A Structural Equation Model (SEM) was developed in this study. SEM is a very powerful tool and is increasingly being used in travel behavior research(Golob, 2003). A complete SEM consists of two components: the structural component and the measurement component. These components are defined by three sets of equations: structural equations, measurement equations for endogenous variables, and measurement equations for exogenous variables. This study includes both of the components and thus a full SEM model. There are several measures employed to assess the goodness-of-fit in SEM. However, in most cases the fit measures do not agree (Fabrigar et al., 2010). Some take parsimony into account and others do not. Based on that, fit indices can be divided into general goodness of fit indices and parsimony fit indices. The first category indices show, roughly speaking, whether the model fits the data better than any other model. Parsimony fit indices address the issue that the model may only be fitting the noise of the data and will not be representative for population-wide application. However chi-square is an essential statistic to report along with the Root Mean Square Error of Approximation (RMSEA) and associated p-value (Hooper et al., 2008). Given the sensitivity of chi-square to model misspecification, additionally Standardized Root Mean square Residual(SRMR) is reported. To represent a good fit the value of SRMR should be less than 0.05 although values up to 0.08 are considered acceptable (Hooper et al.,2008). RMSEA value should be less than 0.05 to indicate a good fit (Golob, 2003;Washington et al., 2009). Given the complexity of the model, this study assessed the model fit on the aforementioned two indices.(www.xing528.com)
免责声明：以上内容源自网络，版权归原作者所有，如有侵犯您的原创版权请告知，我们将尽快删除相关内容。
