrixes: Λy, Λx, Β, Γ, Φ, Ψ, Θε, and Θδ. Γ is the covariance matrix of latent variable ξ, Ψ is the covariance matrix of error term ζ, Θε and Θδ are covariance matrixes of ε and δ. If the assumptions we made are held true, the population covariance matrix will be equal to the sample covariance matrix. Thus, both of the variance and covariance of observed variables (i.e., the indexes of the endogenous variables and exogenous variables) are the parameter functions of the model. Several methods can be used for estimation in SEM. The most common methods for estimation are generalized least square (GLS) and maximum likelihood (ML). The evaluation of SEM

is to examine if the model is a good fit to the data. The constantly used measure is the Chi-Square test, the value of which is calculated by fitting function. As the value of the Chi-Square test is always changing with the sample size, some researchers recommend using several indices to gauge the fitness of the model. There are two types: the incremental fit index (like GFI, AGFI, etc.) and the badness of fit index (like RMR, RMSEA, etc.). All of them can be used to validate the data and sample size. Also they can help to select suitable criteria for assumptions. 3. Survey Investigation The historic district, as a part of the city, whose functional

properties and local socioeconomic attributes are quite different from those of the regular city, has its own particular travel characteristics. So the study of the commuters’ travel characteristics in historic areas cannot adopt the same method as that of the whole city. Research should be carried out, respectively, towards different categories of commuters. In the study, all commuters were classified into two main groups according to their working location, commuters in historic district and commuters out of the district. Therefore, the collection of data was done separately to investigate the differences of their travel characteristics. Data used for this analysis comes from household travel survey in historic district of Yangzhou (2010). The survey was conducted in the form of questionnaire, and we distributed them in every region randomly on weekdays. The content of the questionnaires Batimastat consists of two main parts: (1) individual and household characteristics, such as gender, age, occupation, annual household income, and household size and (2) travel information of all trips in a whole day, including the time taken in each trip, duration of commute time, number of trips, and travel mode choice. A total of 2000 questionnaires were delivered during the entire survey, and 1525 were returned, of which 1221 questionnaires were valid.