Table 2. Fit Indices for Measurement Model and Structural Model
The hypotheses presented in Figure 1 were tested simultaneously using structural equation modeling (SEM) via LISREL 8.72 (Jöreskog and Sörbom 2001). The modeling was undertaken by deploying covariance matrix and the maximum likelihood estimation procedure.
Table 2 presents the model fit measured using the chi-square statistic (c2), the root mean square error of approximation (RMSEA), the goodness of fit index (GFI), the non-normed fit index (NNFI) and the comparative fit index (CFI). Significant results of the chi-square statistic imply that the model was not acceptable. However, as the chi-square statistic is highly dependent on sample size, the fit of models estimated with large samples is often difficult to assess. Thus, caution needs to be exercised in its application and fit indices have been developed to address this problem (Diamantopoulos and Siguaw 2000; Ullman and Bentler 2004). The root mean square error of approximation (RMSEA) is usually regarded as the most informative of the fit indices. Values less than .05 are indicative of good fit, and between .05 and under .08 of reasonable fit (Browne and Cudeck 1993; MacCallum, Browne, and Sugawara 1996). Thus, as seen in Table 2, the model fit is reasonable, as RMSEA does not exceed .08.
The goodness of fit index (GFI) is an absolute fit index, which means that it assesses how well the covariances predicted from the parameter estimates reproduce the sample covariances. Here values greater than .90 reflect acceptable fit (Diamantopoulos and Siguaw 2000), as reflected in the GFI values shown in Table 2.
The last two of the fit measures are relative fit indices, which show how much better the model fits compared to a baseline